Design CPV modules that use Fresnel lenses to concentrate sunlight 500-2000x onto III-V triple-junction cells (InGaP/GaAs/Ge) for 40%+ efficiency. Adjust concentration ratio, DNI and cell temperature to see module efficiency and waste-heat flux update in real time, and the cooling regime needed to keep the cell alive.
Parameters
Concentration ratio X
suns
Optical concentration by Fresnel lens or mirror (X=1 is flat-plate PV)
DNI (direct normal irradiance)
W/m²
Only the direct beam is usable by CPV. Drops sharply under cloud or humidity
Cell efficiency at 1 sun η₁
%
Cell efficiency at one sun (AM1.5G, 1000 W/m²)
Cell area
cm²
Active area of a single III-V chip
Optics efficiency
%
Combined efficiency of primary Fresnel/mirror and any secondary optics
Cell temperature T
°C
Junction temperature after cooling. Corrected against a 25°C reference
Cell technology
Choose the III-V architecture (display reference only)
Results
—
Concentrated flux (W/m²)
—
Cell η at X suns (%)
—
Module efficiency (%)
—
Power per cell (W)
—
Waste-heat flux (W/cm²)
—
Required cooling
—
CPV side view — lens, cell, tracker
A Fresnel lens focuses sunlight onto a small III-V chip, with a heatsink below and a 2-axis tracker on the side. Colour reflects waste-heat flux (green = safe, red = liquid cooling needed).
Efficiency vs concentration (T-corrected)
Waste-heat flux vs concentration
Theory & Key Formulas
$$\eta_{cell}(X) = \eta_1 \cdot \left(1 + \frac{kT}{qV_{oc}}\ln X\right) + \beta(T-T_{ref}),\quad P_{cell} = X\cdot DNI \cdot A \cdot \eta$$
X is the concentration ratio, η₁ is the cell efficiency at one sun, β is the temperature coefficient (~-0.13 %/°C) and A is the cell area. V_oc rises logarithmically with concentration and efficiency drops with temperature; this tool uses voltageBoostFactor = 1 + 0.06·ln(X) as the approximation.
Concentrated flux Φ and waste-heat flux q. η_opt is the optics efficiency. q above 5 W/cm² needs forced-air cooling; above 15 W/cm² liquid cooling is mandatory.
Concentrated photovoltaics (CPV) — III-V triple junctions and cooling design
🙋
I've never heard of "CPV". How is it different from a normal silicon solar panel?
🎓
CPV stands for concentrated photovoltaic. A normal silicon panel just sits there and receives whatever sunlight hits it. A CPV module instead uses a Fresnel lens (or a mirror) to concentrate sunlight 500-2000 times onto a tiny ultra-efficient cell. That cell is a III-V compound semiconductor — typically a stack of InGaP/GaAs/Ge — that absorbs the solar spectrum across three or more bandgaps, so almost no photon energy is wasted. Silicon flat plates get 22-26% efficiency; CPV modules get 35-46% and research cells have hit 47.1% — Spectrolab's six-junction cell, NREL-verified in 2020, still holds the record.
🙋
That's almost double! So why don't we use CPV everywhere? With the default settings at X=500 the tool says "Required cooling = Liquid cooling".
🎓
Good catch. CPV has three big walls. First, it can only use the direct beam — Direct Normal Irradiance, or DNI. On cloudy or humid days the sunlight scatters and the DNI collapses, so CPV only makes sense in dry, sunny places like the Sahara, the Atacama, central Australia or the US Southwest. Second, it needs a 2-axis tracker because the Fresnel focus moves off the cell with only a few degrees of misalignment. Third — which is what you just discovered — there's a brutal amount of waste heat. About half the energy you concentrate ends up as heat. At 500 suns you're dumping ~20 W per square centimetre into the cell. That's an order of magnitude worse than a CPU.
🙋
The waste-heat chart is steep. The flux goes up almost in a straight line with concentration. Is it really linear?
🎓
Yes, it's basically linear. q_heat = X · DNI · η_opt · (1-η) / A_cell. For a fixed cell area, q scales linearly with X. That's why ~100 suns can survive with a passive heatsink, ~500 suns needs forced air, and above ~1000 suns you really do need liquid cooling — microchannel heatsinks or jet-impingement coolers. The HCPV systems Amonix and Soitec commercialised used thick copper spreaders and natural convection, but the cooling-loop cost is exactly what put them out of business once silicon prices crashed after 2015.
