Instrumentation for Quantum Efficiency Measurement of Solar Cells: Part 2

Part 1 of this article [1] by Adnan Adla, PhD, Newport Spectra-Physics GmbH, is a discussion of needed hardware for quantum efficiency (QE) measurement from light beam generation to its absorption by the cell under test. Part 2 continues this discussion for the detection system and bias light source. Furthermore, ways for improving QE measurement accuracy will be discussed.

Bias light source and bias voltage

Because of a linear response of some solar cells to illumination intensity, the quantum efficiency of such cells is constant. A typical example is crystalline Si (c-Si) cells. In these cases, there is no need for a biasing light source for the QE measurement. The measured QE is valid for any illumination. For other cells that have a non-linear response with illumination intensity, e.g., organic cells, it is essential to measure the QE under defined and known bias illumination. The measured QE is then typically reported together with the intensity of illumination or with cell bias voltage. Only then is it possible to compare measurements of different cells at different laboratories. Typically, QE is measured in those cases for a series of illumination intensity and the value at 1 sun (1000W/m2) irradiance is reported. A typical bias light source is a QTH (white light) source of adjustable output intensity that enables uniform illumination of the sample cell. For multi-junction cells, different light beams of a narrow wavelength range are needed for activating only the junction that is currently measured. Typically, the output beam of the bias source is divided into several beams (equal to the number of sub-cells) and the beams are made “monochromatic” using proper bandpass-filters. Alternatively, only one bias beam is used with the entire bandpass filter for each sub-cell (Figure 1).

Figure 1. QTH light source for white and colored light cell biasing
Figure 1. QTH light source for white and colored light cell biasing

For QE measurements under real-world conditions, the output from a solar simulator can be used for white light biasing. This consists of a Xenon light source equipped with an air mass 1.5 global (AM1.5G) filter for best spectral match to sun light. Commonly, 1 sun (100mW/cm2) irradiance is used for reporting purposes.

As an alternative to a bias light source, a bias voltage may be applied to the measured cells, using an adjustable voltage source. This method enables voltage dependent QE measurements and it is preferred, when reporting on real-world conditions is not targeted.

Detection system

A calibrated photo diode or reference cell is used for detection of the monochromatic light that falls on the test cell (from the monochromator). The detector can be read by a lock-in amplifier (AC method) or by a power meter (DC method). The AC method is by far the better choice in terms of excluding diffusion light

Subscribe to Photovoltaics World
Follow Photovoltaics World on via editors Pete Singer, and Debra Vogler, Or join our Facebook group

(contamination), day light and bias light and for detecting low light levels. In this method, a chopper is added to the monochromatic beam. The beam is chopped at low frequencies, typically a multitude of half the line frequency, e. g., 25, 50 or 75Hz in European countries, which helps excluding the light ripple caused by line frequency. Only in cases where the measured cell needs longer time for raising the photo current to its equilibrium value the ac method is not suitable. In such cases, such as most common dye sensitized solar cells (DSSC), a very low chopping frequency (typically bellow 10Hz) is needed, which can not be realized by some lock-in amplifiers. Here a DC detection method is the better choice. For excluding diffusion light and day light the sample is put in a dark box for QE measurements.

Measurement accuracy improvement

Typical QE measurement accuracy is in the range of 5%. Several aspects have to be considered for improving the accuracy to a level of about 1% or lower. The most important factors that affect measurement repeatability and accuracy are the light ripple, beam non-uniformity, and the need to move the sample or reference detector between two wavelength scans in most common QE configurations. The light ripple of arc sources can be up to 0.5%. To minimize the light ripple, a light intensity controller (LIC) can be added to the light source. The LIC senses a portion of the light and sends a feed back to the power supply to change the operation power of the lamp, so that output intensity is kept constant. A light ripple of well below 0.1% can be easily achieved, which clearly minimizes the uncertainty of measured QE (Figure 2).

Figure 2. Reduction of temporal instability of beam intensity of a Xe source using Oriel light intensity controller (LIC) 68950.
Figure 2. Reduction of temporal instability of beam intensity of a Xe source using Oriel light intensity controller (LIC) 68950.

