Microscope with improved resolution

Methods for increasing the optical resolution in the far field via the diffraction-limited resolution are known, which are based on a nonlinear interaction of light with the sample. This includes microscopy using stimulated emission depletion (STED,<patcit id="pcit0001" dnum="US5731588A"><text>US 5,731,588</text></patcit>), by means of ground state depopulation (<nplcit id="ncit0001" npl-type="s"><text>Ground State Depletion - GSD, Hell and Kroug, Appl. Phys. B 60 (1995), 495-497</text></nplcit>) and the optical modification (switching) of fluorescent substances (<patcit id="pcit0002" dnum="US20040212799A1"><text>US 2004/0212799 A1</text></patcit>).

In both cases, a diffraction-limited excitation distribution is modified by means of a diffraction-limited optical light distribution (de-excitation light) through a nonlinear interaction in such a way that light emission can only take place from a subregion. This sub-region (effective point spread function: PSF) can be restricted in all three spatial directions or only in the lateral direction. Useful restrictions of the PSF are:<ol id="ol0001" compact="compact" ol-style=""><li>1. 3D limitation of the excitation volume</li><li>Second Lateral limitation of the excitation volume with extended depth of field</li></ol>

The state of the art, which is not very efficient with regard to the steepness of the gradients, is due to the use of phase masks in the pupil of the de-excitation light with a radial phase jump (literature: / 1 / <nplcit id="ncit0002" npl-type="s"><text>TA Klar, S. Jakobs, M. Dyba, A. Egner, and SW Hell, Proc. Natl. Acad. Sci. USA 97, 8206 (2000</text></nplcit>)) or by phase masks with quadrant-wise phase jump (literature: / 2 / <nplcit id="ncit0003" npl-type="s"><text>E. Engel et al. Appl. Phys. B 77, 11-17 (2003</text></nplcit>)) given.

The use of spiral masks for STED microscopy was described in Ref. / 3 / <nplcit id="ncit0004" npl-type="s"><text>P. Török and PRT Munro. Opt.Expr. 12 (2004), 3605</text></nplcit> proposed. However, the use of a spiral mask as described in the above literature does not produce an excitation distribution that is restricted in all spatial directions, nor one with an expanded depth of field as described below.

The incoherent superposition of two partial beam paths of the de-excitation light is from the document <nplcit id="ncit0005" npl-type="s"><text>TA Klar et al., Physical Review E, Volume 64, 066613, 2001</text></nplcit> known.

It is an object of the invention to implement methods which allow corresponding distributions to be generated comparatively easily and with high efficiency (given by the steepness of the gradients of the distributions).

This object is solved by the features of the independent claims.

Compared to the generation of distributions through the superposition of many point sources (literature: <patcit id="pcit0003" dnum="US5866911A1"><text>US 5,866,911 A1</text></patcit>) the following solutions are much easier. It also enables a much more efficient 3D constraint.

The following realizations take place according to the invention:

<ol id="ol0002" compact="compact" ol-style=""><li>1. Excitation: conventional PSF de-excitation: incoherent superposition of a distribution with a spiral phase in the pupil with one with a radial phase in the pupil</li><li>Second Suggestion: incoherent overlay of conventional PSF with a radial phase in pupil de-excitation: incoherent overlay of a distribution with spiral phase in pupil and one with spiral phase + radial phase</li></ol>

<figref idref="f0001">Figure 3</figref> shows a radial mask (a), a spiral mask (b) and a combination of radial and spiral mask (c), the combination being conceivable as a series of a) and b) or as a finished gray value design. Here phase values ​​are gray-value coded (white = 0, black = 2π):<maths id="math0001" num=""><math display="block"><mi>R</mi><mfenced><mi>r</mi></mfenced><mo>=</mo><mi>exp</mi><mfenced separators=""><mo>−</mo><mi mathvariant="italic">jπ</mi></mfenced><mspace width="1ex" /><mi>for</mi><mspace width="1ex" /><mi>r</mi><mo>&lt;</mo><mi>a</mi><mo>/</mo><msqrt><mn>2</mn></msqrt></math><img file="EP1862839B2_D0001.tif" /></maths><maths id="math0002" num=""><math display="block"><mi>S</mi><mfenced><mi>φ</mi></mfenced><mo>=</mo><mi>exp</mi><mfenced separators=""><mo>−</mo><mi mathvariant="italic">jφ</mi></mfenced></math><img file="EP1862839B2_D0002.tif" /></maths><maths id="math0003" num=""><math display="block"><mi mathvariant="italic">SR</mi><mfenced><mi>r</mi><mi>φ</mi></mfenced><mo>=</mo><mi>R</mi><mfenced><mi>r</mi></mfenced><mo>+</mo><mi>S</mi><mfenced><mi>φ</mi></mfenced></math><img file="EP1862839B2_D0003.tif" /></maths>With <i>r</i>, Radial coordinate and φ: angular coordinate in the pupil (α: pupil radius)

