Solar Simulators

1.1Definition and Purpose

A solar simulator is an instrument that produces artificial illumination approximating the spectral irradiance of natural sunlight under controlled, repeatable laboratory conditions. The core function is to replace the sun as a calibrated light source so that photovoltaic devices, materials, and optical systems can be characterized without dependence on geographic location, weather, time of day, or season [1, 2].

The need for standardized indoor testing arose in the late 1970s when photovoltaic research began scaling beyond isolated laboratory demonstrations. In 1978, ASTM formed Subcommittee E44.09 specifically to address the problem of comparing solar cell efficiency measurements taken under different light sources in different laboratories. The result was ASTM E927, the first standard to define how a solar simulator should be measured and classified [2]. The International Electrotechnical Commission followed with IEC 60904-9 in 1995, and the Japanese Industrial Standards organization published JIS C 8912. These three standards remain the governing documents for solar simulator classification worldwide [1, 2].

1.2Why Indoor Testing Matters

Outdoor solar irradiance varies continuously. Cloud cover, aerosol loading, water vapor content, solar zenith angle, altitude, and surface albedo all modulate the spectral power density reaching a test device. Even on a clear day at a fixed location, the solar spectrum changes from sunrise to sunset as the air mass traverses from high values at the horizon to a minimum at solar noon. This variability makes it impossible to compare photovoltaic measurements taken at different times and places unless a reference standard is imposed.

Solar simulators solve this problem by providing a light source whose spectral content, spatial uniformity, and temporal stability are quantified against a defined reference spectrum. The standard test conditions (STC) adopted by the photovoltaic industry specify a cell temperature of 25 °C, total irradiance of 1000 W/m², and the AM1.5G reference spectrum [1, 3]. Every efficiency number reported for a terrestrial solar cell is measured — or corrected to — these conditions.

1.3Scope Beyond Photovoltaics

Although solar simulators were developed primarily for photovoltaic testing, their applications extend across multiple disciplines. Materials scientists use solar simulators for accelerated weathering studies, exposing polymers, coatings, and composites to calibrated UV and visible radiation to predict outdoor degradation lifetimes. Photochemists and environmental engineers employ them to drive photocatalytic reactions — water splitting, CO₂ reduction, and pollutant degradation — under reproducible illumination. Space agencies rely on AM0-class simulators for thermal vacuum testing of spacecraft components and satellite qualification. In photobiology and cosmetics research, solar simulators provide standardized UV exposure for sunscreen efficacy testing and skin photosensitivity studies [7, 10].

This guide covers the physics of the solar spectrum and air mass convention, the classification standards that define simulator quality, the light source technologies and optical systems used to build solar simulators, the measurement procedures for characterizing their performance, and practical guidance for selecting the right instrument for a given application.

2.1The Sun as a Radiation Source

The sun radiates approximately as a blackbody at an effective surface temperature of about 5800 K. The resulting spectral power distribution spans from the far ultraviolet through the visible and into the mid-infrared, with peak spectral irradiance near 500 nm — consistent with the Wien displacement law prediction of λ_max ≈ 2898 μm·K / 5800 K ≈ 500 nm. At the mean Earth–Sun distance (1 AU), the total irradiance integrated across all wavelengths is the solar constant, approximately 1366 W/m² [3, 5].

This extraterrestrial spectrum is designated AM0 (air mass zero) because the light has traversed no atmosphere. AM0 is the reference for space applications and satellite solar cell qualification [3].

2.2Atmospheric Attenuation

As sunlight enters the Earth's atmosphere, three mechanisms attenuate and reshape the spectrum. Rayleigh scattering by nitrogen and oxygen molecules preferentially removes short-wavelength radiation, reducing UV and blue content while giving the sky its blue color. Molecular absorption by ozone (O₃) removes UV-B and UV-C below approximately 300 nm, while water vapor (H₂O) and carbon dioxide (CO₂) create deep absorption bands in the near-infrared at approximately 720, 820, 940, 1100, 1380, and 1870 nm. Aerosol scattering by dust and particulates further attenuates the beam across all wavelengths, with a wavelength dependence that varies with particle size distribution [5].

The cumulative effect is that the spectrum at the Earth's surface is depleted in the ultraviolet and punctuated by absorption features in the infrared, with a total integrated irradiance substantially lower than the solar constant.

2.3Air Mass Definition

The air mass coefficient quantifies the path length of sunlight through the atmosphere relative to the shortest possible path — that of the sun directly overhead at sea level. For a plane-parallel atmosphere, the air mass is:

Where: θ_z = solar zenith angle (the angle between the sun and the vertical direction).

