A filtered 300 W xenon source gives 80–150 mW/cm2 at 365 nm over 20 cm2 for roughly one third to one half the price of a Class AAA solar simulator. A simulator earns that premium only when data must be quoted at one sun — 100 mW/cm2, AM1.5G spectrum — as in solar-to-fuel efficiency or outdoor comparison.
Both boxes contain a short-arc xenon bulb, which is why they get compared, but everything that separates them sits downstream of it. A general-purpose xenon source puts a lot of photons on a small window; a solar simulator makes the delivered spectrum match AM1.5G band by band and the total land at 1000 W/m2. The simulator's filter set, its reference cell and its class certificate are what you pay for, and none of the three improves a 365 nm quantum yield measurement.
The decision follows from that. For an apparent quantum yield at a named wavelength you need a narrow band, a radiometer reading in mW/cm2 at the window plane and a defensible photon count — a simulator supplies none of it without a filter that discards 90 % or more of its own output. For solar-to-fuel efficiency, or comparison with published outdoor data, a filtered xenon source cannot produce the reference spectrum at all.
| Cost block | Filtered 300 W xenon source | Class AAA solar simulator | Effect on the quote |
|---|---|---|---|
| Lamp and power supply | Short-arc bulb, igniter, current-regulated supply | Same class of bulb, usually 300–450 W | Small difference; the bulb is not the expensive part |
| Spectral shaping | One bandpass filter, 10 nm FWHM, 365 or 420 nm | Multi-element AM1.5G filter set plus UV and IR cut filters | Largest single difference; matched filter stacks are costly to coat |
| Optics and uniformity | Ellipsoidal reflector, collimator, 20–30 % centre hot spot is normal | Integrating optics, ≤2 % spatial non-uniformity over the rated area | Simulator optics cost more and cover a defined area |
| Certification | Radiometer calibration certificate only | IEC 60904-9 class report: spectral match, uniformity, temporal instability | Adds engineering time, a reference cell and annual requalification |
| Consumables | Bulb at 500–1000 h, filter cleaning | Bulb at 500–1000 h, reference cell drift, filter degradation | Comparable per hour; the simulator has one more item to track |
| Cooling | Fan or 200 W chiller, depending on duty cycle | Same or larger, because the filter stack absorbs IR | Similar; both add roughly 0.2 kWh per irradiated hour |
A certified AAA simulator quotes at two and a half to four times a filtered xenon source covering the same illuminated area. The range is wide because certified area drives it: a 50 × 50 mm unit and a 150 × 150 mm unit differ by more than any lamp choice.
Normalise before you compare. With 1000 irradiated hours a year over a five-year life, capital cost per hour is CAPEX divided by 5000. Index the xenon source at 1.0 and the simulator at 3.0: 0.20 and 0.60 index units per 1000 hours. Add one bulb per 800 hours — 1.25 bulbs per 1000 h, identical in both cases — plus electricity at 0.5 kWh per hour for lamp and cooling, plus recalibration. Recalibrating a radiometer is a bench job; requalifying a simulator needs a reference cell and a spectroradiometer run, at five to ten times the cost.
Those additions close part of the gap: roughly 0.32 versus 0.78 index units per 1000 hours, a ratio near 2.4 instead of 3.0. Two effects then decide the real answer, and neither appears in a quote. Bulb ageing costs more on the system running closest to its rated current, and a rejected experiment costs whatever the run cost — light, catalyst, analysis and a day of attention. If the wrong source voids 20 % of runs, the cheaper box loses where it counts.
A class A rating means spectral match within 0.75–1.25 of the AM1.5G reference in every band, spatial non-uniformity at or below 2 %, and temporal instability at or below 2 %. It describes the instrument at the moment of test, at a stated distance, with a clean window. The bulb loses 20–30 % of output over 500–1000 hours, so the instrument drifts out of class long before the certificate expires; requalify annually and log lamp hours beside every dataset.
What no certificate covers is your experiment: window transmission, collimator-to-surface distance, the meniscus and a scratched quartz plate all change the irradiance at the plane where the chemistry happens. A 4 mm borosilicate window loses about 8 % per pass, a 1 mm quartz window about 4 %, and neither figure appears on a supplier document. Measure at the plane, with your window fitted.
Yes, but it is wasteful. Filtering one sun to a 10 nm band at 365 nm leaves roughly 5–15 mW/cm2, an order of magnitude below a dedicated xenon source at the same setting, and the run takes proportionally longer. Use the simulator for one-sun metrics, a filtered source for wavelength-resolved quantum yield.
Every 12 months, and again after a bulb change, filter cleaning or optical realignment. Bulb ageing alone can move spectral match out of class A before the certificate's anniversary, so log lamp hours and requalify early near rated life.
No. An unfiltered xenon arc has strong near-IR lines and a UV-rich continuum that departs from AM1.5G by more than the ±25 % class A tolerance in several bands. You need the AM1.5G filter set and a measured class report; otherwise it is just a xenon lamp.
An AM1.5G-filtered xenon source with a calibrated radiometer and your own nine-point uniformity map. You give up the third-party class certificate, which matters only if the data must be quoted as certified one-sun illumination. For internal comparisons a documented 100 ±10 mW/cm2 with a stated spectrum is enough.