🙋
When I push the cell temperature from 60°C to 100°C the efficiency keeps dropping. Is that linear too?
🎓
Yes — III-V triple-junctions have a temperature coefficient β of about -0.13 %/°C, which is actually a magnitude better than silicon's -0.45 %/°C. That's a big strength of III-V: efficiency degrades slowly with heat. But "doesn't lose much efficiency" is not the same as "doesn't break". Above ~120°C the solder joints and the anti-reflection coatings start to degrade quickly. In practice you target a junction temperature below 80°C. The tool clamps efficiency at 50% even though the underlying formula gives 50.4% at the default settings — that's a deliberately conservative ceiling close to the 47.1% world record.
🙋
So is CPV a dead technology now?
🎓
The commercial flat-utility market collapsed, but research is alive. Space solar arrays (satellites, the ISS) are still almost exclusively III-V triple-junction. Hybrid CPV/thermal systems that take 60-80°C cooling water for district heating or absorption chillers are being demoed by Insolight and others. Ultra-high concentration (>2000 suns) for solar-thermochemical fuel production is being studied at ETH Zurich and DLR. As you can see from the tool, once you understand the DNI dependence, tracking precision and cooling load, CPV is still the highest-efficiency solar system in the world when it's used in the right place.
Frequently asked questions
CPV uses Fresnel lenses or mirrors to concentrate sunlight by 500-2000 times onto a small, ultra-high-efficiency III-V compound semiconductor cell (such as an InGaP/GaAs/Ge triple junction). Where a flat-plate silicon panel reaches 22-26%, CPV modules deliver 35-46% and research cells have hit 47.1% (NREL/Spectrolab 6-junction, 2020). The trade-offs are that only direct normal irradiance (DNI) is usable, a 2-axis solar tracker is mandatory, and the waste-heat density is hundreds of times higher than silicon, so active cooling (often liquid cooling) is required.
Because the open-circuit voltage V_oc grows logarithmically with illumination: V_oc(X) ≈ V_oc(1) + (kT/q)·ln(X). At X=500 suns and room temperature this is about 160 mV, giving roughly a 5% voltage bonus on a triple-junction cell with V_oc=3 V. This tool approximates that with voltageBoostFactor = 1 + 0.06·ln(X). However the gain is partly cancelled by the temperature rise of the cell, so a temperature coefficient β ≈ -0.13 %/°C and adequate cooling are also assumed.
Locations with annual DNI above roughly 2000 kWh/m² — the Sahara, the Atacama, the US Southwest, central Australia and the Middle East. In humid or cloudy regions such as Japan and eastern China, scattered light dominates and CPV (which uses only DNI) loses to flat silicon (which uses GHI). Falling silicon module prices and the cost of 2-axis tracking caused most commercial CPV businesses to wind down after 2015, with Soitec and Amonix exiting the market. Sumitomo Electric and ISFOC continue research and field testing.
At X=500 with DNI=950 W/m², 85% optics and 50% cell efficiency the incident flux at the cell is around 400 kW/m² (40 W/cm²), of which about 20 W/cm² must be removed as heat. A standard silicon panel only sees about 0.07 W/cm² of waste heat, so CPV is hundreds of times worse. The tool uses these thresholds: below 5 W/cm² a passive heatsink is enough, below 15 W/cm² forced-air cooling is needed, above that liquid cooling (microchannel or jet impingement) is mandatory. This is significantly tougher than CPU thermal management (around 1 W/cm² TDP).
Real-world applications
Desert utility-scale plants: Spain (ISFOC), Arizona and California in the US, central Australia and Saudi Arabia all hosted commercial HCPV deployments — Amonix shipped plants up to 30 MW and Soitec rolled out its Concentrix CX-S530 system. In sites with annual DNI above 2200 kWh/m² these arrays out-yielded silicon for several years. Most have now been decommissioned or replaced by silicon as polysilicon prices collapsed after 2015.
