Hyperspectral cameras in semiconductor industry
HERA hyperspectral cameras are widely used in the semiconductor industry, to characterize silicon wafers and new materials (e.g. perovskites), both in macroscopy and microscopy.
A significant endeavor is currently in progress to enhance the efficiency and reduce the costs associated with semiconductors used in photovoltaic and light emitting applications. This entails the exploration and development of novel manufacturing and synthesis methods that yield more uniform materials with fewer defects.
Luminescence, whether induced electrically or optically, serves as a valuable tool in this pursuit. It provides local insight into the recombination processes occurring within the film, without complications of charge extraction through the multiple layers of a complete device.
In this application note, we show three examples of measurement acquired with the HERA cameras, ideal devices for mapping light emission from semiconductors. They enable a fast and quantitative characterization of the semiconductor film, with high spatial and spectral resolutions.
Wide-field imaging: Electro-Luminescence mapping of a silicon solar cell
Defects in photovoltaic devices cause recombination of photogenerated carriers, preventing their extraction and reducing the cell efficiency. Electro-luminescence mapping can reveal the location and nature of these detrimental defects.
Driving a solar cell “backwards” (i.e., applying current) generates electro-luminescence, as carriers are injected at the electrodes and recombine in the active layer. In an ideal cell, all carriers undergo band-to-band recombination, which in silicon generates (with very low efficiency) light near 1100nm. However, defects in the crystal structure can generate other, unfavourable recombination pathways. While often referred to as “non-radiative” recombination, occasionally these processes can generate a photon, typically with lower energy than the band-to-band emission. These very rare photons can be captured to learn about the energetics and distribution of the defects.
In this experiment, we used the HERA SWIR (900–1700 nm), well suited for measuring silicon luminescence. The measurement setup is shown in Figure 1: the HERA is mounted on a tripod, above the solar cell, which was connected to a 10 A current supply. Mounted 75 cm above the sample, the640×512 pixels sensor gives a spatial resolution of approximately 250 µm.
Most importantly, the lack of an input slit in the HERA optical system provides a very high light throughput, making the HERA the ideal choice for measuring the very weak light emission.
Figure 1: Experimental Setup
Figure 2. A and B: Electroluminescence (EL) images at two selected wavelengths (1150 nm and 1600 nm). C: EL spectra corresponding to three different regions in A and B (coloured squares in the images).
Figure 2A and 2B show images of the electro luminescence (EL) at two wavelengths, 1150 nm (band-to-band emission), and 1600 nm (defect emission), which are the average of 4 scans (total acquisition time: 5 minutes). By analyzing the images, one can see that the defective regions are clearly resolved, despite being much less bright than the main emission. Also, the regions with strong defect emission feature relatively weak band-to-band emission.
One can notice several regions that are dark at both wavelengths; this is likely due to damage of the cell received during the shipping of the sample.
Spectra from the small square regions of interest (indicated in Figure 2A and 2B) are shown on a log scale in Figure 2C.
Microscopy imaging: Photoluminescence microscopy of metal - halide perovskite thin films
Metal-halide perovskites enable the fabrication of very efficient solar cells and LEDs from low-tech and cost-effective methods such as spin-coating. One challenge with these methods is maintaining a uniform composition at microscopic length scales. Photoluminescence microscopy is an especially powerful tool for characterizing this in homogeneity.
The HERA hyperspectral camera can be connected to the C-mount camera port of any microscope (upright or inverted), and begin collecting hyperspectral data straight forwardly, without the need of any calibration procedures.
In this experiment, we used the HERA VNIR (400-1000 nm) coupled to a Nikon LV100 upright microscope (Figure 3) to characterize the band-gap distribution of an alloy of two halide precursors. The advantage of alloying two halide precursors is the capability to tune the band-gap of the material; however, these two components often de-mix, leading loss of performance. The goal of this experiment is then to detect this de-mixing: in fact, local changes in the mixing ratio change the local band-gap, leading to the emission of photons of different energy.
