Photoconductance measurements

A common way to measure effective minority carrier lifetimes is by monitoring the photoconductance (PC) of a sample. In the two techniques that were available for this work, the photoconductance decay (PCD) after pulse-like excitation by an external light source is measured with microwave reflection or inductive coupling, respectively.

As the recombination properties in the volume and at the surfaces, quantized byτbulk and S, may be strong functions of the injection level and vary over the sample, particularly in case of multicrystalline silicon, important requirements for a lifetime measurement technique are therefore to provide 1) spatial resolution and 2) absolute effective lifetimes at known injection levels.

The first requirement is met by the microwave detected photoconductance de-cay (µPCD), the second by the quasi-steady-state photoconductance (QSSPC) technique from which a lifetime spectrum over a large injection range is obtained.

The information content of both techniques is different so that they complement each other. For this reason, lifetime samples in this work were usually character-ized by both methods. Firstly, the spatial lifetime distribution was mapped with theµPCD, then absolute lifetime values were determined by QSSPC, typically in the area of best lifetime in theµPCD image, unless noted otherwise.

µPCD

µPCD stands for ”microwave-detected photoconductance decay”, which describes the principle how this machine measures effective lifetimes:

A brief (≈200 ns) light pulse, typically from a laser in the near infrared (e.g.

904 nm for the system at ISC Konstanz, a Semilab WT-2000, corresponding to an absorption length in Si of 32µm), illuminates a small area (≈1 mm²) of the wafer and generates excess minority carriers within the sample which cause an increase in the sample’s conductivity. The transient decay of this excess photoconductance

2.2. Characterisation of surface passivation layers 21 is monitored by measuring the change in microwave reflectivity of the sample, using microwave radiation with a frequency tuneable from 10.05 to 10.4 GHz.

Interpretation of the measured decay time constants as effective lifetimes is not straightforward. The microwave reflectance is not linear with the wafer con-ductance and sensitive to the geometrical arrangement of the sample, microwave antennae and the metallic reflector (”short circuit”) behind the wafer [Schoe95].

The sensitivity even changes sign (and so can the decay curve, leading to vis-ible ”borderlines” or rings of invalid points on the lifetime map) and must be optimized by adjusting the microwave frequency for different samples.

Due to the non-linearity in microwave reflectance and its dependence on sam-ple thickness and the distance in between samsam-ple, the metal back reflector and the microwave waveguide, the resulting photoconductance is not absolute, and thus the injection level is unknown. An iterative procedure to determine the in-jection level and thus the absolute effective lifetime which then agrees well with the effective lifetime determined by QSSPC has been presented [Ber98,Schmi99], but is not practical for implementation when measuring tens or even hundreds of different samples. Therefore, theµPCD is only used for comparative lifetime measurements in this work, while the absolute effective lifetime is subsequently determined by QSSPC, measuring in the area of the best lifetimes visible in the µPCD lifetime map.

The WT-2000 allows for a spatial resolution of down to 62.5µm, correspond-ing to the size of selective emitter or BSF structures. The number of measure-ment points or pixels in the recorded lifetime image is corresponding to the time it takes to record the image, as the method is serial, i.e. only one spot can be measured at a time and the sample area thus has to be scanned following a line pattern. That means, twice the resolution means about 4 times the measurement time. At a resolution of 2 mm, only 2-3 minutes per sample (156x156 mm²) are needed, while a resolution of 500µm requires about half an hour per wafer.

The resolution of choice for a specific measurement is influenced by the nature of the sample and the size of possible structures one wants to observe.

While information in terms of fine structures is naturally lost at coarser res-olutions, the average lifetime of the sample is not affected beyond the range of fluctuations observed when repeating a measurement with identical settings. In practice, the differences in between grains in multicrystalline material are well visible with a resolution of 500-1000µm, and 1-2 mm resolution is sufficient for monocrystalline samples.

QSSPC

The quasi-steady-state photoconductance technique was introduced by R. Sinton [Sin96]. It is now very common in semiconductor and especially photovoltaics research institutions and industry, with more than 600 lifetime-testers currently in use worldwide according to the distributor. This makes the QSSPC well suited

22 Chapter 2: Surface Passivation and Antireflection Coating

Figure 2.5: Schematics of the Sinton WCT-120 series QSSPC instrument

for comparing results with other authors, as it currently is the most commonly quoted source of published effective lifetime values.

The sample is illuminated by a photographic flash with adjustable decay time (≈0.25..12 ms), situated on top of a coil which is electrically insulated from the sample by a thin plastic layer. The coil is part of a 25 MHz RC bridge oscillator circuit. The resistor and capacitor are automatically balanced by a macro included in the MS Excel-based evaluation software. The oscillating EM-field of the coil couples with free carriers in the sample and thus generates eddy currents which induce a counter-acting current in the coil. After removal of the high frequency part by a low-pass filter, this results in a voltage signal which is almost linearly proportional to the photoconductance (PC) of the sample. Due to the diameter of the coil of≈1.5 cm, the measurement includes a circular area of the sample of∅≈3 cm. Naturally, the PC value determined in this way is an average value, which leads to errors in case the sample is smaller than the area covered by the coil, or in the case of mc-Si material with several grains of different lifetimes within the measured area.

The light intensity is measured with a gauged reference solar cell of high series resistance which is connected to a fixed load of 0.1 Ω, yielding a voltage signal with a proportionality factor of 5.80 mV/sun for the setup at ISC Konstanz.

Apart from the potential drawback of very low spatial resolution of the mea-surement, the QSSPC offers the benefits of covering a wide injection range (about 2 orders of magnitude) within a single quick measurement of<1 s as well as using a light source whose spectrum is at least similar to that of the sun, in contrast to the laser used for theµPCD.

The QSSPC also can be used for resistivity measurements, and here is superior to a 4-point probe setup when it comes to mc-Si material, as the averaging over a larger area levels out some of the doping density differences between single grains

2.2. Characterisation of surface passivation layers 23

as well as it reduces the disturbing influence of grain boundaries.