Doping Density Measurements of Textured Solar Cells Using Capacitance-based Techniques Jennifer T. Heath1, Amanda Bowers2, and Emily Van Doozer2 1Department of Physics, Linfield College, McMinnville, OR 97128, United States Introduction Abstract Analysis and conclusions Background Multicrystalline silicon solar cells, similar to those recently installed on T. J. Day Hall, were • Unexpected random scatter in distribution of doping densities investigated using capacitance based techniques. Commercial solar cells have textured o May indicate a non-uniform spread in dopant throughout the bulk region of the p-side • A solar cell is a diode. At its heart is a p-n surfaces to enhance the light trapping. In this project, textured and untextured devices were of the cell. junction, created by doping the n side with studied to determine if the increased surface area created by texturing significantly impacts o Did not control experiment to measure a single crystal grain or determine the region of an electron donor (phosphorus), and the p side the measurement outcomes. The measured doping densities vary from sample to sample, the wafer from which a sample was cut. with an electron acceptor (boron). from 8 x 1015 cm-3 to 2 x 1016 cm-3. This seems to indicate a significant lateral or grain-to- grain variation in doping density within each sample, which overwhelms any possible • Experiment assumed a uniformly doped wafer. • Doping creates a depletion region sandwiched influence of the surface texturing, These results are being further investigated by comparison between bulk materials where electric field with time of flight secondary ion mass spectrometry measurements. • Question regarding texturing effects on experimental results remains unanswered causes current to flow. o Depletion region – an insulating region surrounding the junction where mobile Experiment charge carriers (electrons and positive 0.65 holes) are swept away. Figure 1. Schematic of solar cell as Methods y = 0.311x + 0.3002 basic diode. P-side represents bulk 0.6 • The doping densities are a fundamental material doped with boron, creating • Applied DC bias to change the depletion 0.55 property of the device, and are chosen positive charge carriers. N-side ) width from equilibrium state (Fig. 3) -2 to optimize the solar cell efficiency. represents bulk material doped with 0.5 ) (nF o Doping density- the number of dopant phosphorus, creating negative 2 0.45 • Used lock-in-amplifier, current-to-voltage molecules per unit volume charge carriers. Current flows from preamplifier and DC power supply 0.4 p-side to n-side within depletion Figure 6. Electron backscatter diffraction image from an untextured sample showing a • Real solar cells are textured to increase the measure capacitance 0.35 region. 1/(capacitance grain size on the order of millimeters. efficiency of the device 0.3 o Texturing poses a potential problem when • Non-destructive method using capacitance-based techniques to 0.25 • Other methodologies, such as time of flight secondary ion mass spectrometry (ToF-SIMS) measure doping density. • ImageJ software was used to determine 0.2 can verify accuracy of our results. the macroscopic area of sample from 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 DC Bias (V) digital images (Fig. 4). • TOF-SIMS data at a single location in the untextured sample shows an even higher doping density of 4x1016 cm-3 (Fig. 7). Figure 4. Typical capacitance data. Slope is Figure 2. Cross-sectional Phosphorus- • SEM cross-section (Fig. 2) used to used to compute doping density. image of textured surface, doped (n+) estimate the degree of surface texturing Inset: Photograph of a typical sample with Figure 7. ToF-SIMS data for obtained using scanning the reference ruler used to measure sample textured and untextured electron microscopy. area. samples. Boron-doped (p) Results 16 The B doping density, 4x10 -3 cm , appears larger than that Doping Densities of p-bulk Region measured by capacitance, 1.5 perhaps because of variation Theory between different regions of ) 1.375 -3 the wafer. The slow decrease • Depletion region, of width W, acts like the insulator in a capacitor: !=​$⁄& . in P doping with depth, seen in cm 16 1.25 the textured sample, is an • In this one-sided junction (Na<<Nd), W and C are dominated by the lower doped artifact of the texturing. material. 1.125 textured untextured • W, and therefore Capacitance, varies with applied DC bias. 1 • Capacitance, measured as a function of DC bias, yields the doping density: Doping Density (x10 0.875 Future Work ​↓* =​2/​ε↓- ​↑2 ​​↑−2 / 0.75 0 2 4 6 8 10 12 • Create 3-D maps of doping density using ToF-SIMS to verify non-uniformity of samples and determine its spatial distribution. • Technique theoretically yields an underestimate of the doping density, Na, for textured arbitrary sample name samples because of the added microscopic surface area, which is not easily measured. • Re-measure devices with known uniform doping densities, to determine the real influence Figure 5. Distribution of doping densities for both textured and untextured samples. of surface texturing. • Initial purpose: Determine the way in which texturing affects the capacitance results. Larger samples are portrayed using larger markers. p n • Apply these results to understand the influence of texturing on other capacitance-based 15 -3 16 -3 • Doping densities varied from 9 x 10 cm to 2x10 cm measurements of device properties. Figure 4. Band bending occurs to qVapplied align the Fermi energy levels. The Econduction 16 -3 depletion region is the region in qV • Mean doping density = 1 x 10 cm built-in Acknowledgements which a built-in electric field exists. E By applying external bias, the Fermi • Larger samples had larger doping densities E The authors thank the Keck Foundation for their generous support. We would also like to depletion width W can be varied and valence thank all parties who put in effort toward the installation of solar cells on the roof of TJ Day the doping density can be measured. Hall. W ΔW .