The following arrangements distinguish between excitation and de-excitation beams and refer to STED using short pulses as an example. In this case, the excitation takes place first and then with a time delay the de-excitation by stimulated emission, as is known from the prior art. The remaining excited molecules relax with the emission of fluorescence, which is detected.

These arrangements can also be used for other high-resolution processes such as a) ground-state depletion (lit .: <nplcit id="ncit0006" npl-type="s"><text>SW Hell and M. Kroug, Appl. Phys. B 60 (1995) 495</text></nplcit>) or b) the switching of dyes (<patcit id="pcit0004" dnum="US20040212799A1"><text>US2004 / 0212799 A1</text></patcit>) use. In a), dye is first brought into the triplet state by repeated excitation, for which purpose the de-excitation beam is used. The portion of the dye remaining in the basic state is then excited by the excitation beam and the fluorescence is detected. In b) molecules are switched by the de-excitation beam and thereby put into a non-fluorescent state. The molecules are then excited with the excitation beam, the molecules remaining in the fluorescent state being able to emit fluorescence which is detected. In both cases, the use of pulsed light (as with STED) for the excitation and de-excitation beam is not necessary.

Description of Figures 1 and 2:

The reference symbols mean for both figures:<dl id="dl0001" compact="compact"><dt>ON:</dt><dd>Excitation beam path</dd><dt>FROM:</dt><dd>de-excitation beam</dd><dt>DE</dt><dd>detector</dd><dt>SC:</dt><dd>scanner</dd><dt>O:</dt><dd>lens</dd><dt>ST:</dt><dd>beamsplitter</dd><dt>SV:</dt><dd>beam combiner</dd><dt>PO:</dt><dd>pupil optics</dd><dt>DL:</dt><dd>Delay to extend the light path</dd><dt>SM:</dt><dd>spiral mask</dd><dt>RM:</dt><dd>radial mask</dd><dt>PR:</dt><dd>sample</dd></dl>

In <figref idref="f0004">Fig. 1</figref> there is an excitation AN by means of a conventional PSF, for example a point scanner, via the PO, SC and O in the direction PR. The de-excitation beam path AB is split into two beams by means of a beam splitter ST. One beam path with delay DL (where the length of the extension, possibly over optical fibers, is above the coherence length of the lasers used) contains SM, the other contains RM, the delay ensuring an incoherent superposition of both partial beams behind SV. The masks are imaged in or near a lens pupil via the optics PO.

In <figref idref="f0004">Fig. 2</figref> there is a similar de-excitation beam path AB as in <figref idref="f0004">Fig. 1</figref> before, where RM is replaced by the combination mask RM / SM. The excitation beam path AN is also divided into two beam paths by means of ST2. In the partial beam path with DL there is a radial mask RM, which is imaged in or near the objective pupil. The Delay DL ensures the incoherent superposition of the partial beams behind SV2.

The delay must be greater than the coherence length of the source, which is only sensible to solve with fibers in the case of highly coherent lasers. Another possibility of incoherent superposition is the use of light of slightly different wavelengths (within the excitation spectrum for fluorescence (GSD) / for switching or the emission spectrum (STED)). This is the preferred version especially for cw lasers. For example, Both light at 488 nm and at 477 nm can be used for switching and excitation of the Dronpa protein. Thus, in the de-excitation beam path in arrangement 1, light at 488 nm from the spiral mask and that at 477 nm from the radial mask can be formed and combined using suitable dichroic beam splitters. The same applies to excitation and de-excitation beam paths in arrangement 2. The pupil image must produce a stationary phase distribution in the pupil plane of the lens. Regardless of the type of sample interaction, excitation and de-excitation beams will usually hit the sample one after the other. Either the sample is "prepared" with the excitation beam and then the prepared state is "queried" with the excitation beam (GSD and switching), or the excitation is modified by a delayed de-excitation beam (STED).

