METHOD FOR CHANGING A SURFACE VIA PARTICLE IRRADIATION

Abstract

A method for changing a shape of a surface of an optical element by particle irradiation includes: modelling the problem of determining a resulting change of the surface shape of the optical element from a control variable; determining a predefinition for the control variable of the particle irradiation from a predefined desired change of a surface shape of the optical element by ascertaining an extremum of a merit function; and radiating particles onto the surface of the optical element with a locally resolved effect distribution corresponding to the determined predefinition for the control variable, for the purpose of producing local surface changes at the surface of the optical element. Ascertaining the extremum corresponds to the solution of an Euler equation. The Euler equation defines an integral operator. The eigenvalues of the integral operator are determined, and the predefinition is a linear combination of a finite number of eigenfunctions of the integral operator.

Claims

1. A method, comprising: modelling a problem of determining a resulting change of a surface shape of an optical element from a control variable of a particle irradiation; determining a predefinition for the control variable of the particle irradiation from a predefined desired change of a surface shape of the optical element by ascertaining an extremum of a merit function; and radiating particles onto the surface of the optical element with a locally resolved effect distribution corresponding to the determined predefinition for the control variable, thereby producing local surface changes at the surface of the optical element, wherein: ascertaining of the extremum corresponds to a solution of an Euler equation; the Euler equation defines an integral operator; eigenvalues of the integral operator are determined; and the predefinition is a linear combination of a finite number of eigenfunctions of the integral operator.

2. The method of claim 1, wherein modelling comprises carrying out a convolution of the control variable with a tool function describing an effect potential of a particle irradiation unit effecting the particle irradiation.

3. The method of claim 1, wherein the eigenvalues of the eigenfunctions in each case do not fall below a predefined limit value.

4. The method of claim 1, wherein a magnitude of the eigenvalues of the integral operator corresponds in each case to an efficiency of the particle irradiation unit on the surface of the optical element in relation to a respectively assigned structure of a shape changed to be produced in the surface.

5. The method of claim 1, wherein the integral operator is calculated numerically as a diagonal matrix.

6. The method of claim 5, wherein the diagonal matrix is formed via a tool function describing the effect potential of the particle irradiation unit.

7. The method of claim 1, wherein modelling leads to an ill-defined inverse function configured to select a suitable solution from a cluster of solutions to the ill-defined inverse problem.

8. The method of claim 1, wherein the merit function comprises predefined coefficients which characterize the locally resolved effect distribution.

9. The method of claim 1, wherein the merit function contains a Tichonov regularization.

10. The method of claim 1, wherein the optimization of the merit function is carried out on the basis of external constraints.

11. The method of claim 10, wherein the control variable comprises a residence time of a particle beam on a corresponding point of the surface during the irradiation, and the external constraints comprise a low-pass filter for the residence time of the particle beam.

12. The method of claim 10, wherein the control variable comprises a residence time of a particle beam on a corresponding point of the surface during irradiation, and the external constraints comprise gradient limitations to the residence time of the particle beam.

13. The method of claim 1, wherein the particles are radiated onto the surface of the optical element with a predefined optical footprint and eigenfunctions assigned to the eigenvalues satisfy boundary conditions adapted to the optical footprint.

14. A method, comprising: measuring a wavefront aberration of a projection lens of a microlithographic exposure apparatus; demounting a first optical element of the projection lens; changing the shape of the surface of a second optical element via a method comprising: modelling a problem of determining a resulting change of a surface shape of the second optical element from a control variable of a particle irradiation; determining a predefinition for the control variable of the particle irradiation from a predefined desired change of a surface shape of the second optical element by ascertaining an extremum of a merit function; and radiating particles onto the surface of the second optical element with a locally resolved effect distribution corresponding to the determined predefinition for the control variable, thereby producing local surface changes at the surface of the second optical element; and installing the second optical element instead of the first optical element into the projection lens, wherein: ascertaining of the extremum corresponds to a solution of an Euler equation; the Euler equation defines an integral operator; eigenvalues of the integral operator are determined; and the predefinition is a linear combination of a finite number of eigenfunctions of the integral operator.

15. The method of claim 14, wherein modelling comprises carrying out a convolution of the control variable with a tool function describing an effect potential of a particle irradiation unit effecting the particle irradiation.

16. The method of claim 14, wherein the eigenvalues of the eigenfunctions in each case do not fall below a predefined limit value.

17. The method of claim 14, wherein a magnitude of the eigenvalues of the integral operator corresponds in each case to an efficiency of the particle irradiation unit on the surface of the optical element in relation to a respectively assigned structure of a shape changed to be produced in the surface.

18. The method of claim 14, wherein the integral operator is calculated numerically as a diagonal matrix.

19. The method of claim 14, wherein modelling leads to an ill-defined inverse function configured to select a suitable solution from a cluster of solutions to the ill-defined inverse problem.

20. A method, comprising: modelling a problem of determining a resulting change of a surface shape of an optical element from a control variable of a particle irradiation; determining a predefinition for a control variable of a particle irradiation from a predefined desired change of a surface shape of an optical element by ascertaining an extremum of a merit function; and radiating particles onto the surface of the optical element with a locally resolved effect distribution corresponding to a determined predefinition for the control variable, thereby producing local surface changes at the surface of the optical element, wherein: ascertaining of the extremum corresponds to a solution of an Euler equation; the Euler equation defines an integral operator; eigenvalues of the integral operator are determined; and the predefinition is a linear combination of a finite number of eigenfunctions of the integral operator.

Description

BRIEF DESCRIPTION OF THE DRAWINGS

[0041] The above and further advantageous features of the disclosure are illustrated in the following detailed description of exemplary embodiments according to the disclosure with reference to the accompanying schematic drawings, in which:

[0042] FIG. 1 shows one exemplary embodiment of a device for changing a shape of a surface of an object via particle irradiation in the form of an electron beam figuring device including a control unit, via which a merit function containing eigenvalues is optimized;

[0043] FIG. 2 shows one example of an eigenfunction .sub.1 assigned to an eigenvalue .sub.1;

[0044] FIG. 3 shows one example of an eigenfunction .sub.2 assigned to an eigenvalue .sub.2;

[0045] FIG. 4 shows one example of an eigenfunction .sub.3 assigned to an eigenvalue .sub.3;

[0046] FIG. 5 shows one example of an eigenfunction .sub.6 assigned to an eigenvalue .sub.6; and

[0047] FIG. 6 shows one example of an eigenfunction .sub.7 assigned to an eigenvalue .sub.7.

DETAILED DESCRIPTION OF EXEMPLARY EMBODIMENTS

[0048] In the exemplary embodiments or embodiments or embodiment variants described below, elements which are functionally or structurally similar to one another are provided with the same or similar reference signs as far as possible. Therefore, for understanding the features of the individual elements of a specific exemplary embodiment, reference should be made to the description of other exemplary embodiments or the general description of the disclosure.

[0049] In order to facilitate the description, FIG. 1 indicates a coordinate system which reveals the respective positional relationship of the components illustrated in the figures. In this case, the x-direction extends towards the right, and the z-direction extends upwards. The formulae indicated in FIG. 1 relate to a model which is limited to one dimension (x) for the sake of simplicity. Therefore, FIG. 1 dispenses with indicating the y-coordinate axis. With corresponding extension to a two-dimensional model (x/y), the coordinate system from FIG. 1 should be extended by the y-coordinate axis extending perpendicularly to the plane of the drawing.

[0050] FIG. 1 schematically illustrates one exemplary embodiment of a device 10 for changing a shape of a surface 12 of an object in the form of an optical element 14 via electron irradiation. By way of example, a mirror for the EUV wavelength range, i.e. for electromagnetic radiation having a wavelength of less than 100 nm, in particular a wavelength of approximately 13.5 nm or approximately 6.7 nm, is provided as optical element 14. This can be a mirror for a projection lens or for a mask illumination system of a projection exposure apparatus for EUV microlithography. However, the device 10 is also suitable for highly accurate surface shape production or surface shape change for other optical elements, such as, for example, mirrors for other wavelength ranges, lens elements or optical elements having diffractive structures, or else for objects which do not form an optical element.

[0051] The device 10 contains a particle irradiation unit 16 for generating a particle beam 18 that is directed and focused onto selectable locations of the surface 12. In the case illustrated, electrons are involved and the particle irradiation unit 16 is thus an electron irradiation unit. In other embodiment variants, it is also possible to use ions, for instance in apparatuses for IBF surface processing, or uncharged particles. With the use of ions, the latter are typically neutralized after acceleration in the electric field and before impingement on the surface. The particle beam 18 is configured energetically in particular in such a way that a more or less greatly pronounced local removal or a more or less greatly pronounced local compaction of the material of the optical element 14 at the surface 12 is brought about depending on the energy dose. Energy dose should be understood to mean the energy per area which is introduced into the optical element 14 by the particle beam 18. The energy dose is thus dependent in particular on the residence duration of the electron beam at the selected location and on the intensity of the electron beam.

[0052] Both a material removal and a local compaction bring about a local depression of the surface 12 at the optical element 14. A compaction occurs as a result of a redistribution of electron bonds in particular in amorphous materials. In this case, the local compaction is effected in all spatial directions, that is to say that not only a local surface depression in the region of an area element in the negative z-direction but also a compaction parallel to the surface 12 takes place. In this case, forces acting parallel to the surface 12 arise, as a result of which stresses are induced in the optical element 14. The stresses can bring about a deformation of a surface section that is significantly larger than the area element affected by the local compaction. The surface section can include a part of the surface 12 or else the entire surface 12.

[0053] In order to generate the particle beam 18 in the form of an electron beam, the particle irradiation unit 16 contains an electron source 20 and an acceleration unit 22. By way of example, an incandescent cathode, a crystal cathode or a field emission cathode can be used as the electron source 20. The acceleration unit 22 accelerates and focuses the electrons emitted by the electron source 20. To that end, the acceleration unit 22 can have an anode having a high positive electrostatic potential relative to the electron source 20 and having a small exit opening for the accelerated electrons. For focusing and for setting the intensity of the particle beam 18, the acceleration unit 22 furthermore contains a control electrode, for example a Wehnelt cylinder. The intensity or the beam current indicates the number of electrons which pass through an imaginary area perpendicular to the electron beam per unit time.

[0054] For focusing the particle beam 18 coming from the acceleration unit 22, the particle irradiation unit 16 furthermore includes a focusing unit 24 having suitably configured electrical or magnetic components.

[0055] Via a deflection unit 26 of the particle irradiation unit 16, the particle beam 18 can be deflected in the x-direction in the illustration restricted to one dimension in accordance with FIG. 1, generally both in the x-direction and in the y-direction. For this purpose, the deflection unit 26 likewise contains suitably configured electrical or magnetic components. Depending on the setting of the deflection unit 26, the particle beam 18 impinges on the surface 12 of the optical element 14 at a specific location (x in accordance with the illustration in FIG. 1 or generally (x, y)).

[0056] In this way, a multiplicity of different locations of the surface 12 can be irradiated successively and a locally resolved effect distribution in the form of an energy dose distribution over the surface 12 can thus be achieved. The locally resolved energy dose distribution should be understood here to mean a distribution of the introduced energy per area as a function of the surface coordinate x (generally the surface coordinates (x, y)) of the surface 12 of the optical element 14. In this case, the irradiation can be carried out for example in a raster-like manner or else continuously over the entire surface. Moreover, an irregular or regular arrangement of different locations to be irradiated, for instance in lines, circles, ellipses or the like, is possible.

[0057] In order to avoid absorption of the electrons of the particle beam 18 by air, the particle irradiation unit 16 furthermore has a vacuum chamber, in which the electron source 20, the acceleration unit 22, the focusing unit 24, the deflection unit 26 and the optical element 14 or at least the surface 12 of the optical element 14 are arranged.

[0058] The device 10 furthermore contains a control unit 30 for controlling the particle irradiation unit 16. The control unit 30 is configured in particular to determine, from a predefined desired change 32 for the shape of the surface 12 of the optical element 14, a predefinition 34 for a control variable 36 of the particle irradiation unit 16 for the irradiation of the surface 12 with the particle beam 18, with which the desired change 32 of the shape of the surface 12 is achieved very accurately. The predefinition 34 of the control variable 36 brings about the production of the above-described effect distribution in the form of the energy dose distribution in a suitable manifestation.

[0059] For the purpose of ascertaining the control variable predefinition 34, the control unit 30 is configured for optimizing a merit function 38. The merit function 38 includes eigenvalues 52 of an operator 48 which results from a modelling of the problem of determining a resulting change 42 of the surface shape of the optical element 14 from the control variable 36. The modelling of the problem is carried out via a modelling unit 40. The modelling unit 40 can be part of the device 10 or else form a separate device. In the present embodiment, the modelling unit 40 is not part of the device 10.

[0060] For the purpose of modelling the problem, the change 42 of the surface shape is described as a convolution 46 of a tool function 44 describing the effect potential of the particle irradiation unit 16 with the control variable 36. The modelling results in an operator A and multiplying the latter by its adjoint operator A* results in the operator 48 in the form of a diagonal matrix. By solving the eigenvalue problem 50 described via the eigenvalue equation

K.Math..sub.n(x)=.sub.n19 .sub.n(x)(1)

the eigenvalues 52 are ascertained and transferred to the control unit 30. In expression (1), K denotes the operator 48, .sub.n denotes the eigenvalues 52 and .sub.n(x) denotes associated eigenfunctions.

[0061] Furthermore, the merit function 38 includes coefficients 54 characterizing the locally resolved effect distribution of the particle beam 18 incident on the surface 12. The coefficients 54 are likewise determined by the modelling unit 40 and communicated to the control unit 30. The embodiment of the optimization module 30 and of the modelling unit 40 as illustrated in FIG. 1 is explained in detail below with regard to construction and functioning.

[0062] In many applications in which surfaces 12 are processed with the aid of particle beams 18, the surface change, which can be represented by a function of the spatial coordinates, is linearly dependent on the residence time of the processing tool, in the present case the particle irradiation unit 16, over the surface 12.

[0063] As already mentioned above, changes 42 of the surface shape along only one coordinate x are considered in order to facilitate the description. In the present case, the changes 42 of the surface shape are designated h(x). The effect of the tool in the form of the particle irradiation unit 16 is characterized by the so-called tool function 44, which is designated by f.sub.W(x), wherein the surface changes established as a result of the tool being situated above a point during the residence duration t(x), which represents the control variable 36, can be written as the abovementioned convolution 46 as follows:

h(x)=.sub..sup.f.sub.W(xx)t(x)dx(2)

[0064] It should be noted that the effect range of the particle beam 18 can be limited by stops and other devices. In this case, the integration range is likewise limited. The following form is formulated here for the tool function 44:

[00001]$\begin{array}{cc}{f}_{}\left(x\right)=\frac{1}{\sqrt{2.Math..Math..Math.{}^2}}.Math.e^{-\frac{x^2}{2.Math.{}^2}}.& \left(3\right)\end{array}$

[0065] A typical task in surface processing is the calculation of machine control predefinitions, such as the residence duration t(x), from a predefined desired change 32 of the surface shape, which is designated below by {tilde over (h)}(x). The desired surface change typically need not be obtained for all spatial points, but rather for a specific range, in our case the evaluation interval [L.sub.f, L.sub.f]. As will become clear later, it is not expedient to attempt to ascertain the residence duration t(x) by solving the equation

{tilde over (h)}(x)=.sub..sup.f.sub.(xx)t(x)dx(4)

[0066] This involves a typical inverse problem which is ill defined in the Hadamardian sense.

[0067] The simplest and most common procedure is offered by the method of least squares in the sense of the L.sup.2-norm on the space of the square-integrable functions over the evaluation range [L.sub.f, L.sub.f]. In principle, other methods would be conceivable, too, in which the merit function is defined by an L.sup.1-norm, for example the method of least absolute deviations with the L.sup.1-norm. These variants will not be considered in further detail here. The customary quadratic merit function (t) is defined by way of the L.sup.2-norm of the difference between the surface change obtained via a residence duration t(x) and the desired surface change (h(x){tilde over (h)}(x)):

h{tilde over (h)}.sup.2.sub.L.sub.2.sub.([L.sub.f.sub.,L.sub.f.sub.])=.sub.L.sub.f.sup.L.sup.f(h(x){tilde over (h)}(x)).sup.2dx (5)

and has the form

({t})=.sub.L.sub.f.sup.L.sup.f(.sub..sup.f.sub.(xx)t(x)dx{tilde over (h)}(x)).sup.2dx. (6)

[0068] It should be noted that this merit function is defined apart from a constant. A residence duration t(x) which minimizes this merit function and possibly satisfies additional constraints is regarded as a solution to the inverse problem.

[0069] The present problem can be tackled in a natural way with the aid of functional analytical concepts. The effect of the tool, which acts during the residence duration t(x), on the surface shall be described with the aid of an operator A(L.sub.f,):L.sup.2().fwdarw.L.sup.2([L.sub.f,L.sub.f]):

(A(L.sub.f,)t)(x)=.sub..sup.f.sub.(xx).sub.[L.sub.f.sub.,L.sub.f.sub.](x)dx, (7)

which is taken into account only in the relevant range [L.sub.f, L.sub.f] (X [L.sub.f, L.sub.f] is the indicator function of the interval [L.sub.f, L.sub.f]). A merit function is then written as

({t})=A(L.sub.f,)t{tilde over (h)}.sup.2.sub.L.sub.2.sub.([L.sub.f.sub.,L.sub.f.sub.])==A*(L.sub.f,)A(L.sub.f,)t,A*(L.sub.f,){tilde over (h)},+{tilde over (h)}.sup.2.sub.L.sub.2.sub.([L.sub.f.sub.,L.sub.f.sub.]), (8)

wherein the adjoint operator A*(L.sub.f,):L.sup.2([L.sub.f,L.sub.f]).fwdarw.L.sup.2() is given by

(A*(L.sub.f,)f)(x)=.sub.L.sup.Lf.sub.(xx).sub.[L.sub.f.sub.,L.sub.f.sub.](x)dx(9)

[0070] The extremum of the merit function is determined by the Euler equation

A*(L.sub.f,)A(L.sub.f,)t=A*(L.sub.f,){tilde over (h)}, (10)

[0071] The operatoralso designated by the reference sign 48 in FIG. 1

K(L.sub.f,)=A*(L.sub.f,)A(L.sub.f,)A(L.sub.f,) (11)

is a self-adjoint integral operator with the kernel

[00002]$\begin{array}{cc}\begin{array}{c}{g}_{}\left(x^{},x^{}\right)=.Math.{}_{-L_f}^{L_f}.Math.f_{}\left(x-x^{}\right).Math.f_{}\left(x-x^{}\right).Math.\mathrm{dx}\\ ==.Math.\frac{1}{2.Math.{}^2}.Math.e^{-\frac{{\left(x^{}-x^{}\right)}^2}{4.Math.{}^2}}.Math.{}_{-L_f}^{L_f}.Math.e^{-\frac{{\left(x-\frac{x^{}+x^{}}{2}\right)}^2}{{}^2}}.Math.\mathrm{dx}\\ =.Math.\frac{1}{2}.Math.\frac{1}{\sqrt{4.Math..Math..Math.{}^2}}.Math.e^{-\frac{{\left(x^{}-x^{}\right)}^2}{4.Math.{}^2}}\\ .Math.\left[\mathrm{erf}.Math.\left(\frac{x^{}+x^{}+2.Math.L_f}{2.Math.}\right)-\mathrm{erf}\left(\frac{x^{}+x^{}-2.Math.L_f}{2.Math.}\right)\right].\end{array}& \left(12\right)\end{array}$

[0072] Furthermore, K(L.sub.f, ) is a Hilbert-Schmidt operator. For the solution to equation (10), it is desirable to ascertain the (discrete) spectrum of K(L.sub.f, ) together with the associated eigenspaces, i.e. the eigenvalues .sub.n are ascertained by solving the eigenvalue problem 50 described via the eigenvalue equation indicated under expression (1).

[0073] Since K(L.sub.f, ) is strictly positive in the sense of ordered vector spaces, i.e. for arbitrary strictly positive residence durations the associated change is strictly positive, the highest eigenvalue .sub.n of K(L.sub.f, ) is strictly positive. The associated eigenfunction .sub.n is likewise strictly positive, which follows from the Krein-Rutman theorem. The simple estimation of this eigenvalue follows from Young's inequality:

.sub.1<.sub..sup.f.sub.(x)dx=1 (13)

[0074] Furthermore, K(L.sub.f, ) has no spectral gap, lim.sub.n.fwdarw..sub.n=0.

[0075] That is to say that the numerical calculation considers the kernel (expression (12)) of the operator K(L.sub.f, ) over a finite interval [L; L], wherein L is chosen to be much greater than L.sub.f. The continuous x-coordinate is replaced by an equidistant grid, which is fine enough.

[0076] The eigenfunction .sub.1 with respect to the highest eigenvalue .sub.1 is illustrated in FIG. 2 and is strictly positive (.sub.1=0.98883516485, L=10, L.sub.f=7). It falls to zero at the edge. Higher-order eigenfunctions with respect to eigenvalues .sub.2, .sub.3 . . . have oscillations, wherein the number of oscillation periods increases with the function ordersee FIGS. 3 and 4 for eigenfunctions .sub.2 and .sub.3 with respect to the eigenvalues .sub.2 (.sub.2=0.9560906702, L=10, L.sub.f=7) and .sub.3 (.sub.3=0.90393245233, L=10, L.sub.f=7).

[0077] Within the interval, the eigenfunctions .sub.n are described very well by the sine functions

[00003]$\sin .Math..Math.\frac{2.Math.}{2.Math.L_f}.Math.{}_n.Math.x$

and the cosine functions

[00004]$\cos .Math..Math.\frac{2.Math.}{2.Math.L_f}.Math.{}_n.Math.x;$

see FIGS. 5 and 6 (eigenfunctions .sub.6 and .sub.7 with respect to the eigenvalues .sub.6=0.66807975974 and .sub.7=0.57777692918). Over the evaluation range [L.sub.f, L.sub.f], the eigenfunctions have a well-defined parity, with frequencies rendered approximately by the formula

[00005]$\begin{array}{cc}{}_n=\left(\frac{1}{2}-\frac{}{4}.Math.\frac{}{2.Math.L_f}\right).Math.n+\mathrm{.Math.}& \left(14\right)\end{array}$

[0078] The eigenvalues .sub.n can be described by the formula

[00006]$\begin{array}{cc}{}_n=e^{-.Math..Math.\frac{{}^2}{2.Math.L_f}.Math.{}_n^2}& \left(15\right)\end{array}$

wherein 0.49352.

[0079] It should be noted that eigenvalues of the Weierstrass transform with respect to the eigenfunctions sine and cosine have a similar (but continuous) spectrum:

W(sin x)=e.sup..sup.2 sin x, W(cos x)=e.sup..sup.2 cos x (16)

[0080] As follows from the previous explanations, the spectrum of the operator K(L.sub.f, ) falls relatively rapidly, specifically all the more rapidly, the greater the width a of the tool function. This has the effect that calculating the direct solutions to the Euler equation (10) involves dividing by very small numbers. Consequently, none of the three Hadamardian conditions is met, i.e.: [0081] 1. The problem of ascertaining the residence time does not always have a solution in the L.sup.2 sense. [0082] 2. If there is a solution, it is not necessarily uniquely determined. [0083] 3. Possible solutions are unstable in relation to changes in the input data (predefinitions for the surface changes).

[0084] The problem of ascertaining the residence time can be regularized in at least two ways. The customary method from a mathematical standpoint is Tikhonov regularization, which involves adding the additional term to the merit function presented in expression (6):

.sub.({t})=A*(L.sub.f,)A(L.sub.f,)t,A*(L.sub.f,){tilde over (h)},{tilde over (h)}.sup.2.sub.L.sub.2.sub.([L.sub.f.sub.,L.sub.f.sub.])+(17)

[0085] The parameter a prevents the L.sup.2-norm of solutions from becoming too large. The other approach is to use, during the calculation of the residence time duration t(x), the decomposition thereof with respect to the basis functions of the operator K(L.sub.f, ), which shall be designated as .sub.n(x)L.sup.2():

[00007]$\begin{array}{cc}t\left(x\right)=\underset{n=1}{\overset{}{.Math.}}.Math.a_n.Math.{}_n\left(x\right).& \left(18\right)\end{array}$

[0086] The input data A*(L.sub.f,){tilde over (h)} are likewise expanded with respect to .sub.n(x):

A*(L.sub.f,).Math.{tilde over (h)}=b.sub.n.Math..sub.n(x) (19)

[0087] In this case, the coefficients b.sub.n are the coefficients 54 already mentioned above, which characterize the locally resolved effect distribution of the particle beam 18 incident on the surface 12.

[0088] The merit function can then be written as a sum:

[00008]$\begin{array}{cc}{}_{}\left(\left\{t\right\}\right)=\underset{n=1}{\overset{}{.Math.}}.Math.\left[\frac{{}_n+}{2}.Math.a_n^2-a_n.Math.b_n\right].& \left(20\right)\end{array}$

[0089] In practice, the sum is terminated starting from a specific order N, such that the merit function represents a finite sum:

[00009]$\begin{array}{cc}{}_{}\left(\left\{t\right\}\right)=\underset{n=1}{\overset{N}{.Math.}}.Math.\left[\frac{{}_n+}{2}.Math.a_n^2-a_n.Math.b_n\right].& \left(21\right)\end{array}$

[0090] This merit function is then used by the control unit 30 as merit function 38 for ascertaining the control variable predefinition 34, which is designated as {tilde over (t)}(x) in FIG. 1, via numerical optimization.

[0091] The sought predefinition {tilde over (t)}(x) of the residence duration can also be limited by various constraints. The physical constraint is positivity of the resulting residence duration,

[00010]$\begin{array}{cc}\underset{n=1}{\overset{N}{.Math.}}.Math.a_n.Math.{}_n\left(x\right)0.& \left(22\right)\end{array}$

[0092] Other constraints are conceivable, thus for example limitations of the surface changes from above, or low-pass filters in the residence time, resulting from the dynamic characteristic of the processing machines. It is advantageous to formulate the possible technical limitations as linear constraints.

[0093] The above description of exemplary embodiments, embodiments or embodiment variants is to be understood to be by way of example. The disclosure effected thereby firstly enables the person skilled in the art to understand the present disclosure and the advantages associated therewith, and secondly encompasses alterations and modifications of the described structures and methods that are also obvious in the understanding of the person skilled in the art. Therefore, all such alterations and modifications, insofar as they fall within the scope of the disclosure in accordance with the definition in the accompanying claims, and equivalents are intended to be covered by the protection of the claims.

LIST OF REFERENCE SIGNS

[0094] 10 Device for changing the shape of a surface [0095] 12 Surface [0096] 14 Optical element [0097] 16 Particle irradiation unit [0098] 18 Particle beam [0099] 20 Electron source [0100] 22 Acceleration unit [0101] 24 Focusing unit [0102] 26 Deflection unit [0103] 30 Control unit [0104] 32 Predefined desired change [0105] 34 Predefinition for the control variable [0106] 36 Control variable [0107] 38 Merit function [0108] 40 Modelling unit [0109] 42 Change of surface shape [0110] 44 Tool function [0111] 46 Convolution [0112] 48 Operator [0113] 50 Eigenvalue problem [0114] 52 Eigenvalues [0115] 54 Coefficients