This approximation is valid for zenith angles up to about 75°. At larger angles (sun near the horizon), Earth's curvature becomes significant. The Kasten–Young empirical formula provides better accuracy across the full range [5]:

Key air mass values and their physical meanings:

AM0: No atmosphere — the extraterrestrial spectrum, with integrated irradiance of approximately 1366 W/m² [3, 5].

AM1: Sun directly overhead at sea level (θ_z = 0°). The shortest atmospheric path. Useful for equatorial and tropical regions.

AM1.5: The standard for terrestrial photovoltaic testing, corresponding to θ_z = 48.2°. This was selected in the 1970s because it provides a reasonable annual average for mid-latitude regions where major population centers and solar installations are concentrated [3, 5].

2.4Reference Spectra: AM1.5G, AM1.5D, and AM0

The standard reference spectra for photovoltaic testing are defined in ASTM G173-03 and IEC 60904-3 [3]. Two terrestrial spectra are defined at AM1.5:

AM1.5G (Global): Includes both direct beam and diffuse (sky-scattered) radiation incident on a 37°-tilted, equator-facing surface. The integrated irradiance has been normalized to exactly 1000 W/m² for convenience, although the raw atmospheric model produces approximately 970 W/m². AM1.5G is the standard spectrum for flat-plate photovoltaic module testing and is by far the most commonly used reference for solar simulators [3].

AM1.5D (Direct + circumsolar): Includes only the direct beam from the sun plus the circumsolar component within a 2.5° disk around the solar disk. The integrated irradiance is approximately 900 W/m². AM1.5D is the standard for concentrating photovoltaic (CPV) systems [3].

AM0: The extraterrestrial spectrum with an integrated irradiance of approximately 1366 W/m² (the solar constant). Used for space applications [3, 5].

The atmospheric conditions assumed for the terrestrial spectra include a standard set of parameters representative of the contiguous United States: a water vapor column of 1.42 cm, total ozone of 0.34 atm·cm, and a rural aerosol model with visibility of 23 km. These spectra were generated using the SMARTS (Simple Model of the Atmospheric Radiative Transfer of Sunshine) code developed by Gueymard [5].

3.1The Three Governing Standards

Three international standards define the classification framework for solar simulators used in photovoltaic testing [1, 2]:

IEC 60904-9:2020 — Published by the International Electrotechnical Commission. The most widely adopted global standard, updated in 2020 with significant revisions including the introduction of Class A+ and two new metrics.

ASTM E927-19 — Published by ASTM International. The original solar simulator standard, first drafted in 1979 by Subcommittee E44.09. Widely used in North America.

JIS C 8912 — Published by the Japanese Industrial Standards Committee. Used primarily in Japan; requirements are similar to the IEC standard.

While the three standards differ in detail, their overall approach is the same: a solar simulator is rated on three independent performance metrics, each receiving a letter grade. The combined rating is expressed as a three-letter code [1, 2, 4].

3.2The Three-Letter Classification System

Every solar simulator classification consists of three letters, reported in the following fixed order [1]:

1. Spectral match — how closely the simulator's output spectrum matches the reference solar spectrum across defined wavelength bins.

2. Spatial non-uniformity — how evenly the irradiance is distributed across the illuminated test plane.

3. Temporal instability — how stable the irradiance remains over the measurement period.

Each metric receives a grade of A+, A, B, or C (A+ was introduced by IEC 60904-9:2020 and is not present in all standards). A simulator classified as "ABA" has Class A spectral match, Class B spatial non-uniformity, and Class A temporal instability. The shorthand "Class AAA" means the simulator achieves Class A in all three metrics [1, 2].

Note: ASTM E927-19 uses a single temporal instability metric (≤ 2% for Class A) rather than separate short-term and long-term values. ASTM also defines Class U (unclassified) for simulators falling outside the C thresholds. The ASTM Class A spatial non-uniformity threshold is ≤ 3%, slightly more relaxed than IEC's ≤ 2% [2].

3.3Spectral Match and Wavelength Bins

The spectral match assessment divides the simulator's wavelength range into discrete bins and compares the fraction of total irradiance in each bin against the reference spectrum. The spectral match ratio for each bin is:

A ratio of 1.00 represents a perfect match. For Class A, every bin must fall within 0.75–1.25; for Class A+, within 0.875–1.125. The overall spectral match class is determined by the worst-performing bin [1, 2].

The IEC 2020 revision extended the spectral range to 300–1200 nm and redesigned the bins to contain approximately equal fractions of the total irradiance (~16.67% each). This equal-irradiance approach was adopted because silicon photovoltaic technology had expanded its spectral response range, and emerging technologies such as perovskite–silicon tandems require accurate spectral matching over a broader wavelength range [1].