This ensures an incoherent superposition of the excitation and detection beam.

Also shown below are sections along the lateral coordinate (horizontal) and the axial coordinate (vertical), the distributions being rotationally symmetrical with respect to the axial (optical) axis

<figref idref="f0001">Fig. 4</figref> shows the resulting distributions in the de-excitation and the implementation according to the first embodiment in <figref idref="f0004">Fig.1</figref>, It can be seen that a three-dimensional limitation of the area in focus (object level PR in <figref idref="f0004">Fig.1</figref>) (in the middle of <figref idref="f0001">4 c)</figref>occurs in which there is no de-excitation radiation.

<figref idref="f0002">Figure 5</figref> shows the resulting distribution during de-excitation according to <figref idref="f0004">Fig.2</figref>It can be seen that in the object plane PR (5c) there occurs an area which is extended in the axial direction and is laterally restricted and in which no de-excitation radiation is present.

<figref idref="f0003">Figure 6</figref> shows the distribution of the excitation in <figref idref="f0004">Fig.2</figref>You can see that in <figref idref="f0003">Fig. 6c</figref>) in the object level PR compared to the normal PSF (<figref idref="f0003">Fig. 6b</figref>) elongated excitation light distribution arises.

simulation results

The one used for switching off <i>PSF_s</i> (in Eq. (1) assumed to be normalized to 1) leads to a reduction in the excitability (or excitation) of the dye accordingly <maths id="math0004" num="(1)"><math display="block"><mi>A</mi><mfenced><mi>x</mi><mi>y</mi><mi>z</mi></mfenced><mo>=</mo><mi>exp</mi><mfenced open="{" close="}" separators=""><mo>−</mo><mi>σ</mi><mo>⋅</mo><mi>D</mi><mo>⋅</mo><msub><mi mathvariant="italic">PSF</mi><mi>s</mi></msub><mfenced><mi>x</mi><mi>y</mi><mi>z</mi></mfenced></mfenced></math><img file="EP1862839B2_D0004.tif" /></maths>where σ is the cross section of switching (or de-excitation) and D is the radiation energy per surface. The result of the suggestion with a<i>PSF_a</i> the resulting total PSF is then <maths id="math0005" num="(2)"><math display="block"><mi mathvariant="italic">PSF</mi><mfenced><mi>x</mi><mi>y</mi><mi>z</mi></mfenced><mo>=</mo><mi>A</mi><mfenced><mi>x</mi><mi>y</mi><mi>z</mi></mfenced><mo>⋅</mo><msub><mi mathvariant="italic">PSF</mi><mi>a</mi></msub><mfenced><mi>x</mi><mi>y</mi><mi>z</mi></mfenced></math><img file="EP1862839B2_D0005.tif" /></maths>

Below are xz sections through the PSF for the case of the switchable protein Dronpa (lateral = horizontal, axial = vertical). Switching off with an effective cross-section of 0.07 cm²/ mW / s assumed (literature: <nplcit id="ncit0007" npl-type="s"><text>S. Habuchi et al. Proc. Natl. Acad. Sci. USA 102, 9511 (2005</text></nplcit>)). For an irradiation energy of 3 W / cm² s (ie 3 mW in 10 µs to 1 µm² focused) then result for realization 2 (with the de-excitation accordingly <figref idref="f0002">Fig. 5c</figref>) and the suggestion <figref idref="f0003">Fig. 6c</figref>)) the distribution <figref idref="f0005">Fig. 7a</figref>) and for realization 1 (with the de-excitation accordingly <figref idref="f0001">Fig. 4c</figref>) and the suggestion <figref idref="f0003">Fig. 6b</figref>)) the distribution <figref idref="f0005">Fig. 7b</figref>) compared to conventional PSF (<figref idref="f0005">Fig. 7c</figref>)).

For comparison, the distribution obtained under the same circumstances according to the prior art, namely with a mask with a radial phase shift (<figref idref="f0005">Fig. 8a</figref>)) or with a spiral mask (<figref idref="f0005">Fig. 8b</figref>)) in the de-excitation beam path. It can clearly be seen that in case 8a) an axial, but not an acceptable, lateral limitation of the PSF is obtained. As a result, the resolution is only significantly improved in the axial direction. In case 8b) the same applies with regard to an exclusively lateral improvement of the resolution. In contrast to<figref idref="f0005">Fig. 7a</figref>) there is no application-relevant extension of the depth of field here.