CN104111592B - Method for realizing variable free illumination pupil based on micro-mirror array - Google Patents

Abstract

The invention provides a method for realizing variable free illumination pupil based on a micro-reflector array, which comprises the steps of light spot position measurement, light spot position optimization, light spot position correction and the like. The CCD camera measures the position parameters of the reflection light spots when each micro-reflector unit is in a static state; substituting the parameters into a grid search ant colony algorithm for optimization to obtain a micro-reflector array rotation angle matrix generating the required pupil morphology; and calculating the position difference of light spots of the measured light intensity distribution and the required light intensity distribution on the target surface after the rotation angle matrix is loaded, and taking the light spots as the rotation angle of the micro-reflector array obtained by feedback correction and optimization, so that the result finally meets the illumination requirement on the target surface. The invention can realize various illumination modes such as non-circular symmetric complex illumination pupils, has high optimization speed, good convergence, high accuracy and strong practicability.

Description

A kind of method realizing variable free illumination iris based on micro reflector array

Technical field

The invention belongs to microlithography, it relates to the implementation method of free illumination iris in a kind of lithography illuminating system, it is specifically related to a kind of method realizing variable free illumination iris based on micro reflector array.

Background technology

Micro-lithography manufactures unicircuit, the technology of liquid-crystal display and other micro structural components, in photoetching, two factors that picture quality rises keying action are that resolving power and Jiao are dark, so better resolving power should be obtained to form the figure of critical size, Jiao keeping suitable again is dark. Along with the development of photoetching process, constantly reducing of lithography node, except reducing exposure wavelength lambda and the numerical aperture NA of projection object lens, the shape changing exposure light source is one of important channel improving photoetching resolution, on this basis, the off-aixs illumination proposed from the nineties develops into the multistage lighting engineering of recent polarization, and along with the further reduction of process factor k1, light source-mask optimisation technique becomes the core of resolution enhance technology of future generation gradually. Utilize light source-mask optimisation technique (SourceandMaskOptimization), lighting engineering able to programme (ProgrammableIllumination) is adopted to realize, such as traditional lighting, even comparatively complicated freely the throwing light on of off-axis illumination, obtains more suitably illumination iris and goes out pupil shape. Free lighting engineering able to programme, compared with off-axis illumination, to more complicated chips wire strip, can increase process window, reduces mask error magnification factor MEEF.

In existing technology, in order to obtain instantaneous variable illumination iris shape flexibly, adopt the micro reflector array (MMA based on MEMS, MicroMirrorArray) illumination iris surface is come, each micro-reflector is driven by the electrostatic equipment being similar to DMD (DMD, DigtalMicroDevice), and can tilt to two orthogonal oblique axles, each micro-reflector produces an illumination spot on target illuminated area, along with the change of micro-reflector angle of rotation.Illumination spot can move freely on target illuminated area, the perfect light source distribution required for formation.

In the structure shown in international monopoly WO2005/02684, micro-reflective array comprises the micro reflector array of the plated surface highly reflecting films of more than 10000, the different positions that the various different reflection angle that speculum is produced by the Quasi-straight light-focusing mirror group between micro-reflector from target illuminated area is converted in pupil plane, in above process, core is exactly the angle of rotation being calculated each micro-reflector unit so that the hot spot distribution on target illuminated area distributes close to required perfect light source. To realize ring illumination in US Patent No. 2010/0265482A1, propose a kind of position optimization algorithm determining micro reflector array angle of rotation, namely the intensity distribution function inciding on micro reflector array is first calculated, and obtain its complementary distribution function, the flare of micro-reflector is changed to the rule of complementary distribution function, to compensate the ununiformity of incoming beam each several part light intensity, it is achieved evenly throw light on.