3.4The 2020 IEC Standard Updates

The 2020 revision of IEC 60904-9 introduced several significant changes beyond the Class A+ tier [1]:

Spectral coverage (SPC): A new metric that quantifies what fraction of the reference spectrum is meaningfully reproduced by the simulator. SPC is defined as the percentage of wavelengths within the evaluation range where the simulator's spectral irradiance exceeds 10% of the reference irradiance at that wavelength. A broadband source such as a xenon arc lamp typically achieves high SPC, while an LED array with gaps between emission peaks may have lower SPC despite achieving Class A spectral match in the binned assessment [1].

Spectral deviation (SPD): A metric that captures how much the simulator's spectrum varies from the reference within each bin, beyond what the binned match ratio reveals. SPD is sensitive to spiky or non-smooth spectra that may average out over a wide bin but still differ significantly from sunlight at individual wavelengths [1].

These two metrics are not yet required for classification (a simulator can still be rated AAA without reporting SPC or SPD), but manufacturers are encouraged to report them. They are particularly relevant for LED-based simulators, which can achieve excellent binned spectral match while having low spectral coverage due to the discrete nature of LED emission [1].

4.1Xenon Arc Lamps

Xenon arc lamps are the most widely used light source in solar simulators and have dominated the field since the late 1970s. A xenon arc lamp consists of a sealed quartz envelope filled with high-pressure xenon gas (typically 10–30 atm), with tungsten electrodes that sustain a plasma arc. The resulting emission approximates a blackbody at approximately 5800 K — close to the sun's effective temperature — providing a broadband spectrum from the deep ultraviolet (below 200 nm) through the visible and into the mid-infrared (beyond 2500 nm) [6, 7, 9].

The principal advantage of the xenon arc is its native broadband emission. Even without optical filtering, the unfiltered spectrum provides a reasonable match to sunlight across the UV, visible, and near-IR regions. With appropriate air mass filters (Section 5), the match can be refined to achieve Class A or A+ spectral performance [9]. Xenon lamps also offer high luminous intensity from a compact arc, enabling collimated, high-irradiance beams suitable for both standard (1 sun) and concentrated (multi-sun) testing.

The primary disadvantage is a series of strong emission lines in the 800–1000 nm region, corresponding to electronic transitions in excited xenon atoms. These lines produce excess irradiance in the near-IR that must be filtered to meet Class A spectral match requirements. Additional disadvantages include high power consumption, short lamp lifetime (typically 1000–2000 hours), risk of explosive lamp failure due to the high internal gas pressure, ozone generation from UV emission below 240 nm, and substantial infrared heat load delivered to the test plane [6, 7, 10].

Short-arc xenon lamps (arc gap < 10 mm) are preferred for solar simulators because their compact arc produces a near-point source, allowing efficient collection and collimation by ellipsoidal reflectors. Long-arc versions are occasionally used for large-area applications but produce lower collimation [7, 9].

4.2Metal Halide Lamps

Metal halide lamps are a type of high-intensity discharge (HID) source that uses a mercury vapor arc seeded with metal halide additives to produce a broad spectrum in the 4000–6000 K color temperature range. Approximately 37% of solar simulators built for thermal applications since the 1970s have used metal halide lamps, making them the second most common source type historically [7].

Metal halide lamps offer a lower purchase cost and simpler power supply requirements compared to xenon. However, they produce lower collimation (more divergent beams), making them less suitable for applications requiring tight beam angles or high spatial uniformity over a well-defined test plane. Their spectral output is also less smooth than xenon, with more pronounced emission line structure. Lamp-to-lamp spectral variability is higher than for xenon, and the spectrum shifts more noticeably as the lamp ages [7, 10].

4.3Quartz Tungsten Halogen (QTH) Lamps

Quartz tungsten halogen lamps operate as incandescent sources with a tungsten filament enclosed in a quartz envelope containing a halogen gas that regenerates the filament surface, extending lamp life. They are inexpensive, produce a smooth blackbody-like spectrum, and require only a simple DC power supply [7].

The fundamental limitation is that the tungsten filament temperature cannot exceed approximately 3400 K, well below the sun's 5800 K. This results in a spectrum that is strongly deficient in the blue and ultraviolet regions and has excessive infrared output relative to AM1.5G. As a standalone source, a QTH lamp cannot achieve Class A spectral match. QTH lamps are therefore used primarily as supplementary sources in multi-source (hybrid) solar simulators, where they extend the infrared coverage beyond what a xenon lamp alone provides [7, 9].

4.4LED Arrays

Light-emitting diode solar simulators have gained significant traction since their conceptual introduction in 2003, with rapid development accelerating from 2012 onward as LED luminous efficacy improved dramatically. An LED solar simulator uses an array of many spectrally distinct LED types — typically 10 to 30 different emitter wavelengths — combined using calibrated optics to approximate the target solar spectrum [6, 10].

