News & Views | Published:

Light microscopy

Beyond the diffraction limit?

Naturevolume 417pages806807 (2002) | Download Citation


The wave nature of light manifests itself in diffraction, which hampers attempts to determine the location of molecules. Clever use of microscopic techniques might now be circumventing the 'diffraction limit'.

The two best-known physical limitations are Abbe's resolution limit in optical physics and Heisenberg's uncertainty principle in quantum physics. Each defines a natural limit to the resolution or accuracy with which certain parameters can be measured. But, writing in Physical Review Letters, Marcus Dyba and Stefan Hell1 claim to have taken a step beyond one of these limits.

In 1873, Ernst Abbe2 realized that the smallest distance that can be resolved between two lines by optical instruments has a physical and not just a technical limit. The distance — the diffraction limit — is proportional to the wavelength and inversely proportional to the angular distribution of the light observed. No matter how perfectly an optical instrument is manufactured, its resolving power will always have this natural limit.

About 50 years later, Werner Heisenberg3 realized that the parameters describing a quantum particle, such as its location and momentum, are not independent. The accuracy with which one parameter can be determined is coupled to the accuracy with which the other can be determined: one can have an accurate idea either of the location or of the momentum, but not of both at the same time. Improvement in accuracy in one parameter will always occur at the expense of decreasing the accuracy in the other. Heisenberg's 'uncertainty principle' is probably one of the most thoroughly tested relationships in physics, a firm foundation of modern quantum theory, and there is no reasonable account that suggests it is incorrect. Although not a great surprise, it has only recently been shown that Heisenberg's and Abbe's formulae are related4.

Over the past few years, several groups have claimed to have achieved image contrasts that have taken the resolution of light microscopes beyond that of classical instruments, and beyond Abbe's limit. For example, confocal fluorescence microscopy has had an enormous impact in biology. Using lasers to induce emission from fluorescent chromophores (the molecules responsible for the colour of the material), it combines point-by-point illumination with synchronous point-by-point detection. This technique has proved especially useful for imaging biological objects because of its optical sectioning capability: deep inside optically dense objects (such as embryos), it is possible to record fluorescent images that show the chromophore distribution in just a single focal plane. The advantage with this type of instrument comes from its nonlinear behaviour — it is essentially sensitive to the square of the light intensity, not to the intensity itself, and this discriminates against light that is out of focus.

On the basis of the work of Lukosz5, Gustafsson and colleagues6 have achieved excellent resolution with their images of actin protein networks, collected by illuminating the object with light patterns whose intensity varied in a sinusoidal fashion. These and several other contributions to the field have pushed the image resolution down to about 100 nm — which does not contradict Abbe's resolution limit.

Imaging the common soil bacterium Bacillus megaterium with focused light of wavelength 760 nm, Dyba and Hell1 claim to have observed excited molecules with a resolution of 33 nm — that is, down to a distance considerably smaller than Abbe's limit. The authors' technique relies on two essentially independent technologies. The first is '4Pi confocal fluorescence microscopy', which uses two opposing lenses of high numerical aperture to illuminate an object coherently from two sides7. This creates an interference pattern that is modulated along the optical axis and reduces the observed volume by a half. The main maximum in the pattern of diffracted light — one 'volume element' in the image sought — is surrounded by other peaks of light intensity, which can generally be removed by further computations and by using knowledge gained about the object with other means.

The second technique is more complicated. First, the fluorophores are excited with a regular, well-focused beam of light. Picoseconds later, a second beam operating at the emission frequency of the fluorophores but with a light-intensity distribution that has a gap in its centre, stimulates the emission of the fluorophores outside the gap region. A few more picoseconds later, the regular fluorescence emission, stemming from the volume defined by the gap, can be observed. Dyba and Hell used the 4Pi technique to create the gap pattern for the stimulated emission.

The question is what the observed spot width of 33 nm tells us. Have Dyba and Hell achieved a spatial resolution of 33 nm? Although the experiments shown in their paper seem convincing at first, I question whether this is actually a demonstration of resolution improvement. Their images of membranes of Bacillus megaterium show a two-dimensional view of well-spaced membranes that are already well resolved in the straightforward confocal image. The images recorded with the new technique seem to localize the two membranes more exactly. This is at best an improvement in the precision of their locations, but not actually in the resolution of the image, which is defined as the minimum distance between two features that can be resolved with a certain contrast8.

As already seen in the work of Toraldo di Francia and others9,10, the central diffractive peak can be made much narrower, but this always generates huge side 'lobes' of light, which appear outside the area of interest. This effect is equally visible in the images shown by these authors1. Extensive computations are required to discriminate the information in the central lobe from that in the side lobes. So the improvement is not entirely due to the instrument but in fact stems from combining the imaging techniques and from using a priori knowledge. Another problem is that the signal-to-noise ratio becomes worse.

Nevertheless, although the usefulness of the method introduced by Dyba and Hell is not yet clear, the development is quite fascinating. For the moment one thing is still true: Heisenberg was right and Abbe's limit was certainly not broken.


  1. 1

    Dyba, M. & Hell, S. W. Phys. Rev. Lett. 88, 163901 (2002).

  2. 2

    Abbe, E. Arch. Mikrosk. Anat. 9, 413–468 (1873).

  3. 3

    Heisenberg, W. Z. Phys. 43, 172–198 (1927).

  4. 4

    Stelzer, E. H. K. & Grill, S. Opt. Commun. 173, 51–56 (2000).

  5. 5

    Lukosz, W. J. Opt. Soc. Am. 57, 932–941 (1967).

  6. 6

    Gustafsson, M. G. L., Agard, D. A. & Sedat, J. W. J. Microsc. 195, 10–16 (1999).

  7. 7

    Hell, S. & Stelzer, E. H. K. J. Opt. Soc. Am. A 9, 2159–2166 (1992).

  8. 8

    Stelzer, E. H. K. J. Microsc. 189, 15–24 (1997).

  9. 9

    Toraldo di Francia, G. Atti Fond. Giorgio Ronchi 7, 366–372 (1952).

  10. 10

    Martinez-Corral, M., Caballero, M. T., Stelzer, E. H. K. & Swoger, J. Opt. Express 10, 98–103 (2002).

Download references

Author information


  1. Cell Biology and Biophysics Programme, European Molecular Biology Laboratory, Meyerhofstrasse 1, Heidelberg, 69117, Germany

    • Ernst H. K. Stelzer


  1. Search for Ernst H. K. Stelzer in:

Corresponding author

Correspondence to Ernst H. K. Stelzer.

About this article

Publication history

Issue Date


Further reading


By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.

Newsletter Get the most important science stories of the day, free in your inbox. Sign up for Nature Briefing