The beam uniformity can be improved for uniform illumination of the test cell and reference detector using an integrating sphere. The sphere can be mounted after the monochromator. Unfortunately, using an integrating sphere leads to drastically lower power throughput in the system. The signal-to-noise ratio becomes low and may affect measurement accuracy, especially of low response cells. The larger the sphere, the better is the beam uniformity and the lower the throughput, thus an optimum sphere size has to be considered. An alternative beam homogenization method is the use of diffusion disks in front of the test cell and detector. The power throughput does not suffer markedly in this case, and the cost is by far lower than the cost of integrating spheres, but the achievable beam uniformity is limited. This easy solution is frequently a valuable trade-off.

While difficulties caused by light ripple and beam non-uniformity are easy to solve, it is a delicate operation to prevent moving the test cell and detector between the wavelength scans, as this requires innovative design of the beam delivery engine. Newport Oriel came out recently with a patent-pending solution for this problem [2]. The beam delivery is based on a beam splitter that enables measuring all needed signals simultaneously during a single wavelength scan instead of up to 4 scans (reference, test, reflectance and transmission scan) that are needed in common QE measurement solutions. Figures 3 and 4 show the principle of this turn-key system.

Figure 3. Block diagram of Oriel IQE-200
Figure 3. Block diagram of Oriel IQE-200

The optical layout of the QE-200 (Fig. 4) comprises one spectrally neutral 50-50 beam splitter and four lenses. The output light from the monochromator is first collimated by lens 1. The collimated light is then split into two beams. One beam passes the beam splitter and is focused by lens 2 onto the reference Detector. This detector measures the output light power at any given wavelength. The other beam is reflected by the beam splitter down onto the sample surface through focusing lens 3 for QE measurement. The spot size on the sample is determined by lens 3. For internal quantum efficiency measurement, the (totally) reflected light from the sample is collimated by lens 3 and focused behind the beam splitter by lens 4 onto the reflectance detector.

Figure 4. Patent pending beam delivery engine of Oriel IQE-200.

Figure 4. Patent pending beam delivery engine of Oriel IQE-200.

Figure 5 shows the results for a mono crystalline silicon cell, calibrated by VLSI, an NREL-certified calibration laboratory. The green trace in Fig. 5 shows the curve provided by VLSI. The spectral reflectance measurements performed with IQE-200 have been calibrated and controlled based on measurements made by a qualified reflectance instrument. The statistical analysis of independent measurements shows that IQE spanned a ±0.7% range about the average value. The IQE curve (red trace in Fig. 5) was obtained simultaneously with the QE curve (blue trace in Fig. 5). A good correlation between QE measured with IQE-200 and that measured by VLSI was obtained (green trace in Fig. 5). A similar behavior was obtained using an NREL-calibrated cell (not shown here).

Figure 5. Comparison between QE curves obtained using Oriel IQE-200 (blue curve) and one obtained by a certified calibration lab on the same cell (green trace). IQE curve (red trace) was obtained simultaneously with the blue curve.
Figure 5. Comparison between QE curves obtained using Oriel IQE-200 (blue curve) and one obtained by a certified calibration lab on the same cell (green trace). IQE curve (red trace) was obtained simultaneously with the blue curve.


Using modular components and instruments available on the market enables well-customized QE configurations that offer great flexibility needed by solar cell researchers for testing existing and new solar cells. Only when components are carefully chosen to fit to each other and fulfill the requirements of the application can highly repeatable and accurate QE measurements be made under high signal-to-noise ratio. A carefully chosen QE configuration can achieve measurement accuracy of better than 1%. Innovative beam delivery engine offered by Oriel IQE-200 with no moving parts for IQE measurement and simultaneous detection of all signals during one wavelength scan not only adds repeatability and accuracy to the measurements, but also the measurement duration is reduced by up to 4 times, compared with common QE measurement solutions.

[1] A. Adla, “Instrumentation for quantum efficiency measurement of solar cells,” Photovoltaics World, pp. 28-31, July/August 2010.
[2] R. Ciocan, Z. Li, A. Feldman, J. Donohue “System for High Accuracy Internal Quantum Efficiency Measurement” IEEE 34th Photovoltaic Specialists Conference, pp 1862 – 1863, 2009.

Adnan Adla, PhD, is a product manager at Newport Spectra-Physics GmbH, Guerickeweg 7, 64291 Darmstadt, Germany; ph.: +49 (0)6151/708-355; email Read the first part of this article here.

Previous articleCIS thin film PV at Solar Frontiers will hit 1GW every 2 years
Next article>90% of Americans want more solar energy: Red and blue see green future

No posts to display