EP2979276B1 - Phase contrast x-ray imaging device and refracting grid therefor - Google Patents

Description

The invention relates to a phase-contrast X-ray imaging device, that is to say an X-ray device for phase-contrast imaging. The invention further relates to a diffraction grating for such. The x-ray device and the diffraction grating are provided in particular for phase contrast imaging in the medical field.

The interaction of electromagnetic radiation in general, and x-ray radiation in particular, with a medium is usually described by giving a complex refractive index. The real part and the imaginary part of the refractive index are each dependent on the material composition of the medium to which the complex refractive index is assigned. While the imaginary part reflects the absorption of the electromagnetic radiation in the medium, the real part of the refractive index describes the material-dependent phase velocity, and thus the refraction of the electromagnetic radiation.

Currently used X-ray imaging devices usually detect only the material-dependent radiation absorption in an object to be examined, the intensity of the transmitted through the object X-ray radiation is recorded spatially resolved.

At present, the exploitation of the refraction caused by the object and the associated material-dependent phase shift for the purpose of imaging is still less widespread. Appropriate methods and devices are being developed.

For metrological detection of the phase shift is typically a Talbot-Lau interferometer used, as for example in " X-ray phase imaging with a grating interferometer, T. Weitkamp at al., August 8, 2005 / Vol. 13, no. 16 / OPTICS EXPRESS "is described.

In a conventional Talbot-Lau interferometer, an X-ray source, a coherence grating G ₀ , a phase grating (or diffraction grating) G ₁ , an analysis grating (or absorption grating) G _2, and an X-ray detector constructed of a plurality of pixels are arranged along an optical axis. The coherence grating G ₀ serves to ensure a sufficient spatial coherence of the X-ray source. Accordingly, the coherence grating G ₀ can be dispensed with in the case of an approximately punctiform X-ray radiation source. By means of the phase grating G ₁ , which typically has a uniform striped structure, an interference pattern is generated whose intensity distribution is detected by means of the X-ray detector.

The period of this interference pattern is typically significantly smaller than the size of the pixels of the X-ray detector, so that a direct detection of the interference pattern with the X-ray detector is not possible. In order nevertheless to be able to measure the interference pattern, the X-ray detector is therefore usually preceded by the analysis grid (or absorption grid) G ₂ , with the aid of which the interference pattern can be scanned by spatial-periodic blanking of X-radiation. For this purpose, the analysis grid G _{2 is} displaced in a plane perpendicular to the optical axis and the structure of the interference pattern. As an alternative to the analysis grid G ₂ , the coherence grating G ₀ or the phase grating G _{1 can also be} shifted.

For phase-contrast imaging, the object to be examined is positioned between the X-ray source (and the optional coherence grating) on the one hand and the phase grating G ₁ on the other hand. Alternatively to this The object can also be positioned between the phase grating G ₁ and the analysis grating G ₂ . In both cases, the object causes a location-dependent varying phase shift of the X-ray radiation, which measurably changes the interference pattern generated by the phase grating G ₁ . The changed interference pattern is detected in the manner described above by means of the X-ray detector. From the measured intensity distribution of the interference pattern is then calculated back to the location-dependent phase shift.

The image information is either obtained directly from the phase. Alternatively, the image information may also be determined from the density (i.e., the integrated phase) or the angular spread (dark field). Further, sometimes the phase contrast image is compared with the simultaneously obtained absorption contrast image to reduce the image noise.

The desired advantage of the phase-contrast X-ray imaging is that structures in the soft tissue (especially tissue, water and body fat) in the phase contrast usually stand out more strongly than in the absorption contrast.

However, due to the necessary lattices, Talbot-Lau interferometers either cause a strong loss of intensity (and thus a high X-ray dose) or poor visibility of the interference pattern (and thus a poor phase contrast measurement efficiency), depending on the absorption behavior of the analysis grid used.

Out US 2011/0051889 A1 and US 2012/0201349 A1 In each case a diffraction grating for a phase-contrast X-ray imaging device according to the preamble of claim 1 is known. The known diffraction gratings each have a transversal surface to be aligned essentially transversely to a radiation incident direction, which is spanned by an x-axis and a y-axis perpendicular thereto. The known refraction gratings Furthermore, each have a plurality of refractive webs of an optically comparatively thin base material, which are arranged alternately with optically denser interstices.

Out EP 1 117 010 A2 Further, there is known a diffraction grating for light used in a calibration system for measuring a lateral displacement of a reference point in a lithographic projection apparatus. The different orders of diffraction of the light transmitted through the diffraction grating have different angular positions when leaving the diffraction grating, which are determined by the grating formula. A downstream lens system aligns the different beams in parallel and converts their respective angles to different positions in a plane so that the different orders in that plane are separated. An order aperture is arranged in this plane. In at least some aperture apertures of the order, the order aperture does not only comprise simply blocking selected orders, but wedges to impart a fixed deflection to the respective order rays. The order beams are then focused on fixed reference gratings, behind which are each photodetectors. The wedges are arranged so that the corresponding even and odd orders come together on the same order of the fixed reference gratings.

The invention has for its object to improve the phase contrast X-ray imaging.

With regard to a refraction grating for a phase-contrast X-ray imaging device, this object is achieved according to the invention by the features of claim 1. With regard to a phase-contrast X-ray imaging device, this object is achieved according to the invention by the features of the claim 9. Advantageous and partly inventive in themselves embodiments and further developments of the invention are set forth in the dependent claims and the description below.
The invention is based on the idea of modifying the usual construction of a Talbot-Lau interferometer by positioning, instead of the analysis grating G ₂ or in addition thereto, a diffraction grating G _L (also referred to as a lens grating) in the beam path of the X-ray radiation. In this case, the diffraction grating is configured in such a way that it breaks a plurality of adjacent mutually corresponding structures (eg intensity maxima or intensity minima) of the interference pattern to a common focus, while structures (eg intensity minima or intensity maxima) of the interference pattern lying between them refract to a spatially different focus become. The basic concept of this idea is already in the predecessor application WO 2013/160 153 A1 described.
In this case, the diffraction grating according to the invention has a transverse surface which is spanned by an x-axis and a y-axis perpendicular thereto, and which is to be aligned substantially (ie exactly or at least approximately) transversely to a radiation incident direction. The intended direction of radiation incidence defines a z-axis of the refraction grating, which is aligned in the intended installation position of the refraction grating, in particular parallel to an optical axis of the x-ray imaging device. In this case, the transverse surface can in principle be defined at any z-position (ie position along the z-axis) within the volume of space occupied by the refraction grating. For example only, it is assumed below that the transverse surface is formed by the "front" end face of the refraction grating over which the radiation is incident in the diffraction grating.

The axes introduced above span a Cartesian coordinate system. The spatial directions defined by the orientation of the x, y and z axes are referred to below as (positive) x, y or z direction. The respective opposite spatial directions are designated as (negative) as x, y or z direction. Positions on the x, y, and z axes are labeled as x, y, and z positions, respectively.

In accordance with the desired property of breaking adjacent disparate structures of the interference pattern into different focuses (i.e., focal points or focal lines), the transverse surface is divided into y-directional elongated refraction strips which are juxtaposed in parallel in the x-direction. Adjacent refraction strips differ from one another in that they are always aligned with respect to the refractive properties of the grating material arranged in the region of this refraction strip to different focuses. In other words, the material of the diffraction grating, which is arranged along the z-axis above and / or below a refraction strip, is designed such that it breaks radiation of at least one particular design wavelength into a specific focus, while the material of the refraction grating, along the Z-axis is arranged above and / or below an adjacent refraction strip, the radiation breaks into a different focus. The transverse surface and the refraction stripes formed on it are mathematically abstract structures.

The radiation-refractive effect of the refraction grating is generated by a plurality of refraction webs of an optically comparatively thin base material (ie a solid having a relatively low real part of the refractive index for X-rays), these refractive webs being arranged alternately with optically comparatively dense interspaces. Due to the separating gaps, the refraction webs run within the transverse surface forcibly at least approximately parallel. Preferably, the refractive webs are formed here from gold, nickel or silicon. The intermediate spaces are optionally formed by gaps (air-filled or liquid-filled) or by intermediate webs of an optically comparatively dense material (that is to say a solid having a comparatively large real part of the refractive index for X-rays), for example of photoresist. It should be noted that for X-rays, the real part of the refractive index in all materials is less than one, so that for X-rays - as opposed to visible light - solids always represent the optically thinner medium compared to vacuum or air.

The surface areas respectively occupied by the refraction webs within the transversal surface are not congruent with the refraction stripes, which are defined only by the refractive properties of the refraction grating. Rather, the refraction webs according to the invention are designed such that they extend at least in sections transversely within the transverse surface diagonal (ie obliquely, ie at an angle exceeding 0 ° and 90 ° below the angle to the y-axis). The at least partially diagonal course of the refraction webs in the transverse surface is characterized in that for at least one refraction web, the side surfaces delimiting this refractive web in the x direction extend over at least two refraction stripes within the transverse surface. Since the refraction webs must be at least approximately parallel due to the interspaces, this property necessarily extends to all refractive webs (with the exception of some refraction webs at the edges of the transversal surface whose side surface can extend only over a refraction stripe due to the peripheral position). Apart from any edge effects, the side surfaces of all refraction webs preferably extend over a plurality of refraction stripes, in particular over all refraction stripes.

In another formulation, which is completely equivalent in content, however, the "diagonal course" of the refraction webs in the transverse surface is characterized in that at least one refraction web extends over at least four refraction stripes. This property is due to the at least approximately parallel course of the refraction strips - apart from any edge effects - forcibly for all refractive webs.

Due to the above-described characteristics, the "diagonal layout" of the diffraction gratings of the diffraction grating according to the invention differ qualitatively from the "stripe-shaped layout" of the diffraction bridges of FIG WO 2013/160 153 A1 specified refraction grid. There, the refraction webs extend in the longitudinal direction of the refraction strips, ie in the y direction. In this case, the side surfaces of each refraction web always remain within the space associated with a single refraction stripe, whereby the stripe-shaped refraction webs can extend over only two or at most three refraction striations.

As already in WO 2013/160 153 A1 is described, the diffraction grating can be used according to the invention instead of the analysis grating of the conventional Talbot-Lau interferometer. In this case, the diffraction grating is arranged at its focal distance from the X-ray detector, so that the X-ray radiation is focused by the diffraction grating directly onto the individual pixels or pixel columns of the X-ray detector. Alternatively, the refraction grid, as already shown in FIG WO 2013/160 153 A1 described, the phase grating and the analysis grid are interposed. In this case, the X-radiation is focused through the diffraction grating on the slots and ridges of the analysis grid.

Due to the property of the refraction grating, in each case a plurality of adjacent, mutually corresponding structures (eg intensity maxima or intensity minima) of the interference pattern to break to a common focus, the refraction grating acts as a periodic arrangement of converging lenses, by which the interference pattern is coarsened. This allows the use of a correspondingly coarser and better absorbing in the grid bars analysis grid, whereby the visibility (visibility) is improved. This, in turn, makes it possible to reduce the picture noise and the x-ray dose, in particular in the case of high-energy (short-wave) x-ray radiation. If, in an alternative embodiment of the invention, the analysis grid is omitted, the absorption caused thereby is eliminated, whereby a reduction of the picture noise and the x-ray dose is achieved, in particular with low-energy (long-wave) x-radiation.

Since a diffraction grating can usually be manufactured with a smaller height than an analysis grating, the stripe width of the diffraction grating can be made smaller compared to the typical stripe width of an analysis grating. This allows a greater (i.e., a higher multiple of the Talbot distance) distance to be set between the phase grating and the diffraction grating than between the phase grating and the analysis grating of a standard Talbot-Lau interferometer. This results in a higher (angular) sensitivity, whereby the disadvantage of a slightly lower visibility is outweighed. This allows a further improvement of the picture noise and / or a further reduction of the X-ray dose.

As already in WO 2013/160 153 A1 is described, the diffraction grating is preferably produced in a photolithographic manufacturing process, in particular dam so-called LIGA (lithographic electroplating -) - method or by means of reactive ion etching.

A limiting factor for the production of the diffraction gratings is the aspect ratio, which is limited by the production method, and which is determined for a given grid height in the z direction by the minimum distances that can be established between the side walls of the diffraction webs, namely, depending on the concrete manufacturing process - by the minimum thickness of the refractive webs and / or the minimum thickness of the gaps.

As can be seen, the diagonal layout of the refracting webs for a given grating height and given refractive properties of the refracting strips can ensure particularly large minimum distances between the sidewalls of the refracting webs - both within the refracting webs and between adjacent refracting webs. This in turn allows the production of refractive gratings with a particularly large grating height in the z-direction or a particularly small width of the refraction strips. Such diffraction gratings enable the realization of phase-contrast X-ray imaging devices with a particularly short installation length and particularly high sensitivity.

In a preferred embodiment, the refraction webs are each formed in the manner of oblique prisms inclined in the y-direction, the base surface and top surface of which lie respectively in the end surfaces of the refraction grid parallel to the transversal surface. In this embodiment, the diffraction grating in particular, as in itself already in WO 2013/160 153 A1 is produced by a photolithographic process, in particular LIGA, under oblique exposure of the photoresist layer by X-radiation. The base and the opposite top surface of the prism usually have a complex, polygonal shape here in each case. At the side edges of the refraction grating, the refraction webs-unlike a pure prismatic shape-can be cut off to form edge surfaces oriented in the z-direction. By performing the refractive webs as prisms inclined to the transverse surface, a comparatively high refraction is achieved without substantially impairing other optical properties, in particular the absorption and the visibility.

In particular, the refraction webs are arranged in such a way that a material structure which repeats in the y direction with ay period period is produced in each refraction stripe. The refraction webs are thus designed such that they always occupy parallel-displaced, congruent and uniformly spaced surface sections in each refraction strip. In this case, the refraction webs are inclined in the y direction such that the cover surface of each refraction web opposite the base surface is offset from the base surface by an integer number of period lengths, in particular by exactly one period length. The two opposite in the z-direction end faces of the refraction grid thus have an identical layout, ie an identical formed of refraction webs and spaces material structure.

Preferably, the side surfaces of the refractive webs are each alternately composed of active sub-surfaces with a comparatively strong refractive effect in the x-direction and passive sub-surfaces with a small or vanishing refractive effect in the x-direction. In this case, the active partial surfaces can be regarded as refracting surfaces of (partial) prisms which, with their non-refractive back surfaces, are combined to form a multi-prism forming the respective refraction web.