Each individual LED has a narrow emission bandwidth (typically 10–100 nm FWHM), so the composite spectrum consists of a series of peaks rather than a smooth continuum. The number and spectral diversity of LEDs in the array directly determine the quality of the spectral match. Modern high-end LED simulators can achieve Class A and even Class A+ spectral match across the 300–1100 nm range [6, 10].

The advantages of LED solar simulators are substantial. Lamp lifetime is 50,000–100,000 hours — roughly 50–100× longer than xenon arc lamps — dramatically reducing maintenance cost and downtime. LED arrays can be spectrally tuned by adjusting the drive current to individual LED channels, allowing the user to match different reference spectra (AM0, AM1.5G, AM1.5D) or even create custom spectra without changing physical filters. Turn-on time is nearly instantaneous (no 30–60 minute warm-up as required for xenon), the power consumption is lower, there is no explosion risk, and no ozone is generated. LED arrays can also be pulsed at microsecond timescales for flash-mode operation [6, 10].

The primary limitations are the spectral gaps between LED emission peaks (which reduce spectral coverage even when binned spectral match is Class A), difficulty reproducing the spectrum below 360 nm and above 1100 nm due to limited LED availability in the deep UV and IR, and sensitivity of LED emission wavelength and intensity to junction temperature, which requires active thermal management [6, 10].

4.5Multi-Source Hybrids and Other Sources

To extend spectral coverage, some solar simulators combine multiple source types. A common configuration pairs a xenon arc lamp with one or more halogen lamps: the xenon provides the UV-through-near-IR spectrum, while the halogen extends the simulation further into the infrared beyond 1100 nm. These dual-source systems can achieve the highest spectral fidelity across the broadest wavelength range and are used in calibration laboratories for the most demanding measurements [9].

Supercontinuum lasers — high-power, broadband fiber sources spanning from the visible to the infrared — have been explored as solar simulator sources. They offer high spectral brightness and excellent collimation. However, their illuminated area is extremely small (typically a few millimeters), restricting their use to concentrated photovoltaic testing or single-cell characterization. They remain a niche application rather than a mainstream simulator source [7].

5.1Reflector and Collection Optics

The light source alone does not produce a uniform, collimated beam suitable for testing. The optical system collects the divergent lamp emission, conditions it for spectral content, homogenizes the spatial distribution, and delivers it to the test plane with controlled divergence. The reflector is the first element in this chain [9].

Most xenon-based solar simulators use an ellipsoidal reflector — a mirror whose cross-section is an ellipse — positioned so that the lamp arc sits at one focus. The reflector collects a large solid angle of lamp emission and converges it toward the second focus, where subsequent optical elements are positioned. The ellipsoidal geometry provides high collection efficiency (often 60–80% of the lamp's 4π emission is captured) while concentrating the beam to a manageable cone angle for downstream optics [9].

Parabolic reflectors are used when a more collimated output is needed directly from the reflector, such as in large-area space simulation chambers. A parabolic mirror with the lamp at its focus produces a nominally parallel beam, but the finite size of the arc (not a true point source) introduces residual divergence. Off-axis parabolic configurations avoid obscuration by the lamp housing but introduce coma aberration that can degrade uniformity [9].

5.2Beam Homogenization

The raw beam from a reflector–lamp assembly is inherently non-uniform. The arc has spatial structure (hot spots near the cathode tip), the reflector has a central obscuration (the lamp body blocks some reflected light), and the angular distribution is not Lambertian. Achieving Class A spatial non-uniformity (≤ 2% per IEC) requires active homogenization [9].

Integrating rods (light pipes): A rectangular or cylindrical rod of fused silica or other high-transmission glass captures the converging beam at one end. Light propagates through the rod by total internal reflection, undergoing multiple reflections from the walls. Each reflection remixes the angular and spatial distribution, so the output face of the rod presents a highly uniform irradiance profile. Integrating rods are the most common homogenizer in high-quality benchtop simulators [9].

Fly-eye (microlens) arrays: Two matched arrays of small lenslets — one at the input, one at the output — divide the beam into sub-apertures and superimpose their images at the test plane. The statistical superposition of many sub-beams averages out any spatial non-uniformity in the source. Fly-eye integrators preserve more light than rods (no absorption or scattering losses from wall reflections) and are favored for large-area simulators [9].

Köhler illumination: The classical Köhler geometry, borrowed from microscopy, uses a condenser lens to image the source into the pupil of a projection lens, which in turn projects a uniformly illuminated field onto the test plane. This decouples source non-uniformity from the illuminated field, producing uniform irradiance even from a spatially inhomogeneous source.

