What fluorescence microscopy is
A fluorophore absorbs a photon, loses a little energy to molecular vibration within nanoseconds, and then emits a photon of longer wavelength. A fluorescence microscope is built to send light of the right colour to the sample and pass back only the fainter, redder emission. The gap between the excitation and emission peaks is the Stokes shift: about 20 nm for fluorescein, and well over 100 nm for some dyes and porphyrins. The larger the shift, the easier it is for filters to separate the two.
That separation matters because the excitation light can be several hundred thousand to a million times brighter than the fluorescence. Almost every design decision in a fluorescence microscope, from the filter cube to the camera, serves one goal: removing excitation light while keeping as many emitted photons as possible.
The light path: epi-illumination and the filter cube
Most modern instruments use epi-illumination. Excitation light comes down through the objective, which acts as its own condenser, and the emission is collected back through the same objective. Illumination and imaging therefore stay aligned whenever you change objectives. The colour selection happens in a filter cube, and a microscope usually holds four to six of them on a turret or slider.
- Excitation filter selects a band from the lamp or LED that matches the fluorophore's absorption.
- Dichroic (dichromatic) mirror, tilted at 45°, reflects the shorter excitation wavelengths down to the sample and transmits the longer emission wavelengths up to the detector.
- Emission (barrier) filter blocks residual excitation and stray light and passes the emission band to the eyepiece or camera.
| Component | Typical spec for a GFP/FITC cube | What goes wrong if it is mismatched |
|---|---|---|
| Excitation filter | bandpass around 470 nm (e.g. 450–490 nm) | weak signal, or excitation of a second dye you did not intend to see |
| Dichroic mirror | cut-on around 505 nm | excitation leaks into the image, background rises |
| Emission filter | bandpass such as 520–560 nm, or long-pass | long-pass: brighter but more bleed-through; narrow bandpass: cleaner but dimmer |
The example ranges follow a widely used Nikon B-2E design; check your own cube's datasheet, because filter edges are what decide whether two dyes stay separate.
Light sources and detectors
Classic widefield systems used mercury (50–200 W) or xenon (75–150 W) arc lamps. Mercury has strong spectral peaks, for example at 365, 406, 435 and 546 nm, while xenon is flatter across the visible range. Arc lamps lose output over a rated life of only a few hundred hours, which is one reason many labs have switched to LED illuminators with individually switchable colours and instant on/off. Lasers are standard in confocal and TIRF systems. On the detection side, scientific CMOS and EMCCD cameras dominate widefield imaging, while point-scanning confocals use photomultipliers or hybrid detectors.
Fluorescence microscopy vs confocal microscopy
In widefield fluorescence the whole field is illuminated at once, so emission from above and below the focal plane adds haze to the image. A confocal microscope scans a focused spot and places a pinhole in a conjugate image plane to reject that out-of-focus light. The mechanism is explained in detail in our article on how the confocal pinhole creates optical sections.
| Aspect | Widefield fluorescence | Confocal fluorescence |
|---|---|---|
| Optical sectioning | none (haze from thick samples) | yes, set by pinhole size |
| Speed | fast, full frame per exposure | slower for point scanning; spinning disk is faster |
| Light efficiency | high, gentle on live cells | lower; higher local intensity |
| Best for | monolayers, quick screening, live-cell time lapse | tissue, 3D stacks, colocalization in thick samples |
| Cost | lower | higher |
For flat cultured cells, a good widefield system with deconvolution often matches what users expect from a confocal, at a fraction of the light dose.
Common pitfalls in fluorescence imaging
Photobleaching
Photobleaching is the irreversible loss of a fluorophore's ability to emit, mostly through reactions of the excited molecule with oxygen. Under deoxygenated conditions fluorescein yields roughly ten times more photons before bleaching than in oxygenated saline. In multi-label samples, dyes fade at very different rates: in a published example a Hoechst nuclear stain dropped rapidly after two minutes of continuous illumination, while mitochondrial and actin stains held up longer.
- focus and frame on a neighbouring area or in transmitted light, then move to the region you will record;
- use the lowest excitation power and shortest exposure that give usable signal; add neutral density filters if needed;
- close the field diaphragm to the area you are imaging;
- mount fixed samples in an antifade medium and keep the shutter closed between acquisitions.
Autofluorescence
Many samples glow without any label: plant chlorophyll, elastin and collagen, lipofuscin in aged tissue, and aldehyde-fixed material, which often shows green background. Lens cements, immersion oil and plastic dishes can add their own signal. Always image an unstained control with the same settings; if it is bright in one channel, move your weakest target to a red or far-red dye.
Bleed-through and cross-excitation
Bleed-through (crosstalk) happens when one dye's emission tail reaches another dye's detection channel; cross-excitation happens when one excitation band also excites a second dye. Both create false colocalization. The usual defences are:
- choose fluorophores with well-separated spectra and balance their brightness;
- acquire channels sequentially, typically starting with the longest wavelength;
- image single-stained controls in every channel to see how much signal leaks where it should not;
- use spectral unmixing when dyes share an excitation line and sequential imaging cannot separate them.
Practical fluorophore combinations and staining steps that reduce crosstalk from the start are covered in our guide on how to stain cells for confocal microscopy.
Saturation and misleading images
A saturated pixel carries no quantitative information. Use the histogram or a look-up table that marks clipped pixels, keep acquisition settings identical between conditions you intend to compare, and adjust display contrast only after you have measured intensities.
Before blaming the biology, image three controls: unstained, single-stained for each dye, and secondary antibody only. Most odd fluorescence results are explained by one of them.
Applications of fluorescence microscopy
Fluorescence microscopy is used to localise proteins with immunofluorescence or fluorescent protein tags, to follow live-cell dynamics, to measure ion concentrations with indicator dyes, and to study molecular mobility with FRAP, which deliberately bleaches a region and records recovery. It also underpins super-resolution methods such as STED, SIM and single-molecule localisation, and is used outside biology for minerals, polymers and semiconductor inspection, where intrinsic fluorescence reveals defects.
Conclusion
Fluorescence microscopy works because the Stokes shift lets a filter cube separate bright excitation from faint emission. Good images come from matching fluorophores to filters, choosing widefield or confocal according to sample thickness, and controlling the usual pitfalls: photobleaching, autofluorescence, bleed-through and saturation. Run the right controls once, and most artefacts become easy to recognise.