Device Diagnosis

In general, a low Jsc may be evaluated from the energy dependence of the photocurrent loss. It is useful to measure and model the internal QE. The external QE measurements are described in Chapter 17. The internal QE is modeled (neglecting photon recycling) according to

Figure 8.18 Measured (crosses) and modeled (lines) quantum efficiency of a GaInP solar cell. The contributions from the different layers of the solar cell are labeled and demonstrate how the emitter dominates the blue response, whereas the base dominates the red response. The relatively large contribution from the emitter is a result of the strong absorption of direct-gap materials

Equations (8.2-8.8) (Figure 8.18), and is determined experimentally from

QEInternal = QEexternal/(1 – Reflectivity). (8.30)

Accurate knowledge of the absorption coefficient is essential to successfully model the QE. The absorption coefficients of GaAs and GaInP were discussed above. High-quality samples may exhibit non-negligible photon recycling, implying that the omission of photon recycling in the model may need to be revisited. The application of Equations (8.2-8.8) is most useful when there are significant losses within the cell. Figure 8.19a compares the QE of a typical GaInP-cell (solid line) with what would be expected if there were no loss in the AlInP window (top curve) or

Figure 8.19 Modeled QE of GaInP cell. (a) The solid line, relative to the “no AlInP window” line, shows the effect of absorption of 25 nm of AlInP. The two lower curves show the degradation from an increased front-surface recombination velocity Sfront or decreased emitter diffusion length Lemitter. (b) Comparison of a thin (0.3 |im base) and a thick (3 |im base) GaInP cell


Figure 8.20 (a) Comparison of I – V curves for a “good” GaInP cell and a GaInP cell with an

extra junction caused by the AlInP window. Transmission-line measurements showed that the non­ohmic contact resistance was high through the window layer. (b) I – V curves for good and bad GaInP/GaAs tandem cells. The “bad” curve shows a shunt which appears across the corner of the tandem I – V curve; measurements of the same cell under bottom-cell-limited and top-cell-limited conditions (dashed curves) show that the shunt is in the bottom junction poor collection in the emitter (bottom curves). Although window-absorption losses can easily be distinguished from emitter losses, the similarity of the two lower curves demonstrates the difficulty of distinguishing poor front-surface recombination from poor emitter-material quality. However, it is somewhat easier to differentiate poor base-material quality from poor rear passivation using a series of devices with variable base thickness. A cell with a thick base layer is more sensitive to the diffusion length in the base, whereas a cell with a thin base layer is relatively more sensitive to the quality of the back-surface field (see Figure 8.19b).

There are numerous reasons why the VOC or FF may be degraded. Figure 8.20 illustrates two examples.

Extra junctions are more likely to be problems when working with Ge because III-V elements dope Ge and Ge dopes the III-V materials. The back of the Ge wafer must be etched before processing to avoid an extra junction at the back [135]. Accidental junctions in Ge are often highly shunted, with nearly ohmic I – V characteristics. In this case, these are easiest to observe at high concentration because the VOC of the Ge junction increases faster with photocurrent than that of the intended junction. Spurious junctions in the Ge may add, subtract, or both (in the case of back-to-back junctions) to the VOC.

When a two-junction cell shows evidence of shunting, it is useful to determine which of the two junctions is shunted. This can be determined by measuring the light I – V curve under two different spectra, a red-rich spectrum which reduces the photocurrent of the top junction, and a blue-rich spectrum which reduces the photocurrent of the bottom junction [132]. The example in Figure 8.20b shows a case for which the bottom cell is shunted. This sort of problem is often related to defects originating from particulates or poor wafer quality. Particulate exposure before or during growth is often a bigger problem for GaInP/GaAs cells than for single-junction cells.