How To Measure Each Subcell’s PL in Perovskite–Silicon Tandems

To measure each subcell’s PL in a perovskite–silicon tandem, combine wavelength-selective excitation with spectrally resolved detection and explicit cross-talk checks. A short-wavelength pump can emphasize the perovskite top cell, while a longer-wavelength pump can preferentially excite silicon if it passes through the top stack and is absorbed in the bottom cell. Neither choice guarantees a pure signal. A tandem solar cell characterization system must verify the emission assignment, account for overlapping spectra and optical coupling, and preserve the device’s electrical state. Establish those conditions before converting a camera image into a subcell quality map.

Understand what separates the two emission channels

A perovskite–silicon tandem combines a wider-bandgap perovskite absorber with a lower-bandgap silicon absorber. Their band-edge emission occupies different spectral regions, but real device spectra can contain tails and overlapping contributions.

The top cell absorbs a substantial part of the higher-energy incident light. Silicon uses photons transmitted through the upper stack. In a monolithic two-terminal device, the subcells are electrically coupled through the interconnection and share a series current during operation. Four-terminal structures allow different electrical control, so record the architecture before interpreting a PL measurement.

Perovskite emission depends on composition, bandgap, temperature, and material state. Silicon emission is near its indirect band edge in the NIR and can be weak. Detector response, filter transmission, encapsulation, and the full optical stack change the measured balance. A wavelength region that looks isolated in a schematic may contain substantial background in the real specimen.

Sub-bandgap emission from the perovskite can overlap the silicon detection region. In addition, photons emitted by the top cell can be absorbed by the bottom cell, producing luminescent coupling. These effects differ: one can mix light at the detector, while the other changes the population being excited. A tandem solar cell characterization system should test both rather than relying on a filter label.

Establish selective excitation on the actual device

Select excitation wavelengths from the absorption and transmission of the finished stack. “Visible for the top cell and NIR for silicon” is a useful starting rule, but the relevant absorption edges and parasitic losses are sample-specific.

Emphasize the perovskite top cell

Choose a pump above the perovskite bandgap that is strongly absorbed in the top absorber. Use an emission collection band that rejects excitation and covers the intended perovskite PL. Verify that pump leakage, substrate fluorescence, and camera background do not dominate weak regions.

Some pump light can still reach silicon through incomplete absorption, texture, or local coverage differences. A top-cell excitation recipe should therefore be validated with a full emission spectrum and, where available, appropriate reference structures. Changing excitation wavelength while keeping the detection region fixed helps test the origin of the signal.

Emphasize the silicon bottom cell

Choose photons below the perovskite absorption edge but above silicon’s bandgap, with adequate transmission through the upper layers. Measure transmitted excitation or model stack absorption as needed. Increasing nominal NIR power does not correct an unknown fraction absorbed in contacts or transport layers.

Published wavelength thresholds are not universal recipes. One investigated tandem required excitation of at least 815 nm for a silicon-dominant signal and no longer than 750 nm for a perovskite-dominant signal. The practical lesson is to establish a wavelength-dependent selectivity test for your device, rather than applying those boundaries to every perovskite composition.

For either channel, compare absorbed photon flux when the goal is a quantitative comparison. Equal incident power at two pump wavelengths gives different photon fluxes and different absorption fractions. Record illumination area and uniformity, including whether a local probe is surrounded by unilluminated device area.

Separate spectra before choosing mapping filters

Acquire representative spectra before committing to two fixed imaging channels. Select filters using the measured emission, excitation leakage, and detector response, then verify the filtered signals against those spectra.

A long-pass filter only removes wavelengths below its transition. It does not distinguish all light above that transition by its originating subcell. For silicon mapping, a longer-wavelength collection band can reduce top-cell band-edge light while still transmitting a perovskite sub-bandgap tail. Test the residual contribution rather than assuming it is zero.

Where overlap is important, use a spectrometer or multiple bands and a justified spectral separation model. Reference spectra should be acquired under comparable conditions, and their shapes should remain appropriate across the map. A two-component unmixing fit becomes unreliable if the component shapes vary strongly or an additional emission source is omitted.

Measure filter transmission and detector response across the actual collection region. Silicon detector sensitivity decreases toward its long-wavelength cutoff; an InGaAs channel may be useful depending on the silicon emission band and required signal level. The detector’s nominal wavelength limit alone is insufficient evidence of useful sensitivity at the band edge.