In this configuration, the excitation light was from a mercury lamp, fitered at 350 nm via a bandpass filter, and rejected from the camera using a dichroic beam splitter in the emission path. The high throughput of the HERA enabled collection of the full data cube (1.3 million spectra) within approximately 1 minute measurement time.
Figure 3: HERA VIS-NIR coupled
to a NikonLV100 upright microscope
Figure 4. Spectrally integrated intensity maps of the sample (A: full scale; B: zoom-in).
Figure 4A and 4B shows respectively the full-scale and zoomed-in images of the total integrated signal across all wavelengths (400–1000 nm), revealing bright features with length scales on the order of 1 µm.
When we compare the spectra from a bright vs a dark region (black and red curves in Figure 5B), we see that there is actually emission from the dark regions, and that not only is it less intense, but also centered at a shorter wavelength than the bright regions. In fact, the spectra have a bimodal shape, likely corresponding to emission from the de-mixed precursors. This variation in bandgap is clearly illustrated in the emission map of Figure 5A.
We can understand now why the low-bandgap regions appear brighter – carriers likely relax therefrom the high-bandgap regions and are unable to return before undergoing radiative recombination.
Figure 5. A: Intensity map showing the mean emission wavelength.
B: Emission spectra (normalized) from a bright and dark region.
Macro imaging: Electro-luminescence mapping of a perovskite-silicon tandem solar cell
The macro lens accessory for HERA optimizes light collection for objects with dimensions 2–3 cm, ideal for imaging semiconductor devices.
In this example we collect a map of electroluminescence spectra from a silicon-perovskite tandem solar cell (3×3 cm). Figure 6 shows the very simple setup for this measurement. The tandem cell was connected to a bench-top power supply (set to constant 1.86 V). Carriers injected into the cell by the power supply then recombine in both the perovskite and silicon active layers, emitting photons with an energy (i.e., wavelength) correspond-ing to the optical bandgap of the active layer (~750 nm for perovskite, and >900 nm for silicon).
Figure 6: HERA with macro adaptor, a benchtop power supply, and a stable platform are the only equipment required to spectrally map elettroluminescente from semiconductor devices.
Figure 7: Images of intensity at different wavelengths reveal emission from the upper perovskite layer (panel A) and the lower silicon layer (panel B).
Panel A of Figure 7 shows the emission from the perovskite layer (brightness using a log scale to aid visualization), and panel B the emission from the silicon layer. Both of these images are taken from a single hyperspectral image. We can see inhomogeneity in both layers, although much more variability in the perovskite, and at shorter length scales, revealing the faceted, grain-like structure of the film.
As we have a full spectrum for each pixel, HERA can reveal much more than just the brightness of emission from each layer. Figure 8B shows EL spectra from three regions of interest (the blue, red, and yellow coloured squares shown in inset and also in Figure 7). In these spectra we can see the perovskite emission centered near 750 nm, and the onset of emission from the silicon layer above 900nm. The spectra reveal a redshift of the perovskite emission close to the device electrode. Contrary to the mixed halide PL example shown in the previous section, we also note that the redshifted regions show lower emission intensity.
Calculating the mean wavelength of perovskite emission from each pixel of the full image, we generate the map of emission wavelength shown in Figure 8A. This clearly illustrates the regions around the electrode where the optical bandgap of the perovskite appears to be reduced, potentially due to field-induced halide segregation, or potentially by a reaction between the perovskite and the metal electrode.
Figure 8: (A) Map of peak emission wavelength from the perovskite layer, revealing a reduction in optical bandgap near the center electrode. (B) Insets show emission intensity (top) and wavelength (bottom) near the regions of interest. Spectra from these regions (indicated by coloured squares) clearly show emission from the perovksite (~750nm) and silicon (>900 nm) layers.
Discover the HERA Hyperspectral Cameras
