Calibrating underwater photovoltaic performance
| Type | Research |
|---|---|
| Authors | Nilofar Naaz Shubhrajit Dutta Sanket Goel Kannan Ramaswamy |
| Location | Hyderabad, India |
| Status | Prototyped |
| Made | Yes |
| Tools | Xenon AM1.5G solar simulator (Photoemission Tech SS50AAA), LED AM1.5G solar simulator (Newport LSH-7320), Submersible pyranometer (Apogee MP-200), Source meter (Keysight B2912A), Acrylic water tank with hand-turned power-screw depth adjustment, 3D printer, PDMS, Hot glue gun |
| Uses | Underwater photovoltaics, Floating photovoltaics, Laboratory PV characterisation |
| Links | https://doi.org/10.1016/j.renene.2025.122993 |

Solar simulators are the standard tool for characterising underwater solar cells in the laboratory, but there is no agreed way to report the efficiency of a cell measured through a column of water. This project measured the irradiance and the current-density-voltage (J-V) response of PDMS-encapsulated monocrystalline (m-Si) and polycrystalline (p-Si) silicon cells in deionised water down to 20 cm, using xenon and LED AM1.5G simulators, and compared the measured irradiance losses with spectral calculations. The measured losses were much larger than the calculated ones. The excess was traced to two artefacts: the limited spectral response of the silicon-photodiode pyranometer, and losses intrinsic to the simulator lamps. Three definitions of underwater efficiency were compared. Only the one that uses irradiance corrected for both artefacts gives a physically consistent trend, with efficiency falling by 43% (m-Si) and 56% (p-Si) between the surface and 20 cm. Without the corrections, efficiency at 20 cm is overestimated by 59% (m-Si) and 64% (p-Si). The full study is published in Renewable Energy.[1]
Introduction
[edit | edit source]Solar cells have been proposed as power sources for autonomous underwater vehicles, marine sensors and underwater communication links. Stachiw measured silicon cells in sunlight down to 61 m in 1980,[2] and later work showed that wide-bandgap absorbers such as GaInP suit the blue-green underwater spectrum better than silicon.[3][4] Shallow submerged panels have also been studied outdoors, where water cooling can raise efficiency in the first few centimetres.[5][6]
Much of the device-level work, however, is done indoors with solar simulators shining through a water tank.[7][8] A simulator delivers 1 sun (1000 W/m2) only at its rated working distance and over a small beam area, and its spectrum differs from sunlight. The submersible pyranometers used to measure irradiance underwater are silicon photodiodes that do not respond across the whole AM1.5G spectrum. Both effects change the irradiance value that goes into the efficiency calculation, and so the reported efficiency. No experimental study focused on calibrating underwater PV measurements had been reported, a gap also discussed by Röhr et al.[9] This project addresses that gap.
Theory
[edit | edit source]At the air-water interface, the fraction of normally incident intensity that enters the water is
where and are the refractive indices of air and water. The wavelength dependence of was obtained from Cauchy's dispersion formula. Below the surface, each wavelength is attenuated according to the Beer-Lambert law, so the total irradiance at depth is
where is the absorption coefficient of water. Scattering by suspended particles is neglected. The NREL AM1.5G reference spectrum and published spectra of the two simulator lamps were propagated to depths of 0 to 100 m in pure water. Water removes most of the ultraviolet and infrared, about 57% of the solar spectrum: only visible light remains at 1 m, the red is gone by 10 m, and most of the blue-green is gone by 100 m.
To judge which absorber materials suit a given depth, an absorption efficiency was defined as the fraction of the transmitted energy that can generate electron-hole pairs in a material with cut-off wavelength :
Materials and methods
[edit | edit source]Light sources
[edit | edit source]- Xenon short-arc 150 W AM1.5G simulator (Photoemission Tech., SS50AAA), spectral range 300 to 1100 nm
- LED 150 W AM1.5G simulator (Newport, LSH-7320 ABA), 360 to 1100 nm, with sharp cut-offs below 400 nm and above 1050 nm
- Natural sunlight, measured in a clear university swimming pool on a sunny day
Both simulators have a beam area of about 5 cm × 5 cm. The surface irradiance was set to 1 sun for every source.