In this case, the active and passive partial surfaces are preferably each formed by flat (non-curved) surface sections. The refractive effect of each partial area is determined by the associated gradient. In this case, the gradient g = Δz / Δx is designated as a gradient, which this subarea has in a section along an xz plane (ie a plane spanned by the x axis and the z axis). The greater the gradient g, the steeper the respective partial area is set against the transverse surface, the stronger the incident x-ray light is refracted in the x-direction.

Each active or passive partial surface preferably extends over an integral number of refraction strips within the transverse surface in the x-direction. The transition between active and passive sub-areas of a side surface thus preferably coincides with the transition between two refraction strips. Active and passive partial surfaces are preferably offset from one another on the two side surfaces of a refractive web. In a refraction strip in which a first side surface of each of each refraction ridge has an active sub-area, the other side surface of the same ridge thus has a passive sub-area and vice versa. The exception to this rule is formed by refraction stripes without refractive effect in the x-direction (neutral refraction stripes), in which both side surfaces of a refraction web each have passive sub-areas.

The passive partial surfaces can be aligned exactly in the transverse surface parallel to the x-axis (g = 0). Preferably, however, the passive sub-areas also have a small gradient (offset slope). This offset slope is in particular dimensioned such that the slope Δy caused thereby of the passive partial areas over a refraction strip - measured in the y direction - corresponds approximately to between 20% and 50%, preferably approximately 25% of the strip width s _L measured in the x direction (0.2 ≤ Δy / s _L ≤ 0.5).

Due to the offset slope of the passive sub-areas, given a grid height and given refractive properties of the refraction strips, the minimum distances between the side walls of the refraction webs-both within the refraction webs and between adjacent refraction webs-can advantageously be further increased.

The offset slope of the passive sub-areas also causes a flattening of the angle formed between the active and passive sub-areas, which favors the technical manufacturability of the refraction grid.

Since the offset slope affects equally the upper and lower edges of a column, it does not change the phase shift coded thereby. From a rectangle ("Gradient 0") becomes by the offset increase a parallelogram. The parallelogram acts on the phase of the passing light as a rectangle. The offset slope is expediently chosen to be the same between the upper edge and the lower edge of the material underneath. Different columns may also have different additional gradients for edges that have an x-component (purely vertical jumps in the y-direction are not affected). Also, a column in the left and in the right part (in the x-direction) may have a different offset slope when there is a kink therebetween.

In an expedient embodiment of the refraction grating, each refraction web within the transverse surface extends in alternating sections diagonally in the positive y-direction and in the negative y-direction. The refraction webs thus have kinks. Preferably, the refraction webs are alternately oppositely bent at regular intervals along the x-axis, so that the respective refraction web runs meandering within the transverse surface in the direction of the x-axis. As a result of the single or multiple kinked layout for the refraction webs, the intermediate webs made of photoresist, which were first worked out by exposure to X-ray radiation and subsequent development, are mechanically stabilized in the production of the refraction grid in the LIGA process. The kinks are preferably provided in each case in the region of neutral or weakly refractive refraction strips. The course of each refraction web within the transverse surface thus changes the direction respectively to refraction stripes with a small or vanishing refractive effect.

Expediently, there are always a uniformly predetermined number of refraction strips on a common focus aligned. The focused on a common focus refraction strips are referred to collectively as a focusing group. The focusing groups preferably each comprise an odd number of refraction stripes, for example 3, 5, 7 or 9 refraction stripes. Such a focusing group comprises a neutral refractive index having a disappearing refraction effect, around which the further refraction strips of the focusing group are symmetrically arranged, the refraction effect of which in the x-direction increases with increasing distance to the neutral refraction strip. The refraction stripes of adjacent focusing groups are interleaved in this case.

• F1 - "Gradientenkompression": Dieses Ausgestaltungsmerkmal wird insbesondere auf die stärksten Teilprismen (also die aktiven Teilflächen der Brechungsstege mit den jeweils größten Gradienten) angewandt, wenn bei gegebener Gitterhöhe h und gegebener Streifenbreite s_L der Brechungstreifen durch eine aktive Teilfläche, die innerhalb der xz-Ebene linear über die gesamte Streifenbreite s_L und Gitterhöhe h verläuft (g = h/s_L), eine hinreichend starke Brechung nicht erzielt werden kann. Die Gradientenkompression wird realisiert, indem die aktive Teilfläche nicht über die volle Streifenbreite s_L geführt ist, sondern lediglich über einen Anteil 1/c (mit c > 1) dieser Streifenbreite s_L, während sich die aktive Teilfläche in z-Richtung vorzugsweise über die gesamte Gitterhöhe h erstreckt. Der Gradient g erhöht sich in diesem Fall auf g = c·h/s_L. An der maximalen Visibilität ändert die Gradientenkompression zumindest dann wenig, wenn die Intensität in dem verbleibenden Rand des Brechungsstreifens relativ gering ist. Wenn beispielsweise die Streifen hoher Intensität des durch das Phasengitter G₁ gebildeten Interferenzmusters in dem Brechungsgitter G_L genau auf einen Streifen fallen, so fließt die Hauptintensität in der Streifenmitte (=Spaltmitte) und der Rand bleibt vergleichsweise dunkel. Durch den durch Gradientenkompression erhöhten Gradient wird ermöglicht, den Abstand zwischen dem Patienten und dem Röntgendetektor vergleichsweise klein zu halten, was auch die Empfindlichkeit verbessert. Der kleinere Abstand führt dazu, dass jeder Brechungsstreifen auf einen schmäleren Bereich abgebildet wird, dies verbessert die Sichtbarkeit für geringe Kompression (z.B. bei c² < 2). Bei größerer Kompression (z.B. c² > 2) überwiegt dagegen die Reduktion der Sichtbarkeit durch die Randverluste in den am stärksten brechenden Teilprismen. In zweckmäßiger Dimensionierung ist c im Bereich 4/3 ≤ c ≤ 3/2 wie z.B. c = 2^1/2 gewählt. Die Gradientenkompression wird vorzugsweise symmetrisch bezüglich des zugehörigen Brechungsstreifens durchgeführt. Die in ihrer Breite (in x-Richtung) reduzierte aktive Teilfläche wird also bezüglich des zugehörigen Brechungsstreifens zentriert.

In preferred embodiments of the diffraction grating, one or more of the design features described above are further provided in order to further increase the minimum distances within the refraction webs and between adjacent refraction webs and / or to optimize the optical properties of the refraction grating:

• F1 - "Gradient Compression": This design feature is applied in particular to the strongest subprisms (ie the active faces of the refractive webs with the largest gradients) if, for a given grid height h and given stripe width s _L, the refraction stripe is defined by an active sub-area within the xz Plane linear over the entire stripe width s _L and grid height h runs (g = h / s _L ), a sufficiently strong refraction can not be achieved. The gradient compression is realized by the active partial area is not guided over the full stripe width s _L , but only over a proportion 1 / c (with c> 1) of this stripe width s _L , while the active partial area in the z-direction preferably over the entire grid height h extends. The gradient g increases in this case to g = c · h / s _L. At the maximum visibility, the gradient compression changes at least then little if the intensity in the remaining edge of the refracting strip is relatively low. For example, if the high-intensity stripes of the interference pattern formed by the phase grating G _{1 in} the diffraction grating G _L exactly fall on a strip, the main intensity flows in the middle of the strip (= gap center) and the edge remains relatively dark. The gradient gradient-enhanced gradient makes it possible to keep the distance between the patient and the X-ray detector comparatively small, which also improves the sensitivity. The smaller spacing results in each refraction stripe being imaged on a narrower area, which improves visibility for low compression (eg, at c ² <2). In the case of greater compression (eg c ² > 2), the reduction in visibility due to the edge losses in the most refractive prisms predominates. In appropriate dimensioning, c is chosen in the range 4/3 ≦ c ≦ 3/2 such as, for example, c = 2 ^1/2 . The gradient compression is preferably performed symmetrically with respect to the associated refractive stripe. The reduced in its width (in the x direction) active partial area is thus centered with respect to the associated refraction strip.

Variant: One from the registration US 2012/0041679 A1 The method described aims at aligning all grids in a computed tomography based on phase contrast imaging so that in the absence of the patient in the beam, the bright stripes of the interference pattern generated by the phase grating G ₁ are precisely aligned with the fringe boundaries of the analysis grating. In the case of a full rotation of the computer tomograph (in the presence of the patient), the patient then moves the strip once to the right and (after a 180 ° gantry rotation) into the opposite direction (to the left, for example). This can be scanned without any grid shift and converted into images.

• F2 - Gradienten-Variation: Bei den in WO 2013/160153 A1 beschriebenen Brechungsgittern sind die Teilprismen bzw. Teilflächen stets derart ausgerichtet, dass alle Teilprismen bzw. Teilflächen eine Design-Wellenlänge λ_D bzw. Design-Photonenenergie möglichst exakt auf einen gemeinsamen Fokus fokussieren, in dem die Mitte des zugehörigen Elements des Analysegitters G₂ (Spalt bzw. Steg) bzw. Detektor-Pixels platziert wird. Dies beschränkt allerdings das Layout stärker als notwendig:

In transferring this idea to the phase-contrast X-ray imaging device according to the invention, the gradient compression is expediently carried out asymmetrically. In this case, it is expediently determined beforehand which gap boundaries of the refraction grid should receive the high intensity. The compressed gradient is shifted from the center of the strip to this gap boundary. The asymmetric compression thus has the gradient-free part of the width s _L (1-1 / c ) contiguous on the initially unlit side of the refraction strip and the gradient up to the edge of the illuminated column boundary.

• F2 - Gradient Variation: For the in WO 2013/160153 A1 The partial prisms or partial surfaces are always aligned in such a way that all partial prisms or partial surfaces focus a design wavelength λ _D or design photon energy as precisely as possible on a common focus, in which the center of the associated element of the analysis grating G ₂ (gap or web) or detector pixel is placed. However, this limits the layout more than necessary:

• F3 - Variation der Grundhöhe: Die Grundhöhe (d.h., der über die volle Streifenbreite im Layout vorhandene rechteckige oder parallelogrammförmige Teil des Materials) kann verändert werden, um das Material-Aspektverhältnis zu verbessern. Das Aspektverhältnis wird hierbei verringert, indem die Grundhöhe durch Hinzufügen von Material im Layout über die volle Streifenbreite erhöht wird. Alternativ hierzu kann das Freiraum-Aspektverhältnis verbessert werden, indem die Grundhöhe reduziert wird - dies kann bei gegebenem und übererfülltem Aspektverhältnis auch zur Reduzierung des Material-Anteils und damit der Absorption genutzt werden. Diese Änderung betrifft vorwiegend schwache Teilprismen (mit kleinem Gradienten) und mittelstarke Teilprismen.

Due to dispersion, the strongest subprisms distribute the light intensity over a wide range (often over more than one pixel width or S ₂ gap width). Their gradients are therefore not variable without further loss of visibility in the corresponding refraction stripes. However, the weaker the partial prisms become (the closer the subprisms lie along the x-axis to the corresponding focus), the closer the individual wavelengths are to each other, the narrower the overall illuminated area becomes. This provides the freedom to easily increase or decrease the gradient of these prisms without significantly degrading visibility. The weakest partial prisms are located regularly (along the x-axis) between the strongest prisms and the second strongest prisms the neighboring focusing group. Here, by slightly amplifying these weakest partial prisms, the minimum distances within the refraction webs and between the refracting webs can be increased by compressing the associated partial surfaces in the z direction. Often it is also possible to reduce gradients of equally strong neighboring partial prisms, which reduces absorption by the diffraction grating.

• F3 - Variation of Base Height: The base height (ie, the rectangular or parallelogram portion of the material over the full stripe width in the layout) can be changed to improve the material aspect ratio. The aspect ratio is thereby reduced by increasing the base height by adding material in the layout across the full stripe width. Alternatively, the free space aspect ratio can be improved by reducing the base height - this can also be used to reduce the proportion of material and thus absorption at a given and over-fulfilled aspect ratio. This change mainly affects weak partial prisms (with a small gradient) and medium prisms.

• F4 - Zentrale Spalte: Bei den in WO 2013/160153 beschriebenen Brechungsgittern ist der zentrale (und optisch neutrale) Brechungsstreifen einer jeden Fokussierungsgruppe stets komplett mit Material gefüllt. Dies bedingt eine vergleichsweise starke Absorption in den zentralen Brechungsstreifen, wo eigentlich aus optischen Gründen gar nichts erforderlich wäre (da in den zentralen Brechungsstreifen keinerlei Gradient realisiert werden muss).

In the case of weak partial prisms, it is advisable to use a gradient increase (see F2) instead of increasing the basic height for material-saving optimization of the aspect ratio as long as visibility is not significantly reduced.

• F4 - Central column: For the in WO 2013/160153 The refractive gratings of each focusing group are always completely filled with material. This requires a comparatively strong absorption in the central refraction stripes, where actually nothing would be required for optical reasons (since in the central Refraction strip no gradient must be realized).

• F5 "Gradienten-Verschmälerung": Diese Maßnahme ist sehr ähnlich zu der unter Punkt F1 beschriebenen Gradientenkompression, wird aber nicht auf die stärksten Teilprismen, sondern vorwiegend auf die schwächsten Teilprismen angewandt. Wird deren Gradient im streifenförmigen Layout am äußeren und/oder inneren Material-Rand in Richtung der Streifenmitte des Brechungsstreifens mit Streifenbreite s _L zurechtgeschnitten, so können damit die Mindestabstände erhöht werden.

Here it is conceivable to completely mirror the structure at the central column (ie, structure axisymmetric to the y-axis in the middle of the central column) and to incorporate rectangular material gaps across the full width of the central refraction strips (or slightly more or less) , These gaps are neutral in refraction but reduce absorption. They are to be dimensioned so that minimal material widths are not undershot. Preferably, they correspond to a phase shift of the transmitted light along the z-axis by an integer multiple at full wavelengths (analogous to F6 and F7, respectively, for a wavelength of the spectrum near the design wavelength).

• F5 "Gradient narrowing": This measure is very similar to the gradient compression described under point F1, but is not applied to the strongest subprisms, but mainly to the weakest subprisms. If the gradient in the strip-like layout on the outer and / or inner material edge in the direction of the stripe center of the refraction strip with strip width s _L cropped, so that the minimum distances can be increased.