5.3Air Mass Filters and Spectral Conditioning

Air mass filters modify the raw lamp spectrum to match a specific reference spectrum. For xenon lamps, this primarily means attenuating the excess near-IR emission lines between 800 and 1000 nm and adjusting the UV cutoff to match atmospheric absorption. Filters are typically fabricated from specialized absorption glasses and are designated by their target spectrum: AM0, AM1.0, AM1.5G, AM1.5D, or AM2.0 [9].

Complete elimination of the xenon emission lines while preserving the rest of the spectrum is not physically achievable with absorption filters. The filters reduce the lines to an acceptable level within the binned spectral match tolerances. Over time, prolonged exposure to high-intensity UV and thermal loading can degrade filter transmission, causing a gradual spectral drift that may eventually push the simulator out of its rated class. Regular spectral verification is essential [9].

LED solar simulators do not require air mass filters because their spectrum is constructed additively from individual LED channels. Spectral adjustment is performed electronically by varying channel drive currents. This is one of their principal advantages — no consumable optical filters to degrade [6, 10].

5.4Collimation

Natural sunlight arrives at the Earth's surface with an apparent angular diameter of approximately 0.53° (32 arc-minutes), corresponding to a half-angle divergence of about 0.27°. For many photovoltaic test applications, achieving this tight collimation is not necessary — standard cell I-V measurements are insensitive to modest beam divergence because the cell is a planar receiver. However, for concentrating photovoltaic (CPV) testing, space environment simulation, and solar thermal collector characterization, matching the solar divergence angle is critical [9].

Achieving a collimation half-angle below 1° in a solar simulator requires a large-aperture collimating lens or mirror with a long focal length, combined with a spatially compact source (ideally a point source at the focal point). In practice, the finite arc size of a xenon lamp limits the achievable collimation. Highly collimated simulators (< 0.5° half-angle) typically use parabolic or Cassegrain-style reflective collimators and are significantly larger and more expensive than standard benchtop simulators [9].

5.5Illumination Area and Beam Geometry

The illumination area — the region of the test plane over which the simulator meets its rated spatial uniformity specification — is a primary selection criterion. Illumination areas range from 20 × 20 mm (suitable for single small-area cells) up to 305 × 305 mm (12 × 12 inches) for benchtop research simulators. Production-line module testers can illuminate areas of 1 × 2 meters or larger. Space simulation chambers with collimated beams can reach several meters in diameter [9].

The illumination area is constrained by the available lamp power and the optical system's collection efficiency. Increasing the area while maintaining 1 sun irradiance (1000 W/m²) requires proportionally more optical power delivered to the test plane. For a typical xenon arc system, a 150 W lamp can illuminate a 2 × 2 inch area at 1 sun, while a 1600 W lamp is needed for 12 × 12 inches. LED arrays scale by adding more emitter modules, but thermal management becomes more challenging at larger areas [9].

Spatial non-uniformity generally degrades as the illumination area increases. The optical design must trade off illuminated area against uniformity: a simulator rated Class A non-uniformity over 50 × 50 mm may only achieve Class B over 100 × 100 mm. Manufacturers specify the rated area over which the classification applies.

6.1Spectral Match Measurement

Spectral match is assessed by measuring the simulator's spectral irradiance at the test plane using a calibrated spectroradiometer, then integrating the irradiance over each wavelength bin and computing the spectral match ratio R_SM,i as defined in Section 3.3. Per IEC 60904-9:2020, the measurement must be performed at four locations within the test area (the center and three off-axis positions), and the spectral match is evaluated at each location [1].

The spectroradiometer must be calibrated against a traceable standard (typically an FEL lamp calibrated at NIST or an equivalent national metrology institute). The wavelength resolution should be sufficient to resolve spectral features — typically 1–5 nm — although the classification assessment integrates over broad bins. The measurement must cover the full evaluation range (400–1100 nm for ASTM, 300–1200 nm for IEC A+ classification) [1, 2].

6.2Spatial Non-Uniformity

Spatial non-uniformity quantifies the variation in irradiance across the illuminated test area. The measurement procedure involves scanning a detector (or an array of detectors) across the test plane at multiple positions and recording the irradiance at each point [1, 2].

Where: E = set of irradiance measurements at all test positions (W/m²).

ASTM E927-19 requires a minimum of 64 measurement positions, distributed uniformly across the rated illumination area. The area of each individual measurement aperture must be between 0.5 and 1.0 times the total test area divided by the number of positions [2]. IEC 60904-9:2020 requires a minimum of 64 positions for large areas but permits fewer for small test planes, with specific guidance on the grid layout [1].