Test set-up
[edit | edit source]The measurements used a small variable-depth water tank designed by the research group and fabricated in the BITS Pilani Hyderabad institute workshop. The water level stays fixed. A hand-turned power screw moves the sample platform up and down, so the depth of water above the sample can be set between 0 and 20 cm without adding or removing water.

Components
[edit | edit source]- Tank: 5 mm acrylic sheet, 10 cm × 10 cm × 30 cm, open at the top
- Sample platform: acrylic plate that carries the pyranometer or the encapsulated cell
- Power screw: vertical threaded rod in one corner of the tank, held in a bearing block at the bottom and a bracket with a hand knob at the top
- Nut fixed to the platform, which rides on the screw
- Submersible silicon-photodiode pyranometer (Apogee Instruments, MP-200) with handheld meter
- Source meter (Keysight B2912A) for J-V measurements at ambient temperature
- Deionised water
How it works
[edit | edit source]- Fill the tank with deionised water to a fixed level.
- Place the pyranometer on the platform and raise the platform until the sensor is level with the water surface.
- Position the tank under the simulator at the distance where the pyranometer reads 1 sun (1000 W/m2) at the water surface. Keep this distance fixed for the whole run.
- Turn the knob to lower the platform in steps down to 20 cm, and record the irradiance at each depth.
- Replace the pyranometer with the encapsulated cell and repeat the same depths, recording a J-V curve at each one.
The pyranometer and the cell are measured one after the other on the same platform, so both see the same water column at each depth.
Notes for replication
[edit | edit source]- The screw and nut stay submerged, so use corrosion-resistant hardware.
- Keep the screw in a corner, outside the simulator beam (about 5 cm × 5 cm), so it does not shade the sample.
- The cross-section of the tank should be larger than the beam, so the tank walls do not clip the light.
Devices
[edit | edit source]- Monocrystalline silicon cell (IXYS), VOC 6.5 V, Imax 44.6 mA
- Polycrystalline silicon cell (Kitronik), VOC 3.5 V, Imax 100 mA
Both cells were encapsulated in polydimethylsiloxane (PDMS) to protect them from water, as described below.
Waterproof encapsulation
[edit | edit source]
Each cell was waterproofed with a 3D-printed enclosure and a poured PDMS layer:
- Print an open enclosure (mould) sized to the cell.
- Place the cell in the enclosure and bring the electrical leads out through the enclosure wall.
- Pour PDMS over the cell to fill the enclosure and let it cure.
- Seal all edges, lead exits and any other openings with hot glue.
The encapsulated cells were used in water throughout the group's experiments for nearly five years. Over that period no water entered the enclosures and the PDMS did not yellow. The same approach, a printed mould filled with PDMS, was used to waterproof the electrical connections of a large commercial solar panel.
The group has also trialled a rose-petal surface pattern on the PDMS to make it hydrophobic, with the aim of reducing biofouling on submerged cells.
Separating the irradiance losses
[edit | edit source]The measured loss at each depth, , where is the surface irradiance and the irradiance at depth , was split into three parts:
- , the part of the spectrum the pyranometer cannot detect. This was calculated at each depth by comparing the Apogee spectral response with the propagated spectrum of each source.
- , the loss intrinsic to the lamp: its finite size and divergent beam, which cause intensity to fall with distance independently of the water.
- The remainder, , taken as the real loss caused by the water.
Efficiency definitions
[edit | edit source]Three efficiencies were calculated from the same J-V data:
(absolute: input fixed at 1 sun at every depth)
(modified: input taken as the pyranometer reading at depth )
(corrected: pyranometer reading corrected for and )
Results and discussion
[edit | edit source]Irradiance underwater
[edit | edit source]The measured irradiance losses were well above the calculated ones for every source. For sunlight, the calculated loss rose from 14% at 5 cm to 27% at 20 cm, while the measured loss rose from 29% to 44%. For the xenon simulator the calculated loss was 21% at 5 cm and 37% at 20 cm, against 47% and 73% measured. The LED simulator behaved like the xenon lamp, with close to 80% measured loss at 20 cm.
Spectral differences between the lamps and sunlight account for only a few percent of the integrated irradiance (about 8% for xenon and 6% for LED over 300 to 1100 nm), far less than the gap between measurement and calculation. The pyranometer misses part of the short-wavelength region and everything above 1100 nm. The undetected fraction of sunlight fell from 15% at the surface to about 13% at 20 cm. For the xenon lamp it grew from about 2% to about 10% by 16 cm, and for the LED lamp from 2.5% to about 5%. For sunlight, the pyranometer accounts for almost all of the discrepancy, leaving about 4% that is attributed to impurities in the pool water. For the simulators, a large lamp-intrinsic loss remains.
| Light source | Measured loss | Pyranometer | Lamp | Loss due to water |
|---|---|---|---|---|
| Sunlight | 44.6% | 12.75% | none | 31.85% |
| Xenon simulator | 73% | 10.63% | 24.94% | 37.43% |
| LED simulator | 77% | 4.96% | 38.58% | 33.46% |
Once the instrument and lamp artefacts are removed, the loss due to water at 20 cm is 32 to 37% for all three sources, much closer to the calculated values.
Photovoltaic parameters
[edit | edit source]For both cells and both simulators, JSC decreased exponentially with depth and VOC decreased linearly. The fill factor showed no consistent trend, except for p-Si under the LED lamp, where it fell linearly with depth. The largest spread in VOC was seen for m-Si under the LED lamp.
| Parameter | m-Si, xenon | m-Si, LED | p-Si, xenon | p-Si, LED |
|---|---|---|---|---|
| JSC (mA/cm2) | 1.17 | 1.24 | 2.92 | 2.61 |
| VOC (V) | 6.01 | 6.02 | 3.46 | 3.44 |
| FF | 0.69 | 0.71 | 0.74 | 0.74 |
Underwater efficiency
[edit | edit source]The three definitions give qualitatively different results. rises with depth, which is unphysical: the uncorrected pyranometer reading underestimates the irradiance reaching the cell, so the denominator is too small. falls with depth but ignores the real attenuation and spectral change, so it is not a meaningful efficiency. falls with depth in a way consistent with the calculated absorption efficiency of silicon:
- m-Si: from 12.8% at the surface to 7.2% at 20 cm, a 43% decrease
- p-Si: from 10.4% at the surface to 4.5% at 20 cm, a 56% decrease
Without the corrections, the efficiency at 20 cm would be overestimated by 59% for m-Si and 64% for p-Si. Applying the corrected irradiance to a commercial panel predicts that its maximum available power falls from 500 W at the surface to 262 W at 20 cm, a 45% reduction.
Outdoor studies have reported a peak in efficiency a few centimetres below the surface, attributed to water cooling.[10] No such peak appeared indoors, because the simulators do not heat the cells as the sun does and the lamp-intrinsic losses dominate. Outdoor measurements by the same group did reproduce the shallow-depth gain, which supports the cooling explanation. Silicon panels therefore benefit from submersion only at very shallow depths, below about 4 cm.
Choice of absorber material
[edit | edit source]The absorption efficiency was calculated to 100 m depth for nine absorbers. Low-bandgap materials (about 1 eV or below: Ge, CuInSe2, p-Si) lose absorption efficiency steadily with depth and are poor choices underwater. Mid-bandgap materials (below 1.5 eV: m-Si, CdTe) suit depths down to about 1 m. Wide-bandgap materials (a-Si, GaInP, Cu2O, Se) are the better choice beyond about 10 m.
| Material | Ge | p-Si | CuInSe2 | m-Si | CdTe | a-Si | GaInP | Cu2O | Se |
|---|---|---|---|---|---|---|---|---|---|
| EG (eV) | 0.62 | 1.09 | 1.05 | 1.33 | 1.45 | 1.76 | 1.88 | 2.1 | 2.2 |
| λG (nm) | 2000 | 1140 | 1180 | 927 | 855 | 704 | 659 | 590 | 563 |
Limitations
[edit | edit source]- The small (5 cm × 5 cm), divergent simulator beam and the air gap between lamp and water surface cause the lamp-intrinsic loss. A larger, collimated source would remove most of it.