• F6 - Dispersionskorrektur: Durch die unter Punkt F1 erläuterte Gradientenkompression wird eine Erhöhung des geometrischen Gradienten der stärksten Teilprismen erreicht. Allerdings nimmt die Licht-Ablenkung der Brechungsstreifen nicht in demselben Maße zu. Dieser Effekt lässt sich zur Dispersionskorrektur bei den stärksten Teilprismen einsetzen. Da der Brechungsindex δ ∝ λ² ∝ 1/E² und der Ablenkwinkel γ ≈ δ × g ist, folgt, dass γ ∝ λ² ∝ 1/E² ist. Gegeben sei ein starker Gradient, der die Materialhöhe (und damit die Phase aufgrund der höheren Phasengeschwindigkeit durch das Material) in Richtung positiver x-Achse steigert (und das Licht damit in negativer x-Richtung ablenkt). Der x-Abstand Δx sei nun passend zum Gradient g so gewählt, dass ΔΦ = 2π (δ/λ)g Δx = 2π gilt, dass sich also entlang von Δx die Phase durch den Gradient genau um ΔΦ = 2π ändert bzw. die Wellenfront um Δz = +λ_D in Ausbreitungsrichtung verschoben wird (für eine vorzugsweise nahe der Mitte des Auslegungs-Spektrums liegende einer Design-Wellenlänge +λ_D). Wenn hier in der positiven x-Richtung der steigender Materialhöhe nun die Materialhöhe sprunghaft um Δh = g·Δx reduziert wird, so dass ein Phasensprung von ΔΦ = -2π bzw. Δz = -λ_D entsteht, so ist dies für die Design-Wellenlänge λ_D ohne jede Auswirkung (da nur die Phase modulo 2π relevant ist). Für eine um einen Faktor f geänderte Wellenlänge λ = f·λ_D gilt jedoch δ(λ)/λ ∝ f·λ_D, so dass wegen ΔΦ = 2n (δ/λ)Δh lokal an der Sprungstelle eine Phasen-Abweichung von 2π-(1-f) verbleibt. Falls man Abweichungen von ±π/4 toleriert, so muss also 3/4 ≤ f ≤ 5/4 gelten. Das Spektrum hat also eine Bandbreite vom Faktor 5/3=167% (was gut zur bisherigen Bandbreite des Brechungsgitters vom Faktor 3^1/2=173% passt). Wird alle Δx ein dem Gradient entgegengesetzter Phasensprung erzeugt (wodurch die Materialhöhe eine Sägezahnkurve ohne jede mittlere Steigung beschreibt), so kompensieren sich jedoch die Phasen-Abweichung von 2π durch den Gradient mit der durch den Phasensprung, so dass sich die Phase (bis auf die lokale Abweichung von 2π-(1-f)) im Mittel durch die Gerade ΔΦ = 2π·x/Δx annähern lässt, und zwar unabhängig von f bzw. λ. Für die Röntgenstrahlung sieht es also wie eine fehlerbehaftete kontinuierliche Steigung bei δ ∝ λ¹ oc 1/E¹ aus. Es resultieren Ablenkungswinkel linear zu f: γ ∝ λ¹ oc 1/_E ¹ oc f. Der Farbfehler kann so also um eine Größenordnung reduziert werden. Wird in kürzeren Abständen als Δx ein Phasensprung eingefügt, so wird ggf. überkompensiert, was in geringem Umfang sinnvoll sein kann. Es empfiehlt sich der Einsatz von Phasensprüngen erst zur Korrektur von Farbfehlern, wenn (δ/λ)g·s_L größer oder zumindest nicht wesentlich kleiner als 1 ist.

If the outer edge is cut to size (this is the edge in which there is little or no material in the y direction), then the meander-shaped material strip of two adjacent columns becomes narrower overall. The free space minimum distance can increase in this way. If the inner edge is cut to size (this is the edge in which there is a large amount of material in the y-direction), then ultimately the column boundary between the adjacent refraction strips of the meandering material strip is displaced in the direction of the weaker part prism. The minimum material distance can be increased in this way become. In order to influence the maximum visibility a maximum of 10-15% of the strip width should be influenced.

• F6 - Dispersion correction: The gradient compression explained under point F1 achieves an increase in the geometric gradient of the strongest subprisms. However, the light deflection of the refractive stripes does not increase to the same extent. This effect can be used for dispersion correction in the strongest prisms. Since the refractive index δ α λ ² α 1 / E ² and the deflection angle γ ≈ δ × g, it follows that γ α λ ² α 1 / E ² . Given a strong gradient, the material height (and thus the phase due to the higher phase velocity through the material) in the direction of positive x-axis increases (and thus deflects the light in the negative x-direction). The x-distance Δx should now be chosen so that ΔΦ = 2π (δ / λ) g Δx = 2π, so that along Δx the phase through the gradient changes exactly by ΔΦ = 2π or the wavefront is shifted by Δz = + λ _D in the propagation direction (for a design wavelength + λ _D, preferably close to the center of the design spectrum). If, in the positive x-direction of the increasing material height, the material height is suddenly reduced by Δh = g · Δx so that a phase jump of ΔΦ = -2π or Δz = -λ _D occurs, this is for the design wavelength λ _D without any effect (since only the phase modulo 2π is relevant). For a factor f altered wavelength λ = f · λ _D, however, δ (λ) / λ α f · λ _D applies, so that for ΔΦ = 2n (δ / λ) .DELTA.h locally at the jump point, a phase difference of 2π - (1- f ) remains. If one tolerates deviations of ± π / 4, then 3/4 ≤ f ≤ 5/4 must apply. The spectrum therefore has a bandwidth of the factor 5/3 = 167% (which fits well with the previous bandwidth of the refraction grid of factor 3 ^1/2 = 173%). If all the Δx is generated a phase jump opposite the gradient (whereby the material height is a sawtooth curve without describing any average slope), however, the phase deviation of 2π through the gradient compensates with that due to the phase jump, so that the phase (except for the local deviation of 2π- (1- f )) on average by the Just let ΔΦ = 2π · x / Δx approximate, regardless of f or λ. For the X-radiation, it thus looks like a faulty continuous slope at δ α λ ¹ oc 1 / E ¹ . Distortion angles result linearly with f: γ α λ ¹ oc 1 / _E ¹ oc f. The color error can thus be reduced by an order of magnitude. If a phase jump is inserted at shorter intervals than Δx, it may overcompensate, which may make sense to a lesser extent. The use of phase jumps is only recommended for correcting chromatic aberrations if (δ / λ) g · s _{L is} greater than or at least not significantly less than 1.

Numerical examples: For E = 62 keV photons, a full phase jump corresponds to a gold height of h = 25 μm (transmission 85%) or a nickel height of 43 μm (transmission 95%). By contrast, at E = 29 keV, it is a nickel height of 20 μm (transmission 82%) or a silicon height of 75 μm (transmission 98%). At E = 19 keV photons the jump corresponds to a phase of a silicon height of 50 μm (transmission 95%).

• F7 - Zusätzliche Phasensprünge: Im Gegensatz zu der unter Punkt F6 diskutierten Dispersionskorrektur wird hier die Dispersion erhöht (und ist bei kleinen Gradienten dann näherungsweise unabhängig von einer Variation des Gradienten). Zweckmäßigerweise wird die die Dispersion der schwächsten Teilprismen durch einen oder mehrere zusätzliche 2n-Sprünge (Sprünge um jeweils eine Wellenlänge) erhöht. Dadurch nimmt der Materialanteil im Layout zunächst zu, und der Material-Mindestabstand kann dadurch - erwünschtermaßen - auch zunehmen. Diese Maßnahme erlaubt allerdings, die unter Punkt F3 disktierte Grundhöhe zu reduzieren und so den Freiraum-Mindestabstand zu reduzieren. Letztlich nähert sich die Material/Freiraum-Grenze im Streifen der schwächsten Teilprismen damit der Material/Freiraum-Grenze der gegenüberliegenden stärksten Teilprismen in grober Näherung an. Durch die Wahl der x-Positionen der λ-Phasensprünge innerhalb des Streifens kann man beeinflussen, ob der Freiraum-Abstand (in Richtung der gegenüberliegenden stärksten Teilprismen) stärker oder weniger stark als der Materialabstand (innerhalb des mäanderförmigen Materialstreifens mit der benachbarten Prismenspalte) beeinflusst werden soll. Bei größeren Gitterhöhen kann es auch sinnvoll sein, mehrere komplette Phasensprünge relativ gleichmäßig über die Spaltenbreite zu verteilen (z.B. drei Phasensprünge an den Positionen 12,5%, 50%, 87,5% der Teilprismenbreite oder an den x-Positionen 25%, 50%, 75% von s _L). Dies gilt entsprechend auch für die unter Punkt F6 diskutierte Dispersionskorrektur.

Die unter den Punkten F1 bis F6 diskutierten Ausgestaltungsmerkmale der Brechungsstege und der darin enthaltenen aktiven Teilflächen (Teilprismen) können vorteilhaft sowohl bei diagonalen Layouts der Brechungsstege gemäß der vorliegenden Erfindung als auch bei streifenförmigen Layouts gemäß WO 2013/160153 A1 eingesetzt werden, um die Minimal-Abstände innerhalb der Brechungsstege und zwischen benachbarten Brechungsstegen weiter zu vergrößern und/oder um die optischen Eigenschaften des Brechungsgitters zu optimieren.One problem with this strategy is that enough material has to be left at the jump site to meet the minimum widths of the layout. Therefore, the negative phase jump saves on the one hand material, on the other hand, material must be added to have enough material at the jump sites. In the end, that may mean adding material.

• F7 - Additional phase jumps: In contrast to the dispersion correction discussed under point F6 here the dispersion is increased (and is at small gradients then approximately independent of a variation of the gradient). The dispersion of the weakest partial prisms is expediently increased by one or more additional 2n jumps (jumps of one wavelength in each case). As a result, the proportion of material in the layout initially increases, and the material minimum distance can thereby - desirably - also increase. However, this measure makes it possible to reduce the basic height disputed under point F3 and thus to reduce the free space minimum distance. Finally, the material / free-space boundary in the strip of the weakest partial prisms thus approaches the material / free-space boundary of the opposite strongest partial prisms in a rough approximation. By choosing the x-positions of the λ phase jumps within the strip, one can influence whether the free space distance (in the direction of the opposite strongest subprisms) is influenced more or less strongly than the material distance (within the meandering material strip with the adjacent prism gaps) should. For larger grid heights it may also be useful to distribute several complete phase jumps relatively evenly across the column width (eg three phase jumps at the positions 12.5%, 50%, 87.5% of the partial prism width or at the x-positions 25%, 50) %, 75% of s _L ). This also applies accordingly to the dispersion correction discussed under point F6.

The design features of the refractive webs and the active partial surfaces (partial prisms) contained therein, discussed under points F1 to F6, may be advantageous both in diagonal layouts of the refractive webs according to the present invention and in strip-type layouts according to FIG WO 2013/160153 A1 be used to further increase the minimum distances within the refractive webs and between adjacent refractive webs and / or to optimize the optical properties of the refraction grating.

• F8 - Filterung der Spektren: Je schmäler das verwendete Röntgenspektrum ist, desto besser lässt sich die Geometrie auf dieses Spektrum optimieren (z.B. durch Wahl einer Design-Wellenlänge nahe der Mitte des Spektrums). Der spektrale Breite, die das Prismengitter verträgt, muss aufgrund der Abhängigkeit des Ablenkwinkels von der Wellenlänge (α = δ·g) die Bedingung λ_max/λ_min = E_max / E_min ≤ 3^1/2 erfüllen (aus der fremden Pixelmitte muss mindestens eine halbe Pixelbreite s₂/2 und höchstens 3s₂/2 abgelenkt werden. Es muss daher α_max/α_min ≤ 3 gelten. Je schmäler das Spektrum ist, desto besser fokussiert das Brechungsgitter und desto bessere Sichtbarkeit ergibt sich bei größerem Schlitzanteil von z.B. 25%. Um ein möglichst schmales Spektrum zu erreichen, kann z.B. nahe der Röntgenröhre gefiltert werden (z.B mittels einer flächigen Filterfolie oder Filterplatte) mit Materialien, welche das Spektrum anhand ihrer K_α-Kante begrenzen (also höhere Photonenenergien abschneiden). Im nieder-energetischen Bereich (z.B. bei einer Design-Wellenlänge von λD ≈ 25 keV) bietet sich z.B. eine Filterung mit Antimon an, welches Wellenlängen > 30,5 keV stark absorbiert (beispielsweise Filterung mit 100µm Sb), bei Iod wird dagegen > 33,2 keV gedämpft. Im höherenergetischen Bereich bieten sich z.B. Tantal (67,4 keV), Wolfram (69,5 keV) oder Rhenium (71,7 keV) an, wenn die Design-Wellenlänge z.B. 60-65 keV beträgt. Natürlich gibt es bei anderen Design-Wellenlängen andere sinnvolle Materialien.
• F9 - "Zweidimensionale Formulierung des Phasenkontrastes": Optional werden die Phasenänderungen entlang der x-Achse und entlang der y-Achse gleichzeitig gemessen, z.B. unter Verwendung eines Kohärenzgitters G₀, das zeilenweise und spaltenweise filtert. Bei einem Öffnungsanteil von 0 < q < 1 in jeder der beiden Dimensionen passiert also lediglich q² der Intensität die Löcher in G₀ (z.B. mit q = 30% oder q = 50%). In diesem Fall wird ein zweidimensionales Phasengitter G₁ eingesetzt, so dass ein schachbrettartiges Interferenzmuster (z.B. schwarze Schachbrettfelder dunkel, weiße hell) am Ort des Brechungsgitters G_L entsteht. Ein entsprechendes Brechungsgitter ist beispielsweise aus zwei Brechungsgittern der vorstehend beschriebenen Art mit diagonalem Layout gebildet, wobei diese Brechungsgitter um 90° gedreht übereinandergelegt sind, so dass sich überlappende stückweise Linsen in x-Richtung durch eines der beiden Brechungsgitter und überlappende stückweise Linsen in y-Richtung durch das andere Gitter ergeben.

