Diagram
Spectral balance in broadband measurements
How intensity roll-off across a broadband source limits a single-bench measurement, and how a dual spectrometer system addresses it.
Broadband sources are not spectrally flat. A tungsten-halogen lamp rises through the visible and falls away in the UV, and a deuterium lamp does the opposite. Combining them provides coverage but not balance: the intensity delivered at 250 nm and the intensity delivered at 1600 nm can differ by more than an order of magnitude.
This matters because the dynamic range of a spectrometer is finite and shared across the band. If the integration time is set so that the bright part of the spectrum does not saturate, the weak part falls into the noise. If it is set for the weak part, the bright part clips. On a single bench covering a wide band, one of those two conditions is always present.
Symptoms of insufficient dynamic range
The failure is quiet, which is what makes it expensive. The instrument reports no error. It produces a spectrum that looks plausible and a calibration that does not transfer. Three symptoms are worth recognizing:
- Baseline structure that changes with integration time. If the baseline changes shape when the exposure changes, the measurement is running out of dynamic range somewhere in the band.
- A model that works on standards and fails on real samples. Standards are usually chosen for convenience, often in the well-exposed part of the band. Real samples are not.
- Poor reproducibility only at the ends of the range. The middle of the band behaves and the edges wander.
The dual-spectrometer approach
The direct solution is to stop asking one bench to do two jobs. A dual spectrometer system splits the band and gives each half its own detector, its own integration time and its own gain. Each half then operates in the part of its dynamic range where it performs well.
The cost should be stated plainly: two benches, two sets of optics, more to align, more to keep in calibration, and a stitching region in the middle of the band that has to be handled deliberately. If the measurement occupies a 300 nm window, a single well-chosen bench is the better engineering answer and the second bench is not worth paying for.
The question to ask is not how wide the range is, but how far apart the intensities are that must be measured accurately. If that ratio is larger than the usable dynamic range of the detector, no amount of integration-time tuning will recover the measurement.
