Thin-film solar cells have record conversion efficiencies of nearly 23%, specifically the ones based on Cu(In,Ga)Se2 absorbers. Such solar cells contain a thin-film stack with a standard stacking sequence of ZnO/CdS/Cu(In,Ga)Se2/Mo/glass (see Figure 1a).
Figure 1. SEM (a) and panchromatic CL images (b) acquired on the same identical area on a cross-section specimen prepared from a ZnO/CdS Cu(In,Ga)Se2/Mo/glass solar-cell stack.
Microscopic Analysis of Solar-Cell Devices Through Cathodoluminescence Measurement
An important part of the microscopic study of these highly pertinent solar-cell devices is optoelectronic characterization in scanning electron microscopy (SEM) using cathodoluminescence (CL).
In high-efficiency solar cells, a distinct trait of Cu(In,Ga)Se2 layers is a Ga/In gradient perpendicular to the substrate, with smaller (larger) Ga (In) concentrations close to the CdS/ZnO layers and larger (smaller) Ga (In) concentrations near the Mo back contact.
Since CuGaSe2 has larger band-gap energy (1.68 eV) compared to CuInSe2 (1.04 eV), the optoelectronic characteristics of the Cu(In,Ga)Se2 layers are significantly affected by the Ga/In gradient. Although Ga/In gradients can be measured using energy-dispersive X-ray spectrometry and help in indirectly concluding on local band-gap energies, it is still important to have direct access to these local, optoelectronic quantities.
The Cu(In,Ga)Se2 thin film (2 µm) examined for the current study was co-evaporated from elemental sources on Mo-coated (sputtered, 1 µm) soda-lime glass substrates (2 mm) through a three-stage process. In this process, Se, Ga, and In are co-evaporated (stage 1), Se and Cu are deposited until CuxSe forms over Cu(In,Ga)Se2 (stage 2), and again Se, Ga, and In are co-evaporated until the Cu(In,Ga)Se2 layer becomes Cu-deficient and CuxSe is consumed.
Next, the Cu(In,Ga)Se2/Mo/glass stack was additionally processed toward a completed solar cell. To perform this step, a CdS buffer layer (approximately 50 nm) was deposited in a chemical bath and a ZnO:Al/i-ZnO bilayer (with thicknesses of around 40 and 500 nm) was sputtered. A Ni-Al grid was finally deposited over the ZnO:Al/i-ZnO bilayer to enable the present collection (refer to Figure 1a for the completed solar-cell stack).
CL measurements were carried out in a Zeiss Merlin scanning electron microscope. A DELMlC SPARC CL system, fitted with an iDus InGaAs array as a detector, was also used for this purpose. All CL measurements were conducted at room temperature.
A CL spectrum was obtained in each pixel of the inspected area (point-to-point distance of around 50 nm). A panchromatic CL image, which was recorded on the same area as the one indicated in Figure 1a, is shown in Figure 1b. Energy-dispersive X-ray spectrometry (EDX) was performed to obtain the Ga distribution on the cross-section specimen vertical to the substrate. Figure 2 (red curve) shows a corresponding linescan.
Figure 2. The Ga distribution on the cross-section specimen perpendicular to the substrate obtained by energy-dispersive X-ray spectrometry (red) and the corresponding distribution of the local band-gap energy (black). Also given are the wavelength values for the band-gap energies of 1.2 and 1.3 eV.
As predicted, it shows less Ga toward the CdS/ZnO buffer/window layers, and more Ga towards the Mo back contact, with a local minimum a few 100 nm away from the CdS/Cu(In,Ga)Se2 interface.
The local band-gap energy, which is perpendicular to the substrate, can be determined through Eg(x) = (1 − x)Eg(CuInSe2) + xEg(CuGaSe2) − bx(1 − x), where x = [Ga]/([In] + [Ga]) and b is the bowing factor (the value b = 0.2 was used for the ensuing curve shown in Figure 2).
By estimating the peak shifts in the CL map shown in Figure 1, the distribution of the local band-gap energy can be accessed directly. A linescan (see Figure 3b) was obtained along the arrow shown in Figure 3a, exhibiting a larger peak wavelength and thus a smaller peak energy near the CdS/ZnO layers.
Figure 3. (a) SEM image from a ZnO/CdS/Cu(In,Ga)Se2/Mo/glass solar-cell stack. Along the arrow given in this image, a linescan was extracted from the CL spectral image (b). This linescan is in very good agreement with the distribution of the band-gap wavelength calculated from the Ga distribution measured by EDX.
When the band-gap wavelength values obtained by EDX are plotted against the depth/distance from Figure 2—in the same manner as the linescan shown in Figure 3b (also see Figure 3c)—it becomes evident that both distributions are in excellent agreement. Therefore, the CL measurement suitably reproduces the predicted distribution of the band-gap energy/wavelength.
To sum up, it has been demonstrated that local differences in the band-gap energy of semiconductor materials can be examined through CL at room temperature, and with a spatial resolution of the order of 100 nm.
The spatial resolution is also shown to rely on the diffusion length of the minority charge carriers in the material being analyzed. However, the current study demonstrates that in general, CL can be a highly valuable method for investigating the optoelectronic characteristics of thin-film photovoltaic materials.
This information has been sourced, reviewed and adapted from materials provided by Delmic B.V.
For more information on this source, please visit Delmic B.V.