As mentioned above, the (phase-contrast X-ray imaging) device according to the invention comprises an X-ray source, a phase grating and an X-ray detector which has a one-dimensional or two-dimensional arrangement of pixels. The device according to the invention furthermore comprises a diffraction grating of the type described above arranged between the phase grating and the x-ray detector.
If the X-ray source does not already emit sufficiently coherent X-radiation by itself, the device additionally comprises a coherence grating, which is interposed between the X-ray source and the phase grating.
In an expedient embodiment variant, the device additionally comprises an analysis grid, which is interposed between the refraction grid and the X-ray detector. As mentioned above, however, the diffraction grating may also be connected directly upstream of the x-ray detector in an alternative embodiment of the device.
In preferred method variants, the device further embodies one or more of the design features described below

• F8 - Filtering the spectra: The narrower the X-ray spectrum used, the better the geometry can be optimized for this spectrum (eg by choosing a design wavelength near the center of the spectrum). Due to the dependence of the deflection angle on the wavelength (α = δ · g), the spectral width which can be tolerated by the prism grid must satisfy the condition λ _max / λ _min = E _max / E _min ≤ 3 ^1/2 (from the external pixel center) At least half a pixel width s _2/2 and at most 3s _2/2 must be deflected, so α _max / α _min ≤ 3 be valid. The narrower the spectrum, the better the refraction grating focuses and the better visibility results with a larger slot content of eg 25%. In order to achieve the narrowest possible spectrum, it is possible, for example, to filter near the X-ray tube (for example by means of a flat filter foil or filter plate) with materials which limit the spectrum on the basis of their K _α edge (ie cut off higher photon energies). In the low-energy range (eg at a design wavelength of λD ≈25 keV), for example, filtering with antimony, which absorbs wavelengths> 30.5 keV strongly (eg filtering with 100μm Sb), is more than 33 for iodine , 2 keV steamed. In the higher energy range, for example, tantalum (67.4 keV), tungsten (69.5 keV) or rhenium (71.7 keV) are suitable if the design wavelength is, for example, 60-65 keV. Of course, other design wavelengths have other useful materials.
• F9 - "Two-dimensional formulation of the phase contrast": Optionally, the phase changes along the x-axis and along the y-axis are measured simultaneously, eg using a coherence grid G ₀ , which filters line by line and column by column. With an opening fraction of 0 <q <1 in each of the two dimensions, only q ^{2 of} the intensity passes through the holes in G ₀ (eg with q = 30% or q = 50%). In this case, a two-dimensional phase grating G _{1 is} used so that a checkerboard-like interference pattern (eg black checkerboard fields dark, white light) is formed at the location of the refraction grating G _L. A corresponding diffraction grating is formed, for example, from two diffraction gratings of the type described above with a diagonal layout, wherein these diffraction gratings are superimposed rotated by 90 ° so that overlapping piecewise lenses in the x direction through one of the two diffraction gratings and overlapping piecewise lenses in the y direction through the other grid.

The advantage of this design is that there are now multiple paths for each pixel inside the measured image, along which the phase can be integrated (the farther in the center of the image, the more paths are worthwhile or have larger weights). Accordingly, the image noise can be reduced compared to the one-dimensional case by averaging these results along different paths (usually lines).

Also, the design features disclosed in items F8 and F9 can be applied to both diagonal layout diffraction gratings and stripe-shaped diffraction gratings and are therefore considered to be separate inventions.

FIG 1

in einer grob schematischen Schnittdarstellung eine Phasenkontrast-Röntgenbildgebungsvorrichtung mit einem Brechungsgitter,

FIG 2

in schematischer Ansicht auf eine Transversalfläche ausschnitthaft eine Ausführungsform des Brechungsgitters, wobei die Transversalfläche von diagonal verlaufenden Brechungsstegen und dazwischen angeordneten Zwischenräumen durchzogen ist, und wobei die Transversalfläche in eine Anzahl paralleler Brechungsstreifen gegliedert sind, wobei je fünf Brechungsstreifen auf einen gemeinsamen Fokus ausgerichtet sind (N = 5),

FIG 3

in Darstellung gemäß FIG 2 eine alternative Ausführung des Brechungsgitters für N = 9,

FIG 4

in Darstellung gemäß FIG 2 eine Variante des Brechungsgitters gemäß FIG 3, bei der der Verlauf der Brechungsstege innerhalb der Transversalfläche in gleichmäßig beabstandeten Knickstellen die Richtung ändert,

FIG 5

in Darstellung gemäß FIG 2 eine weitere Variante des Brechungsgitters gemäß FIG 3, bei der der Verlauf der Brechungsstege innerhalb der Transversalfläche in gleichmäßig beabstandeten Knickstellen die Richtung ändert,

FIG 6

in neun Einzeldarstellungen gemäß FIG 2 verschiedene Ausführungsformen des Brechungsgitters, in denen eine Offset-Steigung des Brechungsgitters variiert ist,

FIG 7

in sechs Einzeldarstellungen gemäß FIG 2 verschiedene Ausführungsformen des Brechungsgitters, an denen der Einfluss der Gradientenkompression (Punkt F1) demonstriert ist,

FIG 8

in acht Einzeldarstellungen gemäß FIG 2 verschiedene Ausführungsformen des Brechungsgitters, an denen die Optimierung des Aspektverhältnisses und der optischen Eigenschaften des Brechungsgitters mittels Gradientenvariation (Punkt F2), Änderung der Grundhöhe (Punkt F3), Einführung zentraler Spalte (Punkt F4) und Gradientenverschmälerung (Punkt F 5) demonstriert ist,

FIG 9

in einem schematischen Diagramm die Auslegung von Gradienten bei Einsatz von Gradientenkompression (Punkt F1) und Gradientenvariation (Punkf F2),

FIG 10

in vier Einzeldarstellungen gemäß FIG 2 verschiedene Ausführungsformen des Brechungsgitters, an denen die Optimierung des Aspektverhältnisses und der optischen Eigenschaften des Brechungsgitters mittels Dispersionskorrektur (Punkt F6) und Einführung zusätzlicher Phasensprünge (Punkf F7) demonstriert ist,

FIG 11

in schematischer Darstellung möglichkeiten zur Vermeidung von Materialengstellen bei Brechungsstegen mit negativen Phasensprüngen,

FIG 12

in schematischer Darstellung möglichkeiten zur Vermeidung von Materialengstellen bei Brechungsstegen mit positiven Phasensprüngen,

FIG 13

in schematischer Darstellung für eine zweidimensionale Formulierung einer Phasenkontrast-Röntgenbildgebung die Zusammenwirkung zweier um 90° gegeneinander verdrehter Brechungsgitter,

FIG 14

in einem Diagramm die Sichtbarkeit (Visibilität) für verschiedene Ausführungsformen der Vorrichtung,

FIG 15

in zwei Einzeldarstellungen gemäß FIG 2 verschiedene Ausführungsformen des Brechungsgitters, an denen die Realisierung besonders großer Minimal-Abstände innerhalb der Brehungsstege mittels Gradientenvariation (Punkt F2) demonstriert ist,

FIG 16

in drei Einzeldarstellungen gemäß FIG 2 verschiedene Ausführungsformen des Brechungsgitters, an denen die Realisierung von das Aspektverhältnis übersteigenden Gradienten demonstriert ist,

FIG 17

in zwei Einzeldarstellungen gemäß FIG 2 eine für die Anwendung in der Mammographie ausgebildete Ausführungsform des Brechungsgitter mit N = 17, wobei zur Einhaltung von Mindest-Abständen innerhalb der Brechungsstege zusätzliche Phasensprünge vorgesehen sind,

FIG 18

in zwei Einzeldarstellungen gemäß FIG 2 verschiedene Varianten einer weiteren Ausführungsformen des Brechungsgitters für N = 17, an denen die Realisierung besonders großer Minimal-Abstände innerhalb der Brehungsstege mittels Gradientenvariation (Punkt F2) demonstriert ist,

FIG 19

in zwei Einzeldarstellungen gemäß FIG 2 weitere Varianten des Brechungsgitters gemäß FIG 18, bei denen zusätzlich eine Dispersionskorrektur (Punkt F6) vorgesehen ist,

FIG 20

in zwei Einzeldarstellungen zwei Varianten eines Brechungsstegs in einem zentralen Bereich eines diagonalen Layouts, wobei der Brechungssteg jeweils aufgelöst in einzelne Teilprismen dargestellt ist.

Embodiments of the invention will be explained in more detail with reference to a drawing. Show:

FIG. 1

in a rough schematic sectional view of a phase-contrast X-ray imaging device with a diffraction grating,

FIG. 2

1 shows a cross-sectional schematic view of an embodiment of the diffraction grating, the transverse surface being crossed by diagonal refraction webs and interstices therebetween, and the transverse surface being divided into a number of parallel refraction stripes, each having five refraction stripes aligned with a common focus (N = 5),

FIG. 3

in illustration according to FIG. 2 an alternative embodiment of the diffraction grating for N = 9,

FIG. 4

in illustration according to FIG. 2 a variant of the refraction grating according to FIG. 3 in which the course of the Refraction webs within the transverse surface in uniformly spaced kinks the direction changes

FIG. 5

in illustration according to FIG. 2 a further variant of the refraction grating according to FIG. 3 in which the course of the refraction webs within the transverse surface changes the direction in evenly spaced kinks,

FIG. 6

in nine separate presentations according to FIG. 2 various embodiments of the diffraction grating in which an offset slope of the diffraction grating is varied,

FIG. 7

in six separate presentations according to FIG. 2 various embodiments of the refraction grating, where the influence of the gradient compression (point F1) is demonstrated,

FIG. 8

in eight separate presentations according to FIG. 2 various embodiments of the diffraction grating, in which the optimization of the aspect ratio and the optical properties of the diffraction grating by means of gradient variation (point F2), change of the base height (point F3), introduction of central column (point F4) and gradient narrowing (point F 5) is demonstrated,

FIG. 9

in a schematic diagram the design of gradients when using gradient compression (point F1) and gradient variation (Punkf F2),

FIG. 10

in four separate presentations according to FIG. 2 various embodiments of the refraction grating, in which the optimization of the aspect ratio and the optical properties of the diffraction grating by means of dispersion correction (point F6) and introduction of additional Phase jumps (punk F7) is demonstrated,

FIG. 11

in a schematic representation possibilities for avoiding material galleries in refraction webs with negative phase jumps,

FIG. 12

in a schematic representation possibilities for avoiding material galleries in refraction webs with positive phase jumps,

FIG. 13

in a schematic representation for a two-dimensional formulation of a phase-contrast X-ray imaging, the interaction of two 90 ° against each other twisted refraction gratings,

FIG. 14

in a diagram the visibility (visibility) for different embodiments of the device,

FIG. 15

in two separate presentations according to FIG. 2 various embodiments of the refraction grating on which the realization of particularly large minimum distances within the Brehungsstege by means of gradient variation (point F2) is demonstrated

FIG. 16

in three separate presentations according to FIG. 2 various embodiments of the refraction grating on which the realization of the aspect ratio exceeding gradient is demonstrated,

FIG. 17

in two separate presentations according to FIG. 2 an embodiment of the diffraction grating with N = 17, which is designed for use in mammography, wherein additional phase jumps are provided to maintain minimum distances within the refraction webs,

FIG. 18

in two separate presentations according to FIG. 2 various variants of a further embodiment of the refraction grid for N = 17, at which the realization of particularly large minimum distances within the Brehungsstege by means of gradient variation (point F2) is demonstrated

FIG. 19

in two separate presentations according to FIG. 2 Further variants of the refraction grating according to FIG. 18 in which additionally a dispersion correction (point F6) is provided,

FIG. 20

in two individual representations, two variants of a refraction ridge in a central region of a diagonal layout, wherein the refraction web is shown in each case dissolved in individual prisms.

Corresponding parts and sizes are always provided with the same reference numerals in all figures.

In the FIG. 1 schematically illustrated (phase-contrast X-ray imaging) device 2 comprises an X-ray source 4, a coherence grating G ₀ , a phase grating G ₁ , a diffraction grating G _L , a grating G ₂ and an X constructed of a plurality of pixels P-ray detector 6th

The structure can be a system axis (hereinafter referred to as optical axis 8) assign, which is aligned in the case of the embodiment in a z-direction. The individual optical elements of the X-ray device 2 are configured flat in the exemplary embodiment, arranged along this optical axis 8 and aligned in each case perpendicular to this.

The X-ray device 2 is provided for obtaining medical differential phase contrast images. For image acquisition, a patient between the coherence grating G ₀ and the phase grating G ₁ , preferably immediately before the phase grating G ₁ , positioned. The metrological detection or rather the determination of the spatial distribution of the phase shift caused by the patient takes place in the case of the x-ray device 2 presented here, according to known per se and, for example, in US Pat. X-ray phase imaging with a grating interferometer, T. Weitkamp at al., August 8, 2005 / Vol. 13, no. 16 / OPTICS EXPRESS "described principle.

The coherence grating G ₀ has a grating constant (grating period) p ₀ and a grating height h ₀ (measured in the z direction) and serves to ensure sufficient spatial coherence of the x-ray radiation used for the interferometric measuring method. The coherence grating G ₀ is typically positioned at a distance of about 10 cm to the X-ray source 4 and has in typical dimensioning about the dimensions of a postage stamp. In an alternative embodiment of the device 2, instead of a spatially extended X-ray source 4, a X-ray source which is punctiform to a good approximation is used, which already emits sufficiently coherent X-ray radiation. In this case, the coherence grating G _{0 is} omitted.

In operation of the device 2, the X-ray source 4 emits X-radiation with a photon energy up to about 100keV. For the coherence grating G ₀ , which preferably consists of gold (Au, Z = 79), a height h ₀ of 1000 μm and a grating constant p ₀ of 26.83 μm are selected in an appropriate dimensioning.

At a distance d ₀₁ of, for example, 1000 mm in the z-direction offset from the coherence grating G ₀ , the phase grating G _{1 is} arranged. This serves as in a conventional Talbot-Lau interferometer to produce a strip-like interference pattern and has for this purpose a strip-shaped structure with webs and slots formed therebetween, wherein the webs and slots parallel to each other in a (aligned perpendicular to the z-axis) y Extend direction. In the illustration according to FIG. 1 the y-axis is aligned perpendicular to the plane of the drawing.

The phase grating G ₁ is designed in such a way that the incident X-radiation through the webs at a photon energy of eg 65 keV undergoes a phase shift by a quarter of the wavelength, ie by π / 2, while the X-radiation incident in the region of the slits forms the phase grating G ₁ undergoes no significant phase change. For the height h _{1 of} the phase grating G ₁ , a value of 42 μm was chosen. The value of the lattice constant (grating period) p _{1 of} the phase grating G ₁ made of silicon (SI, Z = 14) is 1.42 μm in this exemplary embodiment. Alternatively, the phase grating G _{1 can} also be designed so that it generates a phase shift of the X-radiation by half a wavelength in the region of its webs. In this case, the lattice constant p ₁ is 2.84 μm.

