What super resolution microscopy actually solves
A conventional light microscope cannot separate two points that sit closer than roughly half the wavelength of the light used. Ernst Abbe expressed this as d = λ / (2 NA). With green light and a good oil immersion objective, the lateral limit lands at around 200–250 nm, and the axial limit is worse, typically 500–700 nm.
Many structures that cell biologists care about are smaller than that: synaptic vesicles, nuclear pores, the spacing of cytoskeletal filaments, protein clusters in membranes. In a standard widefield or confocal image they merge into one blurred spot. Super resolution microscopy is the family of fluorescence methods that break this barrier, either by controlling which molecules emit light or by encoding extra information in the illumination.
The field was recognised with the 2014 Nobel Prize in Chemistry, awarded to Eric Betzig, Stefan W. Hell and William E. Moerner "for the development of super-resolved fluorescence microscopy". Hell was honoured for STED, Betzig and Moerner for the single-molecule work that underpins PALM and STORM.
Super resolution microscopy techniques at a glance
Three families dominate core facilities today. They share one principle: two neighbouring fluorophores must not be read out at the same time in the same way. How they achieve that differs completely.
| Technique | How it beats the limit | Typical lateral resolution | Typical axial resolution | Speed |
|---|---|---|---|---|
| SIM (structured illumination) | Patterned light shifts fine detail into the detectable range; image is reconstructed computationally | about 100–120 nm | about 300 nm (3D-SIM) | Fast, suitable for live cells |
| STED (stimulated emission depletion) | A doughnut-shaped depletion beam switches off fluorescence around the focal centre | about 30–80 nm in cells, lower in ideal samples | about 75–100 nm with 3D depletion | Scanning; fast for small fields |
| PALM / STORM (single-molecule localization) | Only a sparse subset of molecules emits at once; each is localised precisely | about 20–30 nm | about 50 nm with 3D methods | Slow: seconds to many minutes per image |
| Confocal (for reference) | Pinhole rejects out-of-focus light | about 200–250 nm | about 500–700 nm | Fast |
Treat these numbers as practical ranges, not hard specifications. Real resolution depends on labelling density, fluorophore brightness, sample thickness, drift and the analysis pipeline.
SIM: twice the resolution with gentle light
Structured illumination projects a fine striped pattern onto the sample and records several images with the pattern rotated and shifted. Interference between the pattern and the sample (moiré fringes) carries high-frequency information that would otherwise be lost. Software then reconstructs an image with roughly double the resolution of widefield in all three dimensions.
The strength of linear SIM is that it works with ordinary fluorophores and modest light doses, so it is often the first choice for live-cell imaging. Its weakness is reconstruction artefacts: low signal, out-of-focus haze or a mismatch in refractive index can create stripes and "honeycomb" patterns that look like real structure.
STED: switching fluorescence off around the focus
STED is a point-scanning method. An excitation beam excites fluorophores in the focal spot, and a second, red-shifted beam with a zero-intensity centre forces molecules at the rim back to the ground state through stimulated emission. Only molecules in the tiny central region can fluoresce. Because the depletion is non-linear, raising the STED beam power shrinks the effective spot further, so in principle the resolution is not bounded by diffraction.
The trade-off is light dose. High depletion intensities can bleach dyes and stress living samples. STED also needs dyes that tolerate the depletion wavelength, and its images often have lower contrast than SIM or localization images of the same sample. A major practical advantage: the image is acquired directly, with no reconstruction step.
PALM and STORM: building an image one molecule at a time
Single-molecule localization microscopy (SMLM) keeps most fluorophores dark and lets only a sparse, random subset blink on in each camera frame. The centre of each isolated spot can be fitted far more precisely than its width; precision improves roughly with the square root of the number of photons collected. After thousands of frames, the list of coordinates becomes a super-resolved map.
PALM was first shown with photoactivatable fluorescent proteins, while STORM used switchable organic dye pairs; dSTORM later relied on single dyes blinking in special imaging buffers. These methods give the highest routine resolution of the three families, but they need dense, specific labelling, stable buffers, drift correction and patience.
Super resolution microscopy vs confocal: when is confocal still enough?