6.3Temporal Instability

Temporal instability measures the variation of irradiance over time. During a steady-state measurement, the irradiance at the test plane is recorded continuously and the instability is calculated from the extremes of the time series [1, 2]:

IEC 60904-9:2020 distinguishes between short-term instability (STI), evaluated over the duration of a single I-V sweep or measurement, and long-term instability (LTI), evaluated over a longer period (typically 30 minutes to several hours). The STI threshold is more stringent than LTI because rapid fluctuations directly corrupt individual measurements [1].

6.4Spectral Coverage and Spectral Deviation

The 2020 IEC standard introduced two additional informational metrics [1]:

SPC addresses a blind spot in the binned spectral match assessment: an LED simulator could achieve Class A spectral match (correct integrated irradiance in each bin) while having zero emission at certain wavelengths within a bin, provided the total integrated energy is correct. SPC reveals this by reporting the fraction of wavelengths with meaningful illumination. A xenon arc lamp typically achieves SPC > 95%, while an LED array may achieve 60–80% depending on the number and spacing of LED channels [1].

Spectral deviation (SPD) quantifies the wavelength-by-wavelength departure of the simulator spectrum from the reference. It is calculated as a weighted root-mean-square deviation across the evaluation range, sensitive to sharp spectral features that may average out in the broad-bin assessment. SPD complements SPC: a simulator with high SPC but high SPD has broad coverage but poor point-by-point match (e.g., xenon with strong emission lines), while a simulator with low SPC but low SPD has narrow but accurate coverage (e.g., a well-tuned LED array over its emission bands) [1].

7.1Reference Cells

A reference cell is a calibrated photovoltaic device used to set the irradiance level of a solar simulator to the standard test condition of 1000 W/m². The reference cell's short-circuit current (I_sc) has been measured under a primary reference (either a calibrated lamp or outdoor conditions traceable to a world photovoltaic scale) and is certified per IEC 60904-2 [4, 8].

The reference cell is placed at the test plane, and the simulator's intensity is adjusted until the reference cell produces its certified I_sc value. At that point, the irradiance at the test plane is defined to be 1000 W/m² — not because the absolute irradiance has been measured directly, but because the reference cell "sees" the equivalent of 1 sun given its spectral response [4, 8].

This approach works perfectly only when the reference cell has exactly the same spectral response as the device under test (DUT). When the spectral responses differ — as they always do when the reference cell and DUT are different technologies (e.g., a silicon reference cell used to calibrate irradiance for testing an organic solar cell) — the spectral content of the simulator matters. The simulator spectrum is not identical to AM1.5G, and the two devices weight the spectral differences differently. This is the spectral mismatch problem [4, 8].

7.2The Spectral Mismatch Factor

The spectral mismatch factor M quantifies the error introduced when the simulator spectrum and the spectral responses of the reference and test cells are not perfectly matched. It is defined by IEC 60904-7 as [4]:

When M = 1, there is no mismatch — either because the simulator perfectly matches AM1.5G, or because the reference and test cells have identical spectral responses, or both. In practice, M typically deviates from unity by less than a few percent for well-matched systems (|M − 1| < 1–2% for Class A xenon simulators with matched silicon reference cells) [8].

The corrected short-circuit current of the test cell is:

7.3When Mismatch Correction Matters

For routine production testing of silicon cells using a silicon reference cell on a Class A xenon simulator, M is typically within 0.5% of unity and the correction is often negligible relative to other measurement uncertainties. The correction becomes significant in several scenarios [4, 8]:

1. Different absorber technologies: Testing a perovskite or organic cell (strong UV/visible absorption, no near-IR response) using a silicon reference cell (response extending to 1100 nm) on a xenon simulator with strong 800–1000 nm emission lines. The silicon reference cell "sees" excess irradiance in the near-IR that the test cell does not absorb, leading to an underestimate of the test cell's efficiency if M is not applied.

2. Multi-junction cells: Each sub-cell has a different spectral response band. Accurate characterization requires a spectrally adjustable simulator (typically multi-source LED) that can independently set the irradiance in each sub-cell's absorption range.

3. LED simulators with spectral gaps: Even if binned spectral match is Class A, the discrete LED emission peaks may create localized spectral mismatches that are invisible to the binned assessment but affect the DUT response.

8.1Continuous-Wave (Steady-State) Simulators

Steady-state solar simulators operate with continuous illumination — the lamp runs at a stable output for the entire duration of testing. Xenon arc lamps and LED arrays are the most common steady-state sources. The primary advantage is simplicity: the irradiance is constant, data acquisition proceeds at whatever speed the measurement electronics require, and temporal instability is straightforward to characterize over extended periods [2, 9].

The principal disadvantage of steady-state operation is thermal loading. A 1-sun irradiance of 1000 W/m² delivered continuously to a test plane deposits significant thermal energy into the device under test. For small cells with good thermal contact to a temperature-controlled chuck, this is manageable. For large modules or thermally sensitive devices (organic cells, perovskites), the temperature rise during a prolonged I-V sweep can alter the device characteristics during the measurement itself. Active cooling of the test plane and fast measurement techniques mitigate this issue [9].