At a distance d _1L of, for example, 55.91 mm offset from the phase grating G ₁ , the diffraction grating G _{L is} positioned. The geometry of the refraction grating G _L is characterized by three axes, which are designated according to the intended orientation of the refraction grating G _L in the device 2 as x-axis, y-axis and z-axis. According to the invention, the diffraction grating G _{L is} arranged within the device 2 such that its z-axis is arranged parallel to the optical axis 8, and thus to the z-direction and the averaged radiation propagation direction within the device 2. The x-axis and the y-axis of the refraction grating G _L , which are aligned perpendicular to both the z-axis and each other, thus span a transversal surface 10 extending perpendicular to the radiation incident direction. In the z direction, the diffraction grating G _{L has} a grating height h _L of approximately 60 μm.

The diffraction grating G _L is used to manipulate the X-ray field emitted by the phase grating G ₁ and for this purpose has a grating constant p _L of, for example, 1.5 μm, which corresponds to the periodicity of the interference pattern generated by the phase grating G ₁ at the location of the refraction grating G _L corresponds. The periodicity of the structure of the diffraction grating G _L is given by N xp _L , where N is a natural, preferably odd number (N = 3, 5, ...). In the embodiment according to FIG. 2 For example, N is 5 (N = 5).
According to the lattice constant p _L is the in FIG. 2 Sectionally illustrated in greater detail transverse surface 10 of the refraction grating G _L in individual elongated refraction strips 12 (FIG. FIG. 2 ), which extend in the y-direction in each case over the entire transverse surface 10 and are lined up in the x-direction parallel side by side. The refraction strips 12 have a uniform (measured in the x direction) stripe width s _L , which corresponds to half the lattice constant p _L (S _L = 0.5 · p _L ). Each two adjacent refraction strips 12 thus have together one of the lattice constant p _L corresponding width. As mentioned above, the end face of the refraction grating G _{L is} identified by way of example with the transversal surface 10 which faces the X-ray source 4 and at which the X-ray radiation thus enters the refraction grating G _L.
As in FIG. 1 indicated, each refraction strip 12 is aligned with an associated focus F. In other words, the material structure of the refraction grating G _L arranged in the z-direction over the respective refraction strip 12 is chosen such that the X-ray radiation incident in this refraction strip 12 is refracted into the associated focus F. Adjacent refraction strips 12 are always aligned with different focuses F.
The number N describes the number of refraction strips 12, which are aligned to a common focus F. These refractive webs 12 aligned with a common focus F are - as already mentioned above - designated as a focusing group. The number N is uniformly selected for all focusing groups and thus represents a global property of the diffraction grating G _L. In other words, each focusing group of the diffraction grating G _L comprises the same number (in the example shown in FIG N = 5) on respective refractive strip 12.

If the number N is odd, each focus group has a neutral refraction stripe 12a in which incident x-ray radiation is not or only to a negligible extent refracted. The further refraction strips 12 of the focusing group are symmetrically arranged around the neutral refraction strip 12a (two further refraction strips 12b and 12c on both sides at N = 5) whose refractive effect increases with increasing distance to the neutral refraction strip 12a. Since the incident interference pattern is strip-shaped and therefore has little or no change in the y-direction, only the radiation refraction in the x-direction is considered here and below. Any parts of the radiation refraction in the y direction are neglected.

Since the lattice constant of the diffraction grating G p _{L L} of the period of the interference pattern generated by the phase grating P ₁ of the diffraction grating G _L at the place corresponding to the interference maxima and interference minima of the interference pattern are refracted in respectively different foci F. The neighboring focusing groups are here, as in FIG. 1 is indicated schematically, evenly nested.

The said diffraction grating G _L associated focuses F are in a (in z-direction by a distance d of, for example _L2 43,92μm spaced from the diffraction grating G _L) focus plane in which the analyzer grating is positioned G. ₂ The distances of the foci F correspond to half the lattice constant p _{2 of} the analysis grid G ₂ , which has a strip-like structure of webs and slots (columns), so that the foci F lie alternately in the webs and slots of the analysis grid G ₂ . Thus, through the diffraction grating G _L - depending on the relative x-position of the coherence grating G ₁ , the refraction grating G _L and the analysis grating G ₂ - Each N interference maxima (or interference minima) focused on a slot of the analysis grid G ₂ , while the intermediate interference minima (or interference maxima) are focused on the adjacent lands of the analysis grid G ₂ . The grid height h _{2 of} the analysis grid G ₂ is, for example, 400 μm, the grid constant p _2, for example, 7.81 μm.

The expansions of the gratings G ₁ , G _L , G ₂ in the x-direction and in the y-direction are according to the exemplary embodiment FIG. 1 essentially the same. Notwithstanding the schematic representation according to FIG. 1 The extension of the refraction grating G _L and the analysis grating G ₂ in the x direction and in the y direction preferably corresponds approximately to the extent of the x-ray detector 6, more precisely the detector surface spanned by the pixels P of the x-ray detector 6. As with the coherence grating G ₀ , the respective webs are preferably also made of gold in the case of the phase grating G ₁ and the analyte grating G ₂ .

Due to the effect of the refraction grating G _L , to separate interference maxima and interference minima of the interference pattern and to focus in groups, the analysis grating G ₂ can be structured coarser than in a conventional Talbot-Lau interferometer, without thereby the visibility (visibility) of the interference pattern changed significantly. Thus, while in a conventional Talbot-Lau interferometer according to the prior art with a comparable structure, but without diffraction grating G _L , a value for the lattice constant p ₂ of about 1.5 to 2 microns would be selected, the value of the lattice constant p ₂ in the device 2 according to FIG. 1 and N = 5 at five times this value.

If the effect of the refraction grating G _L on five immediately adjacent rays of the same phase is considered, then this corresponds to the effect of a converging lens in whose focus F the analysis grating G _{2 is} positioned. The diffraction grating G _L is therefore also referred to as a lens grid.

To achieve the radiation-refractive effect, the diffraction grating G _{L is formed} from a number of approximately parallel refraction webs 14 made of gold, between which intermediate spaces 16 are formed. The spaces 16 may be air-filled gaps. Alternatively, however, the spaces 16 may also be filled by intermediate webs of photoresist. The refraction webs 14 and the optional intermediate webs are on a base plate 17 (FIG. FIG. 1 ) of the refraction grating G _L , which is aligned parallel to the transverse surface 10 and in the example according to FIG. 1 exemplarily forms the rear (facing away from the x-ray source 4) end face of the refraction grating G _L forms.
The diffraction grating G _L is preferably produced by means of the LIGA method. For this purpose, a radiation-absorbing mask (eg made of gold) is positioned over a photoresist layer applied to the base plate 17, for example about 100 μm thick, and exposed to X-radiation (exposure radiation). By subsequent development, filling areas in the form of holes or trenches, which form a negative mold for the refracting webs 14 to be produced, detach from the photoresist layer. These filling areas are filled with metal, in particular gold, in a subsequent electroplating process step. The remaining in the interstices 16 photoresist can be left after the preparation of the refractive webs 14 to form the intermediate webs or be dissolved out to form the gaps.
The structure of the mask used in the LIGA method corresponds to the material structure that is visible on the transverse surface 10 of the finished refraction grid G _L. An example of this material structure (hereinafter referred to as layout) is fragmentary in FIG FIG. 2 shown. The (corresponding to the gold structures of the mask) refraction webs 14 are shown here as dark areas. The spaces 16 corresponding to the gaps in the mask are shown as white areas.

It can be seen from this illustration that the refraction webs 14 (and correspondingly also the interspaces 16) extend diagonally across the transverse surface 10, that is to say at an angle which exceeds 0 ° and passes 90 ° in relation to the axis. All refraction webs 14 have the same shape except for any edge effects (ie cut-off partial volumes at the edges of the refraction grid G _L ). The refraction webs 14 are arranged parallel to one another in the y-direction, so that the material structure in the transverse surface 10 has a periodicity with a period length p _y .

The two side surfaces 18, over which each refraction web 14 is delimited from the adjacent interspace 16, are structured by a sequence of "steep" sub-surfaces 20, with relatively strong slope through the transverse surface 10 pull, and "flat" faces 22, in the transversal surface 10 have only a slight (offset) slope (or eg in the embodiment according to FIG FIG. 4 ) even partially horizontally (ie in the transverse plane 10 in the x-direction).

In each of the side surfaces 18, the steep partial surfaces 20 and flat partial surfaces 22 usually follow one another alternately, each partial surface 20 or 22 extending over the full strip width s _{L of} a refraction strip 14. The steep partial surfaces 20 and flat partial surfaces 22 of the two side surfaces 18 of a refractive web 14 are arranged in strips offset from one another. In refracting strips 12, in which one of the two side surfaces 18 of a refraction web 14 has a steep partial surface 22, the other side surface 18 has a flat partial surface 20, and vice versa.

An exception to this rule are the neutral refraction strips 12a, in the area of which both side surfaces 18 have flat partial surfaces 22. In the environment of neutral refraction strips 12a, the flat partial surfaces 22 therefore extend over in each case over the full width of two refraction strips 12th

The steep partial surfaces 20 and flat partial surfaces 22 are sometimes interposed vertical jump surfaces 24 which extend in the transverse surface 10 in the y-direction.

The diffraction grating G _L is produced in the LIGA method under oblique exposure. The mask is exposed in this case with exposure radiation whose beam path is aligned obliquely in the z-plane. The inclination of the beam path against the z-direction is in an expedient embodiment between 2 ° and 30 ° and in particular between 5 ° and 15 °.

The refractive webs 14 therefore each have the shape of an oblique prism in three-dimensional space. Apart from any edge effects, the refracting webs 14 therefore have parallel, congruent and polygonal surface sections which move in the y-direction relative to one another in the transverse plane 10 and the end face of the refraction grating G _L lying opposite the z-direction of the refraction grating G _L corresponding to the base surface or top surface of a prism are. The edges of the side surfaces 18 are inclined in the z-plane by an angle corresponding to the angle of incidence of the exposure radiation. This inclination is matched to the grid height h _L such that the edges of the side surfaces 18 extend in y-direction over exactly one period length p _y . This results in the transverse surface 10 and the opposite end face of the refraction grating G _L an identical, in the viewing direction along the z-axis exactly overlapping (aligned) material structure.

• aktive Teilprismen mit jeweils dreieckiger Grundfläche, wobei jeweils eine Kante dieser Grundfläche eine der steilen Teilflächen 20 begrenzt, sowie in
• passive Teilprismen mit parallelogrammförmiger (oder rechteckiger) Grundfläche, wobei jeweils eine Kante dieser Grundfläche eine der flachen Teilflächen 22 begrenzt.

In order to visualize the optical properties of the refraction grating G _L , the refraction webs 14 can be thoughtfully subdivided into

• active partial prisms each having a triangular base area, wherein each edge of this base defines one of the steep partial surfaces 20, and in
• passive partial prisms with a parallelogram-shaped (or rectangular) base surface, wherein each edge of this base defines one of the flat partial surfaces 22.

In the context of the device 2, the steep partial surface 20 of each active subprism generates a comparatively strong refraction of the x-ray radiation incident along the optical axis 8 in the x direction. The steep partial surfaces 20 are therefore also referred to as "active" partial surfaces 20.

On the other hand, the flat partial surface 22 of each passive subprism produces no or only negligible refraction of the x-ray radiation incident along the optical axis 8 in the x direction. The flat partial surfaces 22 are therefore also referred to as "passive" partial surfaces 22. The passive part prisms are used for the mechanical connection of the active part prisms and are provided for manufacturing reasons, to comply with the minimum manufacturable material widths.

The structure of the refractive webs 14 in active and passive subprisms is shown in the illustration FIG. 2 in the region of the second refraction strip 12 from the left - illustrated there at the central refraction web 14. It can be seen from the illustration that the active partial prism there has a material height y _g in the y direction, while the passive part prism has a material height y _m . The adjacent intermediate spaces 16 each have a width y _f measured in the y direction.

In the example according to FIG. 2 y _m be the minimum manufacturable height of material, and _g y to the minimum manufacturable gap width in the y direction.

• erstens aus einem festen y-Anteil an Material, nämlich hier der Materialhöhe y _m,
• zweitens einem variablen Anteil je Brechungsstreifen 12, hier der Materialhöhe y_g, der den eigentlichen Gradienten sowie etwaige Phasensprünge (z.B. im Rahmen einer Dispersionskorrektur gemäß Punkt F6 oder im Rahmen zusätzlicher Phasensprünge gemäß Punkt F7) und die Grundhöhe codiert und zwischen Materialanteilen 0% bis 100% variieren kann, sowie
• drittens einem festen y-Anteil an Freiraum mit der Breite y _f.

Es gilt also: $$\mathrm{py}=\mathrm{ym}+\mathrm{yg}+\mathrm{yf}\mathrm{.}$$

Die Höhe (h/p_y)(y_m + y_f) repräsentiert dabei eine zusätzliche Höhe, die nichts zum maximalen Gradient beiträgt und somit quasi "verschnitt" bildet. Der Verschnitt-Anteil sinkt mit steigender y-Periode p_y, also mit steigender Materialhöhe h bzw. zunehmender Abweichung von der senkrechten Belichtungsrichtung, da er im wesentlichen vom Fertigungsprozess abhängt, also weitgehend unabhängig von Materialhöhe h und Belichtungswinkel ist. Bei z.B. g = 70 (stärkster Gradient Δz/Δx) und einem Belichtungswinkel, der 15° von der Senkrechten abweicht, sowie y_m = y_f = s_L ergibt sich beispielsweise $${\mathrm{p}}_{\mathrm{y}}=\mathrm{20},\mathrm{76}{\mathrm{s}}_{\mathrm{L}}=\mathit{h}\tan \left(\mathrm{15}\mathrm{°}\right)=\mathrm{70}\tan \left(\mathrm{15}\mathrm{°}\right)+\mathrm{2}{\mathit{s}}_{\mathit{L}},$$

also
h =77,46 s_L = (70+2/tan(15°)) S_L
und damit einen Anteil des Gradienten an der Gesamthöhe von $$y_g/p_y={\mathit{g\; s}}_L/h=90,4\%\mathrm{.}$$

Bei einem diagonalen Materialstreifen (z.B. von links unten nach rechts oben in der Transversalebene 10, wie in FIG 2 abgebildet) codiert in jedem x-Streifen der Breite S_L von G_L eine Kante (hier z.B. die obere) von links nach rechts kontinuierlich oder sprunghaft zunehmende Materialdicke, die andere Kante (hier z.B. die untere) von links nach recht kontinuierlich oder sprunghaft abnehmende Materialdicke.The vertical period length p _y in the layout thus exists

• first, from a fixed y-share of material, namely here the material height y _m ,
• secondly, a variable component per refraction strip 12, here the material height y _g , which encodes the actual gradient and any phase jumps (eg in the context of a dispersion correction according to item F6 or in the context of additional phase jumps according to item F7) and the basic height and between material components 0% to 100 % can vary, as well
• third, a fixed y-share of free space with the width y _f .