But this kind of method is only applicable to produce simply, the off-axis illumination of even light distribution, no longer applicable for the free illumination iris producing light intensity and shape complicated change simultaneously. Simultaneously, due to the impact of the various aberration of lighting system able to programme in actual use procedure and optical element bias, the desirable hot spot drawn is distributed with certain difference to the actual hot spot obtained with optimizing, and these differences can affect the score quality in the homogeneity and CD that export hot spot on pupil face.

Summary of the invention

It is an object of the invention to overcome the deficiencies in the prior art, a kind of method realizing variable free illumination iris based on micro reflector array is provided, the method can be applied in the occasion producing the symmetrical illumination iris of complicated non-circumference, further compensates facula position difference, pupil distribution reconstruction accuracy height.

The technical solution used in the present invention is: a kind of method realizing the free illumination iris of continuous variable based on micro reflector array, described method comprises the steps:

Step (1), the location parameter of flare on illumination target face when measure each micro-reflector unit by CCD camera static;

Step (2), the location parameter that step (1) is measured gained flare are brought in grid search ant group algorithm, and optimization produces the micro reflector array angle of rotation matrix of required light distribution;

Step (3), utilize the angle of rotation matrix that step (2) obtains, generate the control signal of each micro-reflector driving mechanism and control micro-reflector and carry out angle rotation, measure now light distribution on illumination target face, using the facula position difference of itself and required light distribution as evaluation index, system is carried out feedback control, until facula position difference meets design requirement.

Further, the method is applicable to utilize micro reflector array to realize the system of any illumination profile, wherein each micro-reflector unit can two dimension continuously rotate, and micro reflector array is imaged onto on target illuminated area by a collimation condensor group, micro reflector array position becomes Fourier's relation with target illuminated area.

Further, by closed loop feedback system, micro reflector array was demarcated before carrying out the measurement of flare position, make each mirror unit of micro reflector array in above-mentioned steps (1), (2), all can normal operation in (3) three processes.

Further, when in described step (1), each micro-reflector unit is in static state, the detailed process measuring its flare location parameter on illumination target face by CCD camera is as follows:

If micro reflector array unit sum is M, provide micro reflector array angle of rotation matrix ��, except micro-reflector unit M to be measured in this angle of rotation matrix_k(k is mirror unit subscript sequence number, and k=1,2, ..., M) angle of rotation be outside 0, controlling other mirror units taking array center by control unit is boundary line, the mirror unit comprising more than boundary line upward deflects �� angle, and following mirror unit deflects down �� angle, and namely �� is:

Wherein �� is the maximum angle of rotation of each mirror unit, target illuminated area is formed one long be the blank region of H, and:

H=2 �� f_collimator��tan(2��)(1-2)

Wherein, f_collimatorFor collimating the focal length of condensor group.

The center position that CCD camera is placed on target illuminated area accepts the emergent light spot of Quasi-straight light-focusing mirror group, it is also carried out data processing by light spot image in computer acquisition CCD camera, each micro-reflector unit is repeated above step, until obtaining the centroid position observed value matrix D of flare corresponding when each micro-reflector unit is in static state⁰(k)��

Further, mirror unit deflection angle more than micro reflector array center can also get other values being less than ��, and the mirror unit deflection angle below center can also get other values being greater than-�� so that only has flare to be measured on effective reception face of CCD.

Further, in step (2), the step of grid search ant group algorithm is:

Step S11: divide Searching Resolution Space region;

Using each micro-reflector unit two dimension angle of rotation as solution space set, by illumination target face grid division, according to the target light intensity distribution situation of required generation, it is determined that the two-dimentional initial rotation angle degree �� of each micro-reflector unit¹(k)=(��_x ¹(k), ��_y ¹(k)) scope as follows:

0 < | ��_x ¹(k) | < ��_upper

0 < | ��_y ¹(k) | < ��_upper(1-3)

Wherein, ��_x ¹(k), ��_y ¹(k) be respectively micro-reflector unit along the x-axis direction with the initial rotation angle degree in y-axis direction,��_cFor the maximum outer coherence factor produced under required target light illumination mode, ��_objFor the area in illumination target face;