For time-resolved work, verify the response and count statistics in each spectral channel independently. A visible-channel timing specification should not be assigned to an NIR configuration without measurement. Fluence, pulse spacing, and the subcells’ electrical coupling can also change the observed decay.

Acquire registered maps with controlled boundary conditions

Map each subcell using a documented excitation/detection recipe and a shared coordinate system. Preserve raw images and calibration data so that instrument nonuniformity can be distinguished from sample nonuniformity.

1. Record the device: architecture, layer stack, contact access, encapsulation, temperature, and light-soaking history.

2. Validate selective excitation: compare emission spectra at selected pump wavelengths and estimate absorbed photon flux.

3. Validate spectral separation: measure pump leakage, detector background, and residual contribution from the other subcell.

4. Acquire the two PL maps: use linear, unsaturated detector settings and document exposure and electrical conditions.

5. Correct and register: apply appropriate dark, flat-field, and illumination corrections, then align the images using stable features.

6. Analyze and recheck: inspect spectral or intensity anomalies and revisit a reference area to detect drift or sample changes.

For open-circuit characterization, ensure the device is actually in the intended state. Selective excitation in a two-terminal tandem can create unusual voltage distributions and coupling between the subcells. If bias illumination is used for the other subcell, record its spectrum and intensity. Do not call a selectively illuminated PL map an operating-point measurement without establishing the corresponding conditions.

For spatial comparisons, use a common field of view and account for chromatic magnification, focus, and point-spread differences. A pixel-by-pixel ratio between poorly registered images can create false boundaries. Retain channel-specific color scales and units; a shared display range does not make the raw intensities quantitatively comparable.

Interpret subcell contrast and choose the required system

Start with relative spatial variations under controlled conditions. Assigning defects or extracting implied voltage requires additional optical, electrical, and calibration information beyond two intensity images.

A dark top-cell region may indicate recombination loss, lower absorption, changed bandgap, or altered collection. A weak silicon region may reflect bottom-cell quality, excitation transmission, top-layer shading, or insufficient detector response. Compare local spectra and complementary electrical measurements before assigning a microscopic mechanism.

Intensity-dependent measurements add another dimension. Suns-PL relates luminescence behavior to varying illumination and can support implied-voltage or pseudo-J-V analysis under suitable calibration and model assumptions. In tandems, define the illumination of both subcells and how their signals are separated. A pseudo-J-V result does not automatically reproduce terminal behavior under current extraction and series-resistance losses.

Time Tech Spectra’s SOLARIUM-L tandem characterization system lists subcell Suns-PL, PL/TRPL, EL, and PL spectroscopy/intensity/lifetime imaging for tandem samples. Its illustrated recipes use 450 nm excitation with a 630 nm long-pass for perovskite imaging, and 825 nm excitation with an 850 nm long-pass for silicon imaging. These are demonstrated measurement conditions, not universal purity guarantees.

For a tandem solar cell characterization system specification, confirm the chosen source wavelengths, NIR sensitivity, filtering, and channel-specific timing. In particular, the listed 400–1100 nm TCSPC APD band should not be interpreted as proof of equally strong detection across the full silicon emission spectrum. Agree on a sample-based selectivity and signal-to-noise test for the intended configuration.

Conclusion

Reliable subcell PL measurement begins with excitation selectivity and ends with evidence that the detected signal belongs predominantly to the intended subcell. Use spectra to choose filters, test cross-talk, control electrical conditions, and register the maps before interpreting defects. A tandem solar cell characterization system should make these checks practical and preserve calibration information. Time Tech Spectra, a spectroscopy company, offers SOLARIUM-L with tandem PL and subcell Suns-PL capabilities. Select the exact source and detection configuration around the actual stack, particularly when weak silicon emission and perovskite spectral tails overlap.

FAQ

Does a tandem solar cell characterization system need two cameras?

Not necessarily. Sequential spectral or imaging configurations can work if their response and registration are calibrated. Simultaneous channels may help when the sample changes during acquisition.

Can the two subcells be compared using raw brightness?

No. Absorbed excitation, emission spectrum, collection efficiency, and detector response differ. Use calibrated quantities appropriate to the comparison.

Is NIR excitation always selective for silicon?

No. Selectivity depends on the top-cell absorption edge and the complete stack. Verify wavelength-dependent emission and possible sub-bandgap contributions.

Can subcell PL identify current mismatch directly?

PL provides useful recombination information, but current matching requires generation and electrical information, such as subcell-resolved EQE and appropriately controlled device measurements.