It therefore applies: $$\mathrm{py}=\mathrm{ym}+\mathrm{yg}+\mathrm{yf}\mathrm{,}$$

The height (h / p _y ) (y _m + y _f ) represents an additional height, which contributes nothing to the maximum gradient and thus forms quasi "waste". The proportion of waste decreases with increasing y-period p _y , ie with increasing material height h or increasing deviation from the perpendicular exposure direction, since it essentially depends on the production process, ie is largely independent of material height h and exposure angle. For example, if g = 70 (strongest gradient Δz / Δx) and an exposure angle that deviates 15 ° from the vertical, as well as y _m = y _f = s _L results, for example $${\mathrm{p}}_{\mathrm{y}}=\mathrm{20}.\mathrm{76}{\mathrm{s}}_{\mathrm{L}}=\mathit{H}\tan \left(\mathrm{15}\mathrm{°}\right)=\mathrm{70}\tan \left(\mathrm{15}\mathrm{°}\right)+\mathrm{2}{\mathit{s}}_{\mathit{L}}.$$

so
h = 77.46 s _L = (70 + 2 / tan (15 °)) S _L
and thus a proportion of the gradient in the total height of $$y_G/p_y={\mathit{gs}}_L/H=90.4\%\mathrm{,}$$

In a diagonal strip of material (eg from bottom left to top right in the transverse plane 10, as in FIG. 2 1) encodes an edge (here eg the upper one) continuously from left to right in each x-stripe of width S _L of G _L or suddenly increasing material thickness, the other edge (here eg the lower one) from left to quite continuously or suddenly decreasing material thickness.

Each of the edges runs initially (in the 'pure' formulation) either from left to right or from bottom to top or in a mixture of the two directions, but preferably never right-down or left-up. This is achieved by fitting adjoining columns together (all structures and thus all edges are of course repeated periodically in the y-direction as required for oblique exposure). If at a strip boundary in y-direction the material length in y in the left column is larger than in the right one, then the columns at the top (positive end in y-direction) are aligned with each other ), otherwise at the lower end (the one in the negative y-direction) joined together (there may be a jump at the top). This will make all angles in the layout more rectangular or flatter, reducing the effects of corner fillets due to the process-related radius.

The diffraction grating G _L has the same optically effective gradient g = Δz / Δx in each refraction strip 12, ie the same slope of the respective active partial surfaces 20 of the refraction webs 14 in the xz plane (for the neutral refraction strips 12a g ≈ 0) , In the refractive stripes 12 associated with a focusing group, the gradient is regularly pronounced differently. To illustrate this, the active sub-areas 20 of the five focusing groups associated refraction strips 12a, 12b and 12c in FIG FIG. 2 each highlighted by rectangular frames.

FIG. 3 shows a variant of the refraction grating G _L with similar layout as in FIG. 2 , but for N = 9.

4 and 5 show variants of the refraction grating G _L according to FIG. 3 (also for N = 9 and an offset slope of 50%), in which kinks DK are provided in the diagonal layout, so that the refraction webs 14 in the transverse plane 10 meander around the x-axis and thus alternate sections diagonally into positive y Direction and in negative y-direction. In the embodiment according to FIG. 4 the kinks DK are provided in the neutral refraction stripes 12a. In the embodiment according to FIG. 5 the kinks DK are respectively provided in the refraction stripes 12 with the weakest active partial prisms, that is, the smallest gradient g different from zero. The weakest active partial prisms is according to FIG. 5 additional basic height (point F3) added for a better material aspect ratio in the direction of the refraction strips 12 with the second strongest subprisms.

In the columns of FIG. 6 For example, different diagonal layouts for N = 5 with different offset slope of the passive sub-areas 22 are shown (offset slope from the left column to the right column: 0%, 50%, 100%). Diagonal layouts without kinks DK are shown in the upper line, in the middle line - analogous to FIG. 4 Diagonal layouts with kinks DK in the neutral refraction strips 12a and in the bottom line - analogous to FIG. 5 Diagonal layouts with kinks DK in the refraction strips 12 with the weakest active partial prisms.

In FIG. 7 the effect of the gradient compression explained above under point F1 is shown. In the left column, this effect is shown for strip-like layouts, as they are basically in WO 2013/160 153 A1 are disclosed. The right-hand column illustrates the effect of gradient compression on diagonal layouts. In the upper line the respective layouts without Grdientenkompression are shown. Underneath, the corresponding layout with gradient compression (c = 150%) is shown in the middle column. The arrows inserted here point to the comparison towards the layouts of the top row shifted corners. In the lower line excerpts of the respective overlying layouts are enlarged.

It can be seen that the gradient .DELTA.y / .DELTA.x of all active partial surfaces 20 formed in the transverse plane 10 is increased to 150% as a result of the gradient compression. With the strongest subprisms ("++ P") and - less pronounced - even with the second strongest subprisms ("+ P"), the gradient is restricted to less than the stripe width. Shown is the case that the gradient is realized in the middle of the respective refraction strip 12, that is to say that the gradient in the x direction is uniformly compressed from both sides. A shift of the gradient to one of the edges of the respective refraction strip 12 is alternatively possible.

• F2: Gradientenvariation,
• F3: Variation der Grundhöhe,
• F4: Einführung zentraler Spalte und
• F5: Gradientenverschmälerung

In FIG. 8 are options for improving the aspect ratios for striped layouts according to WO 2013/160 153 A1 (left column) or diagonal layouts (right column) are illustrated by the following measures:

• F2: gradient variation,
• F3: variation of the basic height,
• F4: Introduction of central column and
• F5: Gradient narrowing

In the top line of the FIG. 8 In this case, regular layouts for N = 9 without aspect ratio-optimizing measures are shown. The arrows added here indicate points at which minimal distances in the material of the refraction webs 14 or in the interspaces 16 are improved in the layouts shown below by means of the measures F2 and F3. Critical minimum distances are illustrated by radii.

The second line from the top shows corresponding layouts in which the critical minimum distances are improved by means of gradient variation (point F2) (here registered arrows indicate changes from the top line). In the diagonal layout (right column), gradients are also reduced to reduce material or to approximate gradients.

The third line from the top shows corresponding layouts in which the critical minimum distances are improved by varying the basic height (point F3) (the arrows entered here indicate changes from the top line). This increases the free space distance to the strongest prism and reduces material.

The bottom line shows corresponding layouts in which the critical minimum distances were improved by gradient narrowing (point F5). The arrows inserted here again indicate changes from the top line). Further, in the striped layout in the left column, gaps were inserted in the central columns, that is, the neutral refraction stripes 12a according to the measure described above under item F4.

FIG. 9 illustrates in a diagram the design of gradients for gradient compression (point F1) and gradient variation (point F2). It is shown how in the standard case ("target position = 0%", thick black line), the gradients of the stripes reduce from c (h / s _L ) linearly to 0 to the central column. In the case of the gradient variation (point F2), it can be seen that the gradients can also be reduced to 0 linearly to positions other than the central stripe. This leads to a situation in which all stripes also reflect the wavelength imaged there by the extreme stripe (ie, the strongest prisms). In the case of weaker subprisms, however, the range in which the different wavelengths are imaged falls sharply, so that in practice, for example, target positions between -35% and + 35% (relative to the radius of a "pixel" in G ₂ , ie N s _L / 2 ) to facilitate the production of the layout or to improve. Optimistically, all target positions between -100% and + 100% allow the intensity maxima of all wavelengths to remain in the target range of width s ₂ for all subprisms. However, the width of the diffraction region and the number of openings in G ₀ often result in a narrower choice of target positions. Particularly suitable target positions between -35% and + 35% are recommended.

• F6: Dispersionskorrektur und
• F7: Einführung zusätzlicher Phasensprünge.

In FIG. 10 is the optimization of striped layouts according to WO 2013/160 153 A1 (left column, S _L ) or diagonal layouts (right column D _L ) illustrated by the following measures:

• F6: dispersion correction and
• F7: Introduction of additional phase jumps.

In the top line of the FIG. 10 In this case again regular layouts for N = 7 without aspect-ratio-optimizing measures are shown. The bottom line shows the corresponding layouts optimized by actions F6 and F7.

As can be seen, measure F6 removes material, which reduces the dispersion but would in itself lead to bottlenecks which are difficult to produce. Measure F7 adds material elsewhere, widening these bottlenecks to facilitate manufacturing. The resulting increase in dispersion is tolerable in the area of small gradients - where this measure is used.

In the illustration a single jump was selected which is in the middle of the gap. Of course this can be varied.

The FIGS. 11 and 12 schematically show possibilities for changing the shape of the refracting webs 14 for the realization of negative and positive phase jumps, avoiding hard-to-make bottlenecks.

FIG. 13 schematically shows two superposed and rotated by 90 ° from each other twisted refraction grid G _L for a two-dimensional formulation of the phase contrast (as explained above under point F9). Shown are areas of action 26a-26c of the individual focusing groups, ie those areas to which they focus. Marked with a cross of solid lines are those areas of action 26a, in which of the two refraction gratings G _L respectively in the FIG. 13 Focus on the dark background refraction stripes 12d. With a cross of dashed lines are those areas of action 26b, in which from two refraction gratings G _L respectively in the FIG. 12 Focus white lined refraction bars 12e. Mixed impact areas 26c from one of the diffraction grating G _L the dark shaded refractive strip 12d, and the white-backed refractive strip 12e focus in the from the other diffraction grating G _L are indicated by a cross of a solid and a dotted line.

In the FIG. 13 shown refraction gratings G _L are each configured with a strip-shaped layout, as shown in FIG WO 2013/160 153 A1 is described. Alternatively, the crossed refraction gratings G _{L can} also be diagonal, eg according to one of the FIGS. 2 to 6 designed layout to be executed.

The FIGS. 14 to 19 show concrete simulations.

FIG. 14 shows first simulation results that illustrate the influence of the refraction grating G _L and the optionally made thereon gradient compression (point F1) on the visibility. Simulated was a Vorichtung 2 with d ₀₁ = 500mm, a design energy of X-radiation of 19.5 keV, a tungsten tube with 30 kVp and a filter foil of 70 .mu.m palladium, wherein in the device 2 partially a refraction grid G _L with N = 5 was used.

• "0% p0 + GL": Punktquelle mit Brechungsgitter G_L,
• "0% p0 Point Source": Punktquelle ohne Brechungsgitter G_L, sondern nur mit den Gittern G₀, G₁ und G₂,
• "1x30% p0 slit": G₀ mit 30% offenen Schlitzen ohne Brechungsgitter G_L,
• "1x30% + GL (IrM, grad=60)": G₀ mit einem offenem 30%-Schlitz mit Brechungsgitter G_L (Photolack, Gradient=60, also sehr schwache Brechung),
• "1x30% + GL 117%": wie "1x30% + G_L (IrM, grad=60)" mit Kompression c=117%,
• "1x30% + GL 133%": wie "1x30% + G_L (IrM, grad=60)" mit Kompression c=133%,
• "1x30% + GL 150%": wie "1x30% + G_L (IrM, grad=60)" mit Kompression c=150%,
• "1x30% + GL 167%": wie "1x30% + G_L (IrM, grad=60)" mit Kompression c=167%,
• "1x30% + GL 183%": wie "1x30% + G_L (IrM, grad=60)" mit Kompression c=183%,
• "1x30% + GL 200%": wie "1x30% + G_L (IrM, grad=60)" mit Kompression c=120%.

Shown is the visibility depending on boundary conditions. In detail, the measurement points listed below from left to right were simulated for the following boundary conditions:

• "0% p0 + GL": point source with refraction grid G _L ,
• "0% p0 point source": point source without diffraction grating G _L , but only with the grids G ₀ , G ₁ and G ₂ ,
• "1x30% p0 slit": G ₀ with 30% open slots without diffraction grating G _L ,
• "1x30% + GL (IrM, grad = 60)": G ₀ with an open 30% slot with refraction grid G _L (photoresist, gradient = 60, ie very weak refraction),
• "1x30% + GL 117%": as "1x30% + G _L (IrM, degree = 60)" with compression c = 117%,
• "1x30% + GL 133%": as "1x30% + G _L (IrM, degree = 60)" with compression c = 133%,
• "1x30% + GL 150%": as "1x30% + G _L (IrM, degree = 60)" with compression c = 150%,
• "1x30% + GL 167%": as "1x30% + G _L (IrM, degree = 60)" with compression c = 167%,
• "1x30% + GL 183%": as "1x30% + G _L (IrM, degree = 60)" with compression c = 183%,
• "1x30% + GL 200%": like "1x30% + G _L (IrM, grad = 60)" with compression c = 120%.

It can be seen from the diagram that optimum visibility for a gradient compression of c = 117% is achieved when using the refraction grating G _L. However, a stronger gradient compression of c = 133% only slightly degrades the visibility (about 2%).

FIG. 15 shows the result of a simulation for s _L = 1.0 .mu.m, in which by means of gradient variation (point F2) with a diagonal layout minimum spacings of 1.7 .mu.m were realized both in the interstices and in the material of the refraction webs 14 (In the illustration, the lengths in the y-direction were compressed to 80% of their actual length for the greatest possible representation).

The upper picture shows a layout with a gradient compression of c = 133% and low base height, but without gradient variation. Inserted 1.7μm radii illustrate the required minimum distances. The arrows also inserted indicate violations of these minimum distances. The picture below shows a layout optimized by additional gradient variation (point F2). The minimum distance of 1.7μm is maintained everywhere.

FIG. 16 shows stripe-shaped layouts for N = 7, a stripe width of 1.0μm, minimum widths of 0.88μm and a height of 87μm, where the maximum gradient is greater than the aspect ratio. In the pictures again minimum radii are registered. Inserted arrows indicate their violation.

The upper left image shows a theoretical layout in which the refraction webs 14 form only gradients and no ground level is present. It can be seen that here at many points of the transverse surface 10, the minimum distances are not met. The upper right image shows a layout which is improved by gradient variation (point F2), variation of the basic height (point F3) and gradient narrowing (point F5), but which does not yet have any phase jumps. The picture below shows an optimized layout in which additional phase checks (point F7) ensured compliance with the minimum distances. With this layout, a maximum gradient of 130 with a symmetrical aspect ratio of 100 can be realized.