8.2Pulsed (Flash) Simulators

Pulsed solar simulators deliver a brief, intense flash of light — typically lasting 1–100 milliseconds — during which the I-V curve is swept electronically. The flash is generated by discharging a capacitor bank through a xenon flash tube (similar to photographic flash units) or by pulsing an LED array [2, 9].

The principal advantage is dramatically reduced thermal loading. The total energy delivered to the DUT during a flash is orders of magnitude less than during a continuous measurement, allowing testing of large modules and thermally sensitive cells without significant temperature rise. Flash simulators are the standard for production-line module testing, where throughput requires fast cycle times [2, 9].

The principal disadvantage is the inherent transience of the illumination. The spectral content and intensity of a xenon flash change during the pulse as the plasma heats and cools. Multi-flash techniques capture different portions of the I-V curve on successive flashes, requiring careful synchronization between the flash trigger and the electronic load. LED flash simulators mitigate the spectral transience issue because LED emission is less sensitive to drive current variations on short timescales [2, 6].

Flash simulators can achieve very high irradiances — up to several thousand suns — by concentrating the flash energy into a short pulse. This makes them essential for characterizing concentrator photovoltaic (CPV) cells that are designed to operate under concentrated sunlight [9].

8.3Illumination Area Considerations by Type

Illumination area capabilities differ by simulator type and are closely tied to the operating mode:

Steady-state xenon simulators range from 2 × 2 inches (51 × 51 mm) at 150–450 W lamp power to 12 × 12 inches (305 × 305 mm) at 1600 W. Larger areas require multiple lamp modules or moving the test stage through a smaller beam. Steady-state LED systems scale to approximately 200 × 200 mm with current technology before thermal management of the dense LED array becomes limiting [9].

Pulsed xenon flash simulators can illuminate areas up to 2 × 2 meters for production-line module testing because the short pulse duration limits the total optical power that must be sustained. The tradeoff is that spatial uniformity is typically harder to maintain over such large areas (Class B or C uniformity is common for large-format flash testers) [2, 9].

9.1Photovoltaic Cell and Module Characterization

The primary application of solar simulators is measuring the current–voltage (I-V) characteristics of photovoltaic devices under standard test conditions. From the I-V curve, all key performance parameters are extracted: short-circuit current (I_sc), open-circuit voltage (V_oc), maximum power point (P_max), fill factor (FF), and power conversion efficiency (η). Every published solar cell efficiency — from silicon to perovskite to multi-junction — is measured under solar simulator illumination referenced to AM1.5G at 1000 W/m² [1, 2, 8].

For standard single-junction silicon cells, a Class AAA xenon simulator with a matched silicon reference cell is the most common configuration, and spectral mismatch correction is typically small. For emerging technologies — organic photovoltaics, perovskite cells, and especially perovskite–silicon tandems and III-V multi-junction cells — spectral fidelity becomes critical. Organic and perovskite absorbers respond primarily in the UV and visible range, making them sensitive to the UV content of the simulator. Multi-junction cells require independent control of irradiance in each sub-cell's absorption band, which has driven the adoption of spectrally tunable LED simulators for this class of devices [6, 8].

9.2Accelerated Weathering and Materials Durability

Solar simulators are used for accelerated aging studies in which materials, coatings, textiles, and polymers are exposed to calibrated UV and visible radiation to predict outdoor degradation lifetimes. The relevant standards (ASTM G154, G155, ISO 4892) specify irradiance levels and spectral windows for weathering tests. Xenon-based weathering instruments with daylight or window-glass filters are the industry standard for this application, as they most closely replicate the full solar spectrum including UV-A and UV-B [10].

9.3Photocatalysis and Photoelectrochemistry

Photocatalysis research — water splitting for hydrogen production, CO₂ reduction, pollutant degradation — requires reproducible illumination at defined spectral content and intensity. Solar simulators provide the controlled light source, typically using a xenon lamp with an AM1.5G filter to simulate terrestrial conditions. The irradiance may be varied from sub-sun to multi-sun levels to study intensity-dependent kinetics [10].

9.4Space Environment Simulation

Spacecraft components, satellite solar panels, and thermal control surfaces are qualified in ground-based space simulation chambers that replicate the extraterrestrial radiation environment. These facilities use AM0-class solar simulators — large-area, highly collimated beams matching the extraterrestrial solar spectrum. The collimation requirement (< 1° half-angle to approximate the solar divergence) and the need for large beam diameters (meters-scale) make these the most optically demanding solar simulator systems [9].

