Plate readers in bioproduction

A plate reader is a photometer that has given up the cuvette, and most of its characteristic errors follow from that single change.

Absorbance without a fixed path length

Beer–Lambert makes absorbance proportional to concentration and to path length. In a cuvette the path is a fixed centimetre. In a microplate the beam passes vertically through the liquid column, so the path is whatever the fill volume and meniscus make it — typically a few millimetres, and different in a well that has evaporated. Absorbance values are therefore not portable between plate formats or fill volumes unless the instrument measures the path optically and normalises, which good readers do by comparing water absorbance bands.

For cell density there is a second, deeper problem: an OD600 of a bacterial suspension is not absorption at all. Cells scatter light out of the collection cone, and the reader records the loss as if it were absorbance. That proxy is roughly linear only while scattering is single; above an optical density of about 0.5 photons are scattered more than once and some are scattered back into the detector, so the curve bends and the reading understates cell mass. The correct handling is dilution into the linear regime, and any growth curve read undiluted through the stationary phase compresses the top of itself.

Fluorescence, and what limits it

Fluorescence measures emitted photons against a dark background, so it is intrinsically more sensitive than absorbance — the limit is background, not signal. Culture media supply plenty: riboflavin, phenol red, NADH and serum components all fluoresce in the blue-green, which is why blue-excited assays in medium have a floor that no gain setting improves. At high fluorophore concentration the inner filter effect reverses the calibration: excitation light is absorbed before it reaches the middle of the well and emitted light is reabsorbed on the way out, so signal saturates and then falls, in the same non-monotonic trap that catches sandwich immunoassays.

Two designs attack the background rather than the signal. Time-resolved fluorescence uses lanthanide chelates whose excited state lives hundreds of microseconds instead of nanoseconds; the reader pulses, waits, and reads after all short-lived autofluorescence has decayed. Ratiometric FRET formats divide acceptor by donor emission in the same well, which cancels pipetting variation, quenching by coloured samples and lamp drift, because both channels are affected equally.

Luminescence removes the excitation source entirely — no lamp, no scattered excitation light, no autofluorescence — and is usually the most sensitive mode available. The cost is that the photons are being generated by consuming a substrate, so the signal is a reaction in progress: flash chemistries decay in seconds and must be injected in the reader, glow chemistries trade peak intensity for a stable plateau.

Optics and the statistic that matters

Filter-based readers pass a wide band with high throughput; monochromators buy wavelength freedom and lose light, which matters exactly where sensitivity is scarce. Neither choice rescues an assay with poor separation. The number that decides usability is the Z′ factor, which combines the gap between positive and negative controls with the spread of both; below about 0.5 a screening assay cannot reliably call hits, however elegant its chemistry.