FIG. 17 shows a stripe-shaped layout of a refraction grating optimized for use in mammography G _L with N = 17, s _L = 1.00 microns, symmetrical minimum distances of 0.76 microns, a symmetrical aspect ratio of 48 and a height of 36 μm. The simulation was carried out for a diffraction grating G _L with refraction webs 14 of nickel and photoresist-filled interstices 16.

The upper picture shows a variant of the layout already optimized by gradient compression (point F1), gradient variation (point F2) and setting of the basic height (point F3). The weakest and second-weakest subprisms are assigned a target position of 35%. Despite the above-mentioned measures - as is again illustrated by registered radii - the minimum distances are not respected everywhere.

The middle picture shows an improved variant of the layout, in which an additional phase jump (point F7) was introduced to maintain the minimum distances in the weakest prisms.

The picture below shows an enlarged section of the middle picture.

The FIG. 18 and 19 show variants of a diagonal layout of a refraction grid G _L with N = 17 optimized for use in mammography. The in the upper picture of the FIG. 18 The layout variant shown here was realized without gradient variation (point F2). The in the lower picture of the FIG. 18 Layout variant shown was realized (to achieve greater minimum distances) with gradient variation (point F2). The picture above FIG. 19 shows a layout variant in which the refractive webs 14 was additionally provided with a negative phase jump (point F6) to reduce color aberrations. The bottom picture off FIG. 19 finally shows the reduction of color errors in 6 of 17 strips per pixel. In each case three successive refractive webs 14 within the focusing group, a phase jump is inserted for dispersion correction.

FIG. 20 Finally, schematically shows in two variants a part of a refraction ridge 14 in a central region of a diagonal layout. Different partial prisms of the refraction web 14 are highlighted for clarity by different filling or hatching. As in the preceding examples, these partial prisms are also made of the same material and are connected together in one piece to form the refraction web 14.

The variant of the refraction web 14 shown in the right-hand half of the figure differs from the variant shown in the right-hand half of the figure by added material for realizing a dispersion correction (point F6) and additional phase jumps (point F7).

It is shown that the dispersion correction can be achieved by adding material on the half of the refraction strip 12 with the lower material fraction (in particular the refraction stripe 14 with the strongest prism part) and that the use of stripes can be useful in which the effects of the measures Cancel F6 and F7. This is shown in the layout variant shown on the right for the weakest subprism, where in the left half additional material for the dispersion correction (point F6), and in the right half as much additional material for the realization of an additional phase jump (point F7) has been added.

Concrete example A (high energy, large minimum distances):

FIG. 15 shows for s _L = 1.0μm that with a diagonal layout despite gradient compression (c = 133%) minimum distances of 1.7μm (> s _L ) can be realized symmetrically (= both for free space and for material). The upper part of the picture shows the situation without gradient variation. Here, minimum distances of only 1.3 μm for the material result (arrows indicate violations of the minimum distance of 1.7 μm).

Refraction grid frame data: N = 7, diagonal layout with offset slope of 50%, exposure angle α = 12 °, S _L = 1.0μm, p _y = 18.47μm = y _m (= 5.2μm) + (= 7, 97μm) + y _f (= 5.3μm), aspect ratio symmetrical (ie, for both space and material) r = 53, minimum spacings symmetric 1.7μm, grid height h = 86.9μm (maximum material height 62μm, average material height 42.8μm), Maximum material gradient 50: 1 = c × 37.5: 1. In the graph, 100% gradient corresponds to the gradient 37.5: 1. The top image (partial image a) uses target positions of 0%. The lower image (partial image b)) uses target positions of 0% (strongest partial prisms), -37% (medium-strength partial prisms are attenuated) and + 35% (weakest partial prisms are amplified).

Overall structure Frame data of the optical simulation: X-ray tube with tungsten anode, 100 kVp, filtering area 20 μm gold + 200 μm rhenium (K (α) = 71.7 keV).

G ₀ : P ₀ = 20,881μm, ridges simulated as completely absorbing ("black"), two adjacent ones of q = 7 slits in G ₀ opened to 30% opening fraction (ie, G ₀ repeats in the x direction all q P ₀ = 146.168μm, the more slits are opened, the brighter the image, but the more the stripes widen in G2 and degrade the visibility). d ₀₁ = 1000mm.

G1: p ₁ = 1.8252 μm, h ₁ = 41.93 μm silicon (alternatively 11.4 μm nickel) for a phase jump of π / 2 ↔ λ / 4 at λ _D = 65 keV; d _1L = 95.78 mm.

G _L : material gold, air as clearance, refraction of light 1: 14100 for 47keV to 1: 32600 for 71keV; d _L2 = 116.071 mm.

G ₂ : p ₂ = 15.4829 μm or detector pixel width about 7.74 μm, height h ₂ = 774 μm gold, aspect ratio 100.

Optical power (simulated with 21 discrete wavelengths 38keV, 41 keV, ..., 98keV with 3keV energy gap respectively), here on the basis of visibility of various boundary conditions and sections (columns). For the visibilities for one strip (the central) or two strips (two strips of the same prism geometry), the other columns were simulated as completely absorbing ("black"). Boundary condition (diagonal layout F1 c = 133% 1. Overall visibility Visibility in the columns with the ... partial prisms headquarters 2nd strongest 3rd middle 4th weakest 5th column a) Without F2 55.8% 39.4% 58.8% 64.5% 64.5% b) With F2 53.7% 52.4% 64%

If F1 is omitted (c = 100%) and d _L2 = 169.7mm in the 2nd (strongest prisms), the result is again 2.39.4%. The strongest prisms always have to deal with dispersion or color errors. Gradient compression is not the reason for the poor performance.

If instead a phase jump is introduced for dispersion correction (point F6) by -24μm height gold, the visibility of the columns of strongest prisms increases from 39.4% to 42.2%. In addition, if the distance is increased from 116.071 mm to d _L2 = 137.85 mm (which is now possible because of the smaller color errors), the visibility increases strongly to 48.6%. The dispersion correction is thus the dominating effect here.

Concrete example B (high energy, high gradient):

FIG. 16 shows that in strip-like layouts, as they are basically in WO 2013/160 153 A1 are disclosed, gradients are possible that exceed the aspect ratio.

N = 7: By means of the measures F1, F2, F3, F5 and F7 (phase jump) a maximum gradient of 130 with a symmetrical aspect ratio of 100 can be realized.

Diffraction grating frame data: N = 7, stripe-shaped layout, exposure angle α = 9 °, s _L = 1.0 μm, p _y = 13.73 μm, aspect ratio symmetrical r = 100, minimum distances symmetrical 0.88 μm, grid height h = 86, 7 μm (maximum material height 86.7 μm), maximum material gradient 130: 1 = c × 86.7: 1 with compression (point F1) c = 150%. In the graph, 100% gradient corresponds to the gradient 86, 67: 1. The upper left image (partial image a) uses target positions of 0%. The upper right image (partial image b))) and the lower image (partial image c)) use target positions of 0% (strongest partial prisms), -20% (medium partial prisms are attenuated) and + 35% (weakest partial prisms are amplified). The lower part of the picture additionally contains an additional phase jump (point F7) of Δh = 25μm gold.

• G₀: Jeweils vier benachbarte von q=17 Schlitzen in G₀ geöffnet zu 30% Öffnungsanteil (d.h., G₀ wiederholt sich in x alle q p₀=17 x 20,881µm = 354,98µm), Stege wiederum perfekt absorbierend ("schwarz") simuliert.
• G_L: Lichtbrechung 1:5400 für 47keV bis 1:12500 für 71keV; d_L2=44, 9907mm.
• G₂: p₂=14,575 µm bzw. Detektor-Pixelbreite ca. 7,3 µm, wie G₀ als perfekt absorbierend angenommen zur Vergleichbarkeit mit der Situation ohne Brechungsgitter (vgl. Zeile 4 in folgender Tabelle).

Overall structure Frame data of the optical simulation (only differences to the FIG. 15 underlying simulation indicated):

• G ₀ : four adjacent each of q = 17 slots in G ₀ open to 30% opening fraction (ie, G ₀ repeats in x all qp ₀ = 17 x 20,881μm = 354,98μm), webs again perfectly absorbent ("black" ) simulates.
• G _L : refraction of light 1: 5400 for 47keV to 1: 12500 for 71keV; d _L2 = 44, 9907mm.
• G ₂ : p ₂ = 14.575 μm or detector pixel width about 7.3 μm, as G ₀ assumed to be perfectly absorbent for comparability with the situation without diffraction gratings (compare line 4 in the following table).

Optical power (spectrum as in example A, FIG. 15 ), here on the basis of visibility of various boundary conditions and sections (columns): Boundary condition [G ₀ , G ₂ perfectly absorbent] total visibility Visibility in the columns with the ... partial prisms Central column most middle vulnerable a) Gradients only 58.4% 39.3% 64.5% 65.6% 65.3% b) With F2, F3, F5, without F7 57.2% 39.3% 64.0% 64.4% 67.0% c) With all / 4x30% G ₀ 56.4% 39.3% 64.0% 62.7% 67.0% Without refraction grid 65.0% --- --- --- 65.0%

To show the loss of visibility through the diffraction grating G _L , the last line in the above table is the situation without diffraction grating G _L (ie, G ₂ instead of G _L ). It was decided not to model the absorption of G ₂ in particular.

The visibility loss is mainly due to the high compression c = 150% and the number of open columns in G ₀ (an equivalent layout with c = 117%, in which only one of then 14 columns in G _{0 is} opened, has a total visibility of 62 , 7%).

It can be seen that the central strip of the layouts according to the partial images b) and c) has higher visibility than the situation without diffraction grating G _L. The reason is that hardening takes place through 87 μm of gold, which removes the low energy fractions and over 80.7 keV (Gold Kα), which are less well represented by G _L due to diffraction between G _L and G ₂ . Therefore, relatively more intensity remains near the design wavelength.

Concrete example C (lower energy, large minimum distances, limited height):

FIG. 17 shows a mammographic configuration with N = 17 and stripe layout, gradient compression (point F1), gradient variation (point F2) and base height (point F3) and additional phase jump (point f7), s _L = 1.0μm, minimum distances symmetric 0.76 μm, aspect ratios symmetrical 48, height 36μm. The refraction webs 14 are made of nickel. The spaces 16 are filled with photoresist.

The larger N is, the stronger are the gradient differences between the smallest gradient strip (weakest sub prism) and the neighboring strip (the one adjacent strip is separated by space, the other connected to material).

Refraction mesh frame data: N = 17, strip-like layout as always in WO 2013/160 153 A1 exposure angle α = 10 °, s _L = 1.00 μm, p _y = 6.348 μm, aspect ratio symmetrical r = 48, minimum distances symmetrical 0.76 μm, grid height h = 36.00 μm (average material height 18.7 μm , with smaller minimum distances this and thus the absorption can be reduced), Maximum material gradient 48: 1 = c × 36: 1 with c = 133%, in the graph 100% gradient corresponds to the gradient 36: 1.

• Oberes Bild (Teilbild a)), Minimal-Abstände 0,63µm (0,76µm eingezeichnet):
  • Zielpositionen 1. 0%, 2. 0%, 3. 0%, 4. 0%, 5. 25%, 6. 35%, 7. 35%, 8. 35%
• Mittleres Bild (Teilbild b)), Minimal-Abstände wie eingezeichnet 0,76µm:
  • Zielpositionen 1. 0%, 2. 0%, 3. -30%, 4. +25%. 5. +25%, 6. +25%, 7. +35%, 8. +25%,
    dabei bei 7. (zweitschwächste Prismen-Spalten): Gradienten-Verschmälerung (Punkt F5) auf 88% der Spaltbreite (6% der Streifenbreite rechts und links abgeschnitten), sowie bei 8. (schwächste Prismen-Streifen) ein Phasensprung (Punkt F7) von Δh=20µm Höhe (Δy=3,53µm), aufgeteilt in zwei Teile je p_y, um die Spalte der Geometrie den benachbarten Spalten anzunähern.

The target positions for the design of the prisms are descending (from the column or the stripe of the strongest prisms ("1.") to the stripe of the weakest prisms ("8."), central stripe 9. always 0%):

• Upper image (partial image a)), minimum distances 0.63μm (0.76μm drawn):
  • Target positions 1. 0%, 2. 0%, 3. 0%, 4. 0%, 5. 25%, 6. 35%, 7. 35%, 8. 35%
• Middle image (partial image b)), minimum distances as shown 0,76μm:
  • Target positions 1. 0%, 2. 0%, 3. -30%, 4. + 25%. 5. + 25%, 6. + 25%, 7. + 35%, 8. + 25%,
    with 7. (second weakest prism columns): gradient narrowing (point F5) to 88% of the slit width (6% of the strip width cut to the right and left), and 8 (weakest prism strips) a phase jump (point F7) of Δh = 20μm height (Δy = 3.53μm), divided into two parts per p _y , to approximate the column of geometry to the adjacent columns.

Overall structure Frame data of optical simulation: X-ray tube with tungsten anode, 40 kVp, filtering 100μm "Sb" antimony (K (α) = 30.5 keV).

G ₀ : p ₀ = 26,6573μm, webs are completely absorbent ("black") (or are in good approximation of gold with height h ₀ ≥ 130μm). Five adjacent each of q = 17 slots opened in G ₀ to 30% opening fraction (ie, G ₀ repeats in the x direction all qp ₀ = 453.174 μm); d ₀₁ = 1500mm.

G1: p ₁ = 3.7208 μm, h ₁ = 8.4 μm nickel for a phase jump of n ↔ λ / 2 at λ _D = = 25keV; d _1L = 112.54 mm.

G _L : As described, material nickel, photoresist as clearance (reduces refractive power of nickel by about 2/13 = 15.5% to about 11/13), refraction of light 1: 5000 for 19 keV to 1: 12500 for 30 keV ; d _L2 = 130,796mm.

G ₂ : p2 = 36.758μm resp. Detector pixel width approx. 18.38μm, as perfectly absorbing simulated for comparability with situation without lens grid.