9.5Photobiology and Cosmetics

UV solar simulators are used in dermatology and cosmetics to determine sun protection factors (SPF), study skin photosensitivity, and evaluate UV-protective materials. These instruments are optimized for UV-A (320–400 nm) and UV-B (280–320 nm) output and are calibrated per standards such as ISO 24444 [10].

9.6Concentrated Solar Power (CSP)

High-flux solar simulators — capable of delivering thousands of suns of irradiance — are used to test receivers, absorbers, and thermochemical reactors for concentrated solar power and solar fuel applications. These systems typically use arrays of xenon lamps with ellipsoidal reflectors focused onto a small target, achieving irradiances up to 10,000 kW/m² or more [7].

10.1Matching Simulator Class to Application

Not every application requires Class AAA performance. The required class should be matched to the measurement uncertainty budget of the intended application:

Research and calibration laboratories: Class A+AA or AAA (IEC 2020). These facilities publish certified efficiency values that define the state of the art. Spectral fidelity, uniformity, and stability must all be at the highest level, and spectral mismatch correction is routinely applied [1, 8].

Production-line testing: Class ABA or ABB is typically sufficient. The spectral match must be Class A to avoid systematic efficiency errors across production lots, but slightly relaxed uniformity (Class B) is acceptable because production testers measure relative performance (binning cells by efficiency) rather than absolute calibration [2].

Materials weathering and photocatalysis: Spectral match requirements depend on the specific standard invoked. Many weathering standards require only that the UV content match within specified tolerances, and the visible/IR match is less critical. A Class B spectral match may suffice [10].

Educational and demonstration: Class B or C may be adequate when the goal is to demonstrate photovoltaic concepts rather than measure certified efficiencies.

10.2Illumination Area Sizing

The illumination area must fully cover the device under test with margin. For cell-level testing (typical cells are 156 × 156 mm or smaller), a 6 × 6 inch (152 × 152 mm) or 8 × 8 inch simulator provides adequate coverage. For module testing, flash simulators with 1 × 2 meter or larger areas are standard. Overfilling the DUT area ensures that edge effects from the beam profile do not corrupt the measurement.

10.3Lamp vs. LED Decision

The choice between lamp-based and LED-based solar simulators involves several tradeoffs:

Choose a xenon arc lamp when: broadband spectral coverage beyond 1100 nm is needed (e.g., for Ge-based multi-junction bottom cells), the highest spectral coverage (SPC > 95%) is required, or the application demands a single-point source for high collimation.

Choose an LED array when: spectral tunability is required (multi-junction cell testing, variable AM spectra), long-term operating cost must be minimized, UV and ozone hazards must be avoided, or rapid on/off cycling (flash testing) with microsecond control is needed.

Choose a hybrid (xenon + halogen) when: the application requires calibration-laboratory accuracy across the widest possible wavelength range (300–2500 nm), such as for comprehensive spectral mismatch characterization.

10.4Common Pitfalls

Filter aging: AM filters in xenon simulators degrade over time due to UV exposure and thermal cycling. The spectral shift is gradual and may not be noticed without periodic spectral verification. A simulator that was Class A at installation can drift to Class B over months of heavy use.

Lamp aging: Both xenon and metal halide lamps change their spectral output as they age. The UV content generally decreases while the infrared fraction increases. Irradiance also decreases, requiring higher drive current to maintain 1 sun, which further accelerates spectral change. Monitoring the reference cell reading alone is insufficient — it corrects for total irradiance change but not for spectral shifts.

Non-uniform aging: The spatial irradiance profile can change as the lamp ages or as optical elements degrade. A simulator that was Class A uniform at installation should be re-verified periodically.

Reference cell mismatch: Using a reference cell technology that differs significantly from the DUT technology without applying spectral mismatch correction introduces a systematic error in the reported efficiency.

Thermal effects: Insufficient thermal management of the DUT — especially for steady-state measurements — causes the cell temperature to rise above the 25 °C STC, reducing V_oc and overstating the thermal-coefficient-corrected efficiency.

10.5Selection Workflow

A step-by-step approach to selecting a solar simulator:

1. Define the application and identify the governing standard (IEC 60904, ASTM E927, or application-specific standard).

2. Determine the required classification (A+AA, AAA, ABA, etc.) based on the uncertainty budget.

3. Determine the required illumination area from the largest device to be tested.

4. Determine the required reference spectrum (AM1.5G for terrestrial, AM0 for space, custom for weathering).

5. Evaluate source technology (xenon, LED, hybrid) based on spectral coverage, tunability, lifetime, and cost requirements.

6. Select the appropriate reference cell technology, matched to the DUT technology where possible.

7. Establish a periodic calibration and re-verification schedule for spectral match, uniformity, and temporal stability.