A confocal microscope is not obsolete. If your question concerns which compartment a protein localises to, whether two markers overlap at the scale of organelles, or how a tissue is organised in depth, a well-optimised confocal often answers it faster and with fewer artefacts. Our guide to how the confocal pinhole creates optical sections explains where its limits come from.
Super resolution becomes necessary when the biological answer depends on distances below about 200 nm. Typical examples:
- counting or sizing protein clusters, receptors or nanodomains in membranes;
- resolving the periodic architecture of the neuronal cytoskeleton or synaptic proteins;
- mapping the arrangement of components in nuclear pores, centrioles or cilia;
- measuring the true separation of two signals that look colocalised in confocal images;
- following fast dynamics of small structures in living cells (usually SIM or fast STED).
Detector-based improvements such as pixel reassignment (image scanning microscopy) sit in between. They give a modest gain, up to about a factor of 1.5–2, while keeping the familiar confocal workflow, which makes them a sensible step before a full super resolution project.
Choose the technique from the biological question backwards: required resolution first, then live or fixed sample, then number of colours and imaging depth. The "best" microscope on paper is the wrong one if your labelling cannot support it.
Sample requirements for super resolution imaging
Most failed super resolution experiments fail at the bench, not at the microscope. The resolution you can claim is limited by the size of your probe and the density of your labelling. A primary plus secondary antibody complex can place the dye 10–20 nm away from the target, which matters when the optical resolution is 20 nm.
| Requirement | SIM | STED | PALM / STORM |
|---|---|---|---|
| Fluorophores | Standard bright dyes and fluorescent proteins | Photostable dyes matched to the depletion laser | Photoswitchable proteins or blinking dyes |
| Imaging buffer | Standard mounting medium | Mounting medium with matched refractive index | Special blinking buffers (oxygen scavenging, thiols) for dSTORM |
| Sample thickness | Best below about 10–20 µm | Works deeper, but aberrations degrade the doughnut | Best close to the coverslip (TIRF or HILO illumination) |
| Live-cell suitability | Good | Possible with care and low doses | Limited, mainly for slow processes |
| Main artefact risk | Reconstruction patterns | Bleaching, low contrast | Undersampling, drift, multiple blinking |
Use high-precision No. 1.5H coverslips, clean them thoroughly and keep refractive indices matched. Our article on fluorescence microscopy filters and common pitfalls covers bleed-through and background problems that become even more visible at higher resolution.
How to choose a super resolution technique: step by step
- Define the length scale. Write down the distance or structure size you need to resolve. Around 100 nm points to SIM; 30–80 nm to STED; below 30 nm to SMLM.
- Decide live or fixed. Live imaging favours SIM or gentle STED; fixed samples open the full range, including SMLM.
- Count your colours. Multicolour SIM is routine; multicolour STED and SMLM are possible but require careful dye selection and correction for chromatic offsets.
- Check probes and labelling density. Smaller probes such as nanobodies or self-labelling tags reduce linkage error; labelling must be dense enough to sample the structure.
- Plan controls. Image known reference structures (for example, microtubules or DNA origami rulers) to verify the resolution you actually achieve.
- Talk to the facility early. Core facility staff know which systems, dyes and buffers work locally and can save weeks of trial and error.
New directions: MINFLUX and expansion microscopy
Two newer approaches push the limits further. MINFLUX, developed in Stefan Hell's group, localises single molecules with a doughnut-shaped excitation beam, using the point of minimal emission rather than maximal signal. It reaches localization precision in the range of a few nanometres with relatively few photons.
Expansion microscopy takes a different route: the sample is embedded in a swellable hydrogel and physically enlarged, roughly fourfold in the original protocol, so that a conventional confocal can resolve features that were below the diffraction limit. Combining expansion with SIM or STED multiplies the gains, though it raises questions about distortion that must be controlled.
For electron-level detail of the same structures, correlative light and electron microscopy (CLEM) links fluorescence to ultrastructure, an area that connects optical and electron microscopy communities.
Conclusion
Super resolution microscopy is not a single instrument but a toolbox. SIM offers a gentle twofold gain for living cells, STED delivers tens of nanometres directly from the microscope, and PALM/STORM reach around 20 nm at the cost of speed and sample preparation effort. Start from the biological question, check whether a confocal already answers it, and invest in labelling and controls before chasing the smallest number on a specification sheet.