Optical power (21 discrete photon energies at 18 keV, 19 keV, ..., 38 keV with intervals of 1 keV each), here on the basis of visibility of various boundary conditions and sections (stripes). The transmission is about 74% with lens grid and about 94.5% without lens grid. # Boundary condition [G ₀ , G ₂ perfectly absorbent] total visibility Visibility in the columns with the ... partial prisms Central column 1st strongest 5. middle 8. weakest a) Without F7 / target position 35% 46.1% 33.0% 50.2% 50.1% 51.2% b) With F7, minimum distance 0.76μm 45.1% 47.2% Without refraction grid 51.7% --- --- --- 51.7%

Concrete example D (lower energy, dispersion correction):

The FIG. 18 and 19 show for N = 17 diagonal layouts with the aim to coarsen directly to the pixel size of a fine detector in the lower energy range. The upper picture (partial picture a)) in FIG. 18 is realized without gradient variation (point F2). The lower image (partial image b)) in FIG. 18 is realized to achieve larger minimum distances with gradient variation. The upper picture (partial picture c)) in FIG. 19 is additionally provided with a negative phase shift to reduce color aberrations (point F6). The lower image (partial image d)) off FIG. 19 finally shows the reduction of color errors in 6 of 17 strips per pixel.

• Teilbild a) ohne Gradientenvariation (Punkt F2) und ohne Dispersionskorrektur (Punkt F6):
  • r=50, Δxy=0,74 µm, h=36 µm, durchschnittl. Materialhöhe 14,94 µm, alle Zielpositionen 0%.
• Teilbild b) mit Gradientenvariation (Punkt F2), aber ohne Dispersionskorrektur (Punkt F6):
  • r=46, Δxy=0,80µm, h=36µm, durchschnittl. Materialhöhe 14,38 µm, Zielpositionen 1. 0%, 2. 0%, 3. -20%, 4. -20%. 5. -25%, 6. +7%, 7. +35%, 8. +35%.
• Teilbild c) mit Gradientenvariation (Punkt F2) und mit 1x Dispersionskorrektur (Punkt F6):
  • r=97, Δxy=0,50 µm, h=47,4 µm, durchschnittl. Materialhöhe 13,53 µm, Zielpositionen wie bei Teilbild b).
• Teilbild d) mit Gradientenvariation (Punkt F2) und Dispersionskorrektur (Punkt F6) in 3 Spalten:
  • r=97, Δxy=0,50µm, h=47,5µm, durchschnittl. Materialhöhe 14,46 µm, Zielpositionen 1. 0%, 2. 0%, 3. 0%, 4. +10%. 5. -5%, 6. -15%, 7. +25%, 8. +30%.

Diffraction grating frame data: N = 17, diagonal layout, exposure angle α = 12 °, s _L = 1.00 μm, p _y = 7.653 μm (partial images a) and b)) or 10.083 μm (partial image c)) or 10.098 μm ( Partial image d)), aspect ratio r and minimum distances Δxy symmetrical, but different depending on the partial image (as well as the grid height), maximum material gradient 30: 1 = c × 22.5: 1 with c = 133%. In the FIG. 18 and 19 corresponds to 100% gradient respectively to the gradient 22.5: 1. aspect ratios r, minimum distances Δxy, grid heights h, target positions per partial image:

• Partial image a) without gradient variation (point F2) and without dispersion correction (point F6):
  • r = 50, Δxy = 0.74 μm, h = 36 μm, avg. Material height 14.94 μm, all target positions 0%.
• Partial image b) with gradient variation (point F2), but without dispersion correction (point F6):
  • r = 46, Δxy = 0.80μm, h = 36μm, avg. Material height 14.38 μm, target positions 1. 0%, 2. 0%, 3. -20%, 4. -20%. 5.-25%, 6. + 7%, 7. + 35%, 8. + 35%.
• Partial image c) with gradient variation (point F2) and with 1x dispersion correction (point F6):
  • r = 97, Δxy = 0.50 μm, h = 47.4 μm, avg. Material height 13.53 μm, target positions as in sub-picture b).
• Partial image d) with gradient variation (point F2) and dispersion correction (point F6) in 3 columns:
  • r = 97, Δxy = 0.50μm, h = 47.5μm, avg. Material height 14.46 μm, target positions 1. 0%, 2. 0%, 3. 0%, 4. + 10%. 5. -5%, 6. -15%, 7. + 25%, 8. + 30%.

• G0: Jeweils drei benachbarte von q=11 Schlitzen in G₀ geöffnet zu 30% Öffnungsanteil (d.h., G₀ wiederholt sich in x-Richtung alle q p₀=11 x 26,6574µm = 293,23 µm).
• G₁: wie im Beispiel C
• G_L: Nickel/Photolack wie im Beispiel C, Lichtbrechung 1:8000 für 19 keV bis 1:20 000 für 30 keV; d_L2=211,496 mm.
• G2: p₂=38,4593µm bzw. Detektor-Pixelbreite ca. 19.23µm (perfekt absorbierend simuliert).

Overall structure Frame data of the optical simulation (only differences to the FIG. 17 given underlying simulation, in particular p ₀ , p ₁ , d ₀₁ , d _{1L are} unchanged):

• G0: Each three adjacent of q = 11 slots in G ₀ open to 30% opening fraction (ie, G ₀ repeats in the x direction all qp ₀ = 11 x 26,6574μm = 293,23 microns).
• G ₁ : as in example C
• G _L : nickel / photoresist as in example C, refraction of light 1: 8000 for 19 keV to 1:20 000 for 30 keV; d _L2 = 211.496 mm.
• G2: p ₂ = 38.4593μm or detector pixel width approx. 19.23μm (perfectly absorbing simulated).

Optical performance (spectrum as in example C), again based on visibility of various boundary conditions and sections (columns). The transmission is about 75.8% with diffraction grating (77% in panel c)) and about 94.2% without diffraction grating. Boundary condition [G ₀ , G ₂ perfectly absorbent] total visibility Visibility in the columns with the ... partial prisms Central column 1st strongest 5. middle 8. weakest b) Without F6, Δxy = 0.8μm 42.5% 29.4% 43.6% 48.1% 49.8% b) With F6, Δxy = 0.5μm 43.6% 36.6 Without refraction grid 50.9% --- --- --- 50, 9%

The visibility of the strongest subprisms increases by about 7%. The entire grid only benefits with 1% gain, as only two out of 17 columns are among the strongest. It will now be shown how the visibility distance to the situation without diffraction grating can be reduced. The layout and the number of open slots per qp ₀ in G _{0 are} vertically varied according to the following table. For the previous table this value was 3 (a reduction of the gradient compression from c = 4/3 to c = 7/6 would bring a further improvement). Boundary condition [G ₀ , G ₂ perfectly absorbent] Total visibility at ... columns in G ₀ per qp = 203,23μm 1 2 3 4 a) Without F2 / F6, 0.74μm 48.4% 46.5% 43.4% 39.3% b) Without F6, Δxy = 0.8μm 47.5% 45.6% 42.5% 38.4% c) 1x F6, Δxy = 0.5μm 48.9% 46.8% 43.6% 39.3% d) 3x F6, Δxy = 0.5μm 50.4% 48.5% 45.4% 41.1% e) as d), but d = 191,78mm _L2 ↔ Q = 12 QP = ₀ 319,89μm 50.5% 48.8% 46.2% 42.6%

The shortening of the distance improves the visibility only slightly and only between d) and e), because the highest photon energies are a little more broken.

The invention is not limited to the embodiments described above. Rather, other variants of the invention can be derived therefrom by the person skilled in the art without departing from the subject matter of the invention. In particular, all the individual features described in connection with the exemplary embodiments can also be combined with one another in other ways, without departing from the subject matter of the invention. The object of the invention is defined by the claims.

Appendix: Physical Basics

The phase contrast of an X-ray image visualizes different phase velocities c _p = c ₀ / (1-δ) = c ₀ (1 + δ) due to the material-dependent refractive index n = 1-δ + iβ.

The transmission T behind the patient is determined by the intensity ratio T = I / I ₀ , where I _{0 is} the average intensity without G ₁ and G ₂ . The noise (represented by the standard deviation σ _φ ) is proportional to (Δt I ₀ ) ^-2 . To halve the noise, so the quadruple dose .DELTA.t I _{0 is} required. In differential phase contrast (as well as in dark field recordings), there is a proportional dependence of the noise σ _φ on the strip phase φ (where V is the (stripe) visibility or visibility $$\mathrm{V}=\left({\mathrm{I}}_{\mathrm{Max}}-{\mathrm{I}}_{\min }\right)/\left({\mathrm{I}}_{\mathrm{Max}}-{\mathrm{I}}_{\min }\right)$$

wobei Sichtbarkeit V und die Transmission T jeweils in den Grenzen 0-100% variieren können.The quantities I _max and I _min denote the maximum / minimum intensities depending on the x position of a displaceable G ₀ : $${{\mathrm{\sigma }}_{\mathrm{\phi }}}^{\mathrm{2}}\alpha \mathrm{1}/\left({\mathrm{\Delta t\; I}}_{\mathrm{0}}{\mathrm{V}}^{\mathrm{2}}\mathrm{T}\right).$$

visibility V and transmission T may each vary within the limits 0-100%.

In order to keep the noise at a certain value, Δt I ₀ V ² T must remain unchanged. An efficiency η of the optics behind the patient can be defined as: $${\mathrm{\eta }}_{\mathrm{\phi }}={\mathrm{V}}^{\mathrm{2}}\mathrm{T}$$

An aim of the present application is to optimize η _φ , ie to achieve a dose minimization for given σ _φ (as opposed to σ _{Φ '} or σ _Φ ), where Φ is the phase of the actual Wavefront and Φ 'whose spatial change Φ' = ∂Φ / ∂x denote. Φ 'is given by $$\mathrm{\Phi }\text{'}=\partial \mathrm{\Phi }/\partial \mathrm{x}={\mathrm{\phi \; p}}_{\mathrm{2}\mathrm{L}}/\left({\mathrm{\lambda \; d}}_{\mathrm{1}\mathrm{L}}\right)$$

With the definition of the sensitivity S as S = d _1L / P _L , for a given design wavelength λ = λ _D $${\mathrm{\sigma }}_{\mathrm{\Phi }\text{'}}\alpha {\mathrm{\sigma }}_{\mathrm{\phi }}/\mathrm{S}$$

folgt wie oben bei vorgegebenem Rauschen σ_Φ' dass die Dosis Δt·I₀ proportional ist mit η_Φ mit $$\mathrm{\eta }={\mathrm{\eta }}_{\mathrm{\Phi }}={\mathrm{S}}^{\mathrm{2}}{\mathrm{V}}^{\mathrm{2}}$$

The higher the sensitivity, the lower the noise. Out $${{\mathrm{\sigma }}_{\mathrm{\Phi }\text{'}}}^{\mathrm{2}}\alpha {{\mathrm{\sigma }}_{\mathrm{\phi }}}^{\mathrm{2}}/{\mathrm{S}}^{\mathrm{2}}\alpha \mathrm{1}/\left({\mathrm{\Delta t\; I}}_{\mathrm{0}}{\mathrm{S}}^{\mathrm{2}}{\mathrm{V}}^{\mathrm{2}}\mathrm{T}\right)$$

follows as above with given noise σ _{Φ '} that the dose .DELTA.t · I _{0 is} proportional to η _Φ with $$\mathrm{\eta }={\mathrm{\eta }}_{\mathrm{\Phi }}={\mathrm{S}}^{\mathrm{2}}{\mathrm{V}}^{\mathrm{2}}$$

Claims (11)

1. Refracting grid (G_L) for a phase contrast X-ray imaging device (2) having a transverse surface (10) to be aligned essentially crosswise to a direction of incident radiation, said transverse surface (10) being spanned by an x axis and a y axis perpendicular thereto, and having a plurality of refracting bars (14) made of a visually comparatively thin base material, said refracting bars (14) being arranged in alternation with visually thicker intermediate spaces (16), characterised in that the refracting bars (14) are designed such that

   - they divide the transverse surface (10) into refracting strips (12) elongated in the y direction in each case, which are arrayed in the x direction parallel to one another, wherein adjacent refracting strips (12) are always aligned to different focus points (F), and

   - within the transverse surface (10) they run diagonally at least in sections, wherein the lateral surfaces (18) of at least one refracting bar (14), that delimit said refracting bar (14) in the x direction, in each case extend over a plurality of refracting strips (12).
2. Refracting grid (G_L) according to claim 1,
   wherein the refracting bars (14) are each formed in the manner of oblique prisms inclined in the y direction, whose base and top surface lie in the end surfaces of the refracting grid (G_L) parallel to the transverse surface (10).
3. Refracting grid (G_L) according to claim 2,
   wherein the refracting bars (14) are arranged such that in each refracting strip (12) a material structure repeated in the y direction with a y period length (p_y) results, and wherein the refracting bars (14) are inclined in the y direction such that a top surface of each refracting bar (14) opposing the base is offset in respect of the base by a whole number of period lengths (p_y), in particular exactly one period length (p_y).
4. Refracting grid (G_L) according to one of claims 1 to 3,
   wherein each refracting bar (14) adjoins the intermediate spaces (16) arranged between the refracting bars (14) by two lateral surfaces (18) in each case, and wherein the lateral surfaces (18) are alternately composed of active part surfaces (20) with a comparatively strong refractive effect in the x direction and passive partial surfaces (22) with a low or negligible refractive effect in the x direction.
5. Refracting grid (G_L) according to claim 4,
   wherein each active or passive part surface (20,22) extends in the x direction across a whole number of refracting strips (12).
6. Refracting grid (G_L) according to one of claims 1 to 5,
   wherein each refracting bar (12) within the transverse surface (10) runs in alternating sections diagonally in the positive y direction and in the negative y direction.
7. Refracting grid (G_L) according to claim 6,
   wherein the course of each refracting bar (12) within the transverse surface (10) in each case changes direction at refracting strips (12) with a low or negligible refractive effect.
8. Refracting grid (G_L) according to one of claims 1 to 7,
   wherein in each case a uniformly predetermined number of refracting strips (12) is aligned on a common focus point (F).
9. Phase contrast X-ray imaging device (2) having an X-ray source (4), having a phase grid (G₁), having an X-ray detector (6) which has a one-dimensional or two-dimensional array of pixels (P) and having a refracting grid (G_L) according to one of claims 1 to 8 arranged between the phase grid (G₁) and the X-ray detector (6).
10. Phase contrast X-ray imaging device (2) according to claim 9, having an additional coherence grid (G₀) which is connected between the X-ray source (4) and the phase grid (G₁).
11. Phase contrast X-ray imaging device (2) according to claim 9 or 10, having an additional analysis grid (G₂) which is connected between the refracting grid (G_L) and the X-ray detector (6).

Images