By each ��_x ¹(k), ��_y ¹K () is divided into N decile, if the sequence number in certain solution space corresponding to micro-reflector unit is (p, q), when algorithm starts, each micro-reflector finite element rotation angle random is dispersed in solution space etc. somewhere, subregion, obtain initial rotation angle degree as follows:

${{\mathrm{\&theta;}}_x}^1\left(k\right)=0+\frac{{\mathrm{\&theta;}}_{\mathrm{upper}}}{N}\&times;(p+\mathrm{rand})$

${{\mathrm{\&theta;}}_y}^1\left(k\right)=0+\frac{{\mathrm{\&theta;}}_{\mathrm{upper}}}{N}\&times;(q+\mathrm{rand}),(p,q=\mathrm{1,2},...,N)---(1-4)$

Wherein, rand is the randomized number between [0,1];

Step S12: initialize population and individual information element distribution;

Iteration is initial, and setting ant sum is M, and in optimization, maximum cycle is NC_maxInformation Meter volatilization parameter is �� (0 < �� < 1), convergence precision is �� (0 < �� < 1), the difference of the target light distribution that target illumination field light distribution produces with plan after making objective function G equal the loaded and optimized gained angle of rotation matrix of micro reflector array. By the position calculation matrix of each flare of gained in step (1)When correction iteration is initial, each ant k in the position of target illuminated area is:

${D_x}^1\left(k\right)=f_{\mathrm{collimator}}\&times;\tan \left(2{{\mathrm{\&theta;}}_x}^1\left(k\right)\right)+D_x^0\left(k\right)$

${D_y}^1\left(k\right)=f_{\mathrm{collimator}}\&times;\tan \left(2{{\mathrm{\&theta;}}_y}^1\left(k\right)\right)+D_y^0\left(k\right)---(1-5)$

The evaluation function of definition ant k is the objective function G of u point on the target illuminated area residing for it_uWith the objective function G of v point residing for ant l in its field_vSum, that is:

��G_uv=G_u+G_v(1-6)

By position u point to the mobile state transition probability formula of position v point it is:

${P_{\mathrm{uv}}}^k=\frac{{\mathrm{\&tau;}}_{\mathrm{uv}}/d_{\mathrm{uv}}}{\underset{v\&Element;{\mathrm{allowed}}_k}{\mathrm{\&Sigma;}}{\mathrm{\&tau;}}_{\mathrm{uv}}/d_{\mathrm{uv}}},u\&NotEqual;v---(1-7)$

Wherein, d_uvFor distance between u point and v point in target face, ��_uvFor the pheromone intensity of ant k on mobile route, allowed_kThe space networks lattice point set of paths that next step allows away for ant k, the initial pheromone concentration of every ant k position u point is:

��_u ⁰=e^-G(u)(1-8)

Step S13: elitist ants and general ant are determined in grouping;

After loading the micro reflector array initial rotation angle degree matrix Ru shown in (1-4) formula, select the elitist ants of satisfied | G |�ܦ� and its present position on target illuminated area is added in taboo list, by the mobile each ant of formula (1-7), unidirectional search mechanisms is adopted to be moved from the position that objective function is big to the position that objective function is little by ant, when a loop ends, the pheromone intensity on ant k institute mobile route is pressed following formula and is upgraded:

${{\mathrm{\&tau;}}_{\mathrm{uv}}}^{\mathrm{new}}=\mathrm{\&rho;}\&times;{\mathrm{\&tau;}}_{\mathrm{uv}}^{\mathrm{old}}+e^{-G_{\mathrm{uv}}}---(1-9)$

Possible pheromone strength range on each bar optimizing path is limited in [��_min,��_max]:

${{\mathrm{\&tau;}}_{\mathrm{uv}}}^{\mathrm{new}}=\left\{\begin{array}{c}{\mathrm{\&tau;}}_{\min },{{\mathrm{\&tau;}}_{\mathrm{uv}}}^{\mathrm{new}}\&le;{\mathrm{\&tau;}}_{\min }\\ {{\mathrm{\&tau;}}_{\mathrm{uv}}}^{\mathrm{new}},{\mathrm{\&tau;}}_{\min }\&le;{{\mathrm{\&tau;}}_{\mathrm{uv}}}^{\mathrm{new}}\&le;{\mathrm{\&tau;}}_{\max }\\ {\mathrm{\&tau;}}_{\max },{{\mathrm{\&tau;}}_{\mathrm{uv}}}^{\mathrm{new}}\&GreaterEqual;{\mathrm{\&tau;}}_{\max }\end{array}\right.---(1-10)$

Step S14:NC > NC_max, then continue to optimize, otherwise go to step S15.Often taking turns after iteration terminates, upgrade target function value G and the taboo list of ant group, if the target function value of every ant is less than the convergence precision value �� of definition, then goes to step S15, otherwise go to step S13, start next and take turns iterative search;

Step S15: terminate optimizing, according to the two-dimensional position matrix D of current ant k on target illuminated area^optimalK () calculates best miles of relative movement matrix h^optimal(k) and corresponding two-dimentional angle of rotation matrix ��^optimal(k). Wherein,

$\begin{array}{cc}{h_x}^{\mathrm{optimal}}\left(k\right)=D_x^{\mathrm{optimal}}\left(k\right)-D_x^1\left(k\right)& {{\mathrm{\&theta;}}_x}^{\mathrm{optimal}}\left(k\right)=\frac{1}{2}\arctan \left(\frac{{h_x}^{\mathrm{optimal}}\left(k\right)}{f_{\mathrm{collimator}}}\right)\end{array}$

$\begin{array}{cc}{h_y}^{\mathrm{optimal}}\left(k\right)=D_y^{\mathrm{optimal}}\left(k\right)-D_y^1\left(k\right)& {{\mathrm{\&theta;}}_y}^{\mathrm{optimal}}\left(k\right)=\frac{1}{2}\arctan \left(\frac{{h_y}^{\mathrm{optimal}}\left(k\right)}{f_{\mathrm{collimator}}}\right)\end{array}---(1-11)$

Further, angle of rotation matrix �� after step (3) middle correction grid search ant group algorithm optimization^optimalK the step of () is:

Step S21: provide micro reflector array angle of rotation matrix ��, except tested micro-reflector M in this angle matrix_kTwo-dimentional angle of rotation be ��^optimalK () outward, other mirror units are by control unit control, and taking array center position as boundary line, the mirror unit comprising more than boundary line upward deflects �� angle, and following mirror unit deflects down �� angle, and matrix �� is:

Step S22: CCD camera is placed on D on target illuminated area^optiamlK () position, measures each micro-reflector M_kThe physical location of flare barycenter, calculates itself and the upper predetermined position D of target illumination^optimalThe position difference d at (k) place_diff, select d_diffIt is greater than the speculum set M to be corrected that target illumination field allows maximum facula deviation amount �� L_s(s < k, and M_s��M_k), in these speculums, its angle of rotation �� is got from 0���� n point (n >=20), measure micro-reflector unit M respectively_sIn the facula position parameter at this n some place, it is designated as P ' (��, s), is used nonlinear function f_s(��) it is carried out data fitting, represents and be:

$f_s\left(\mathrm{\&theta;}\right)=\underset{j=0}{\overset{n}{\mathrm{\&Sigma;}}}{a}_j{f}_j\left(\mathrm{\&theta;}\right)$

|f_s(��)-P��(��,s)|�ܦ�₁(1-13)

Wherein, a_jFor the coefficient of polynomial fitting, ��₁For data fitting low precision, obtain S function expression from formula (1-13), the flare M to be corrected grid ant group algorithm optimization drawn_sPosition D^optiamlS () substitutes into f_s(��) in expression formula, that is:

$D^{\mathrm{optimal}}\left(s\right)=\underset{j=0}{\overset{n}{\mathrm{\&Sigma;}}}{a}_j{f}_j\left(\mathrm{\&theta;}\right)---(1-14)$

From upper formula, solve �� (s), replaced ��^optimalThe value of identical position in (k).