- An alternative is to set 1 sun at each depth in air before filling the gap with water, so that only the pyranometer and water losses remain. This was not practical here: changing lamp power shifts the spectrum, the simulators cannot reach 1 sun at greater depths, and a small air gap remains in the current set-up.
- Measurements were made in still water. Spectral fluctuations from waves and surface disturbance were not included.
- A submersible pyranometer whose spectral response matches AM1.5G would reduce the pyranometer correction.
Conclusions
[edit | edit source]Underwater PV efficiencies measured with solar simulators should be reported using irradiance corrected for both the pyranometer's spectral response and the lamp-intrinsic loss. With these corrections, m-Si and p-Si cells lose 43% and 56% of their surface efficiency at 20 cm, and the trend agrees with the calculated absorption efficiency of silicon. Without them, efficiency appears to rise with depth or is overestimated by about 60%. The procedure can be applied to other absorber materials and to floating and shallow-submerged PV tested in the laboratory.
Related work
[edit | edit source]- Influence of water turbidity on the irradiance and response of underwater a-Si cells.[11]
- Author page: User:Nilofar Naaz
References
[edit | edit source]- ↑ Naaz, N., Dutta, S., Goel, S., Ramaswamy, K. (2025) Calibrating underwater photovoltaic performance: Demonstration using monocrystalline and polycrystalline silicon solar cells. Renewable Energy, 247, 122993. https://doi.org/10.1016/j.renene.2025.122993
- ↑ Stachiw, J.D. (1980) Performance of photovoltaic cells in undersea environment. Journal of Engineering for Industry, 102(1), 51–59. https://doi.org/10.1115/1.3183829
- ↑ Jenkins, P.P., et al. (2014) High-bandgap solar cells for underwater photovoltaic applications. IEEE Journal of Photovoltaics, 4, 202–206. https://doi.org/10.1109/JPHOTOV.2013.2283578
- ↑ Röhr, J.A., Lipton, J., Kong, J., Maclean, S.A., Taylor, A.D. (2020) Efficiency limits of underwater solar cells. Joule, 4, 840–849. https://doi.org/10.1016/j.joule.2020.02.005
- ↑ Lanzafame, R., et al. (2010) Field experience with performances evaluation of a single-crystalline photovoltaic panel in an underwater environment. IEEE Transactions on Industrial Electronics, 57, 2492–2498. https://doi.org/10.1109/TIE.2009.2035489
- ↑ Rosa-Clot, M., Rosa-Clot, P., Tina, G.M., Scandura, P.F. (2010) Submerged photovoltaic solar panel: SP2. Renewable Energy, 35, 1862–1865. https://doi.org/10.1016/j.renene.2009.10.023
- ↑ Enaganti, P.K., Dwivedi, P.K., Srivastava, A.K., Goel, S. (2020) Study of solar irradiance and performance analysis of submerged monocrystalline and polycrystalline solar cells. Progress in Photovoltaics, 28, 725–735. https://doi.org/10.1002/pip.3264
- ↑ Zhang, A., Sartor, B.E., Röhr, J.A., Taylor, A.D. (2024) LED-based characterization of solar cells for underwater applications. STAR Protocols, 5. https://doi.org/10.1016/j.xpro.2023.102833
- ↑ Röhr, J.A., Sartor, B.E., Lipton, J., Taylor, A.D. (2023) A dive into underwater solar cells. Nature Photonics, 17, 747–754. https://doi.org/10.1038/s41566-023-01276-z
- ↑ Sheeba, K.N., Rao, R.M., Jaisankar, S. (2015) A study on the underwater performance of a solar photovoltaic panel. Energy Sources, Part A, 37, 1505–1512. https://doi.org/10.1080/15567036.2011.619632
- ↑ Naaz, N., Enaganti, P.K., Wani, W.A., Goel, S., Ramaswamy, K. (2021) Influence of water turbidity upon the solar radiation and photovoltaic response of amorphous silicon based underwater solar cells. Energy Proceedings, 23, 571. https://doi.org/10.46855/energy-proceedings-9473
| Authors | Nilofar Naaz |
|---|---|
| License | CC-BY-SA-4.0 |
| Cite as | Nilofar Naaz (2026). "Calibrating underwater photovoltaic performance". Appropedia. Retrieved November 16, 2026. |