Space solar arrays: Satellites, the International Space Station and deep-space probes still rely almost entirely on III-V triple-junction cells. Weight is the binding constraint, so the 30-35% efficiency of InGaP/GaAs/Ge cells is essential. Mini-concentration (10-20 suns) is being studied for space too; AZUR SPACE, SolAero, Sharp and Spectrolab are the main suppliers.
Hybrid CPV/thermal (CPV-T): Pulling 60-80°C cooling water off the back of the cell and using it for district heating, absorption chillers or low-temperature process heat raises the overall electrical+thermal efficiency to 70-80%. Hospitals, hotels and food-processing plants are typical pilots; Insolight, Solar Junction and several university spin-offs are pushing this market.
Sun-to-fuels: Ultra-high concentration (>2000 suns) drives reactor temperatures past 1500°C for direct thermochemical hydrogen production or CO₂ reduction. ETH Zurich, DLR and Synhelion combine III-V cells with thermal receivers in these hybrid concepts; CPV optics is the front-end of all of them.
Common misconceptions and pitfalls
The biggest misconception is that CPV is always more economical than silicon because it is more efficient. CPV's efficiency advantage only applies to the direct beam (DNI). In cloudy or humid climates the annual yield falls below silicon, which can use diffuse light too. Japan's annual DNI is around 1300 kWh/m², below the GHI silicon receives (~1500 kWh/m²). The economics depend on "efficiency × usable irradiance × cost of tracking + cooling". Site selection is the single biggest call you make with CPV. Drop the DNI slider in this tool from 950 to 400 (cloudy-day equivalent) and the instantaneous power roughly halves.
The second pitfall is assuming more concentration is always better. V_oc does grow logarithmically with X, but cell temperature also rises and the temperature coefficient β ≈ -0.13 %/°C eats into the gain. On top of that, series-resistance losses scale with the square of the current, so practical CPV systems plateau around 500-1000 suns. The tool clamps efficiency at 50% even when you push X to 2000 — that is a deliberately conservative limit that respects the 47.1% world record (6-junction at 143 suns).
The third trap is underestimating the cooling load. At the default settings (X=500, DNI=950, 1 cm² cell) the tool returns 20 W/cm² of waste heat. Only liquid cooling with heat-transfer coefficients of hundreds of W/cm²·K can handle that. A naive passive heatsink would push the junction past 200°C in minutes and the solder joints, anti-reflection coatings and ARC films would degrade rapidly. In design you must evaluate the cell-to-coolant thermal resistance R_th [K·cm²/W] together with the cooling loop and verify that the target junction temperature (typically below 80°C) is met. Fraunhofer ISE has demonstrated R_th ≈ 0.5 K·cm²/W with copper microchannel coolers.
How to Use
Set concentration ratio (500–2000x) using the slider or numeric input; this determines how many suns of direct normal irradiance (DNI) strike the III-V triple-junction cell.
Enter direct normal irradiance (DNI) in W/m², typically 800–900 W/m² under peak desert conditions, and cell efficiency at 1 sun (reference ~42% for InGaAs/GaAs/Ge).
Specify active cell area in cm² (common values: 0.3–1.5 cm² for Fresnel-lens designs); simulator calculates concentrated flux, efficiency derating, electrical output, and thermal load requiring active cooling.
Worked Example
A CPV module with 800x concentration ratio, DNI = 850 W/m², cell efficiency at 1 sun = 42%, and active cell area = 0.8 cm²: concentrated flux = 6800 W/cm² (680 MW/m²), cell efficiency drops to 38% at 800 suns due to series resistance, electrical output per cell ≈ 2.06 W, waste-heat flux ≈ 3.4 W/cm², necessitating microchannel cooling at flow rates ~0.5–1.0 L/min per cell to maintain junction temperature below 85°C.
Practical Notes
Fresnel lens tracking tolerance: misalignment >1° typically reduces concentration uniformity by 15–25%; use dual-axis trackers with ±0.1° accuracy in commercial installations.
Spectral response: InGaP/InGa(As)P/Ge cells benefit from higher DNI in desert sites (direct beam >0.8 kWh/m²/day); diffuse irradiance contributes <5%, limiting CPV to arid regions.
Thermal management cost: silicon carbide or copper microchannel heat sinks add 8–12% to module BOM; junction temperature rise of 10 K typically reduces efficiency by 0.5–0.8%.