Step S23: repeating step S22, until the position difference of each mirror unit flare is all less than target illumination field allows maximum facula deviation amount, export the last micro-reflector two dimension angle of rotation matrix �� meeting actual optical system requirement^finnal(k)��

The useful effect of the present invention is:

The present invention is compared with previous free illumination iris implementation method, it is to increase the practicality of optimization method so that it is better combine with light source mask optimizing process, it is to increase the resolving power of etching system, is specifically:

(1) when the present invention adopts each micro-reflector unit of CCD camera record static, the location parameter of flare, ensure that the accuracy inputting data in subsequent algorithm;

(2) the present invention utilizes grid search ant group algorithm to obtain micro reflector array angle of rotation matrix, and algorithm speed is fast, and convergency is good, can produce to comprise traditional lighting, off-axis illumination and the light illumination mode such as multiple that freely throws light on;

(3) the present invention proposes the angle of rotation that algorithm optimization draws to be compensated, and adopts and first screens, and the rear mode measured, simplifies measuring process, and the position of correction flare, reduces the difference going out pupil with ideal.

Accompanying drawing illustrates:

Reference is below in conjunction with accompanying drawing to the detailed description of embodiment, and the various feature and advantage of the present invention can be easier to understand, wherein:

Fig. 1 is the method flow diagram that application the present invention realizes variable free illumination iris based on micro reflector array;

Fig. 2 is free lamp optical system schematic diagram able to programme in the photolithographic exposure system that is suitable for of the present invention; Wherein, excimer laser 1, single micro-compound eye array 2, micro reflector array mirror unit 31, micro reflector array substrate 32, turnover speculum 4, Quasi-straight light-focusing mirror group 5, CCD camera 6, computer 7 and control unit 8;

Fig. 3 is that the CCD camera that the present invention relates to gathers static lower each micro-reflector unit flare schematic diagram on illumination target face; 6 is CCD camera, and B represents flare region, and O is the center in illumination target face, and H is the length in the blank region that flare is formed;

Fig. 4 is the schema based on grid search ant colony optimization algorithm of the present invention;

Fig. 5 is micro reflector array schematic diagram before and after flare position correction on target illumination field, for the flare that micro reflector array center one in the Y direction is listed in target face, Fig. 5 (a) is the position difference curve of light distribution and desirable light distribution before correction; Fig. 5 (b) is the position difference curve of light distribution after correction with desirable light distribution;

Fig. 6 is the free light illumination mode pupil simulate effect figure realized according to the present embodiment; Fig. 6 (a), 6 (b) is the free illumination iris distribution realized; Fig. 6 (c), 6 (d) is the RMS difference value curve of the desirable illumination intensity distribution freely thrown light on intend producing realized.

Embodiment

For making the object of the present invention, technical scheme and advantage are clearly understood, below in conjunction with specific embodiment, and with reference to accompanying drawing, the present invention is described in more detail.

Embodiment 1:

Fig. 1 is the method flow diagram realizing the free illumination iris of continuous variable based on micro reflector array of the present invention, comprises facula position and measures, facula position optimization, and facula position corrects three processes, and concrete steps are as follows:

Step (1), the location parameter of flare on illumination target face when measure each micro-reflector unit by CCD camera static;

Step (2), the location parameter that step (1) is measured gained flare are brought in grid search ant group algorithm, and optimization produces the micro reflector array angle of rotation matrix of required light distribution;

Step (3), utilize the angle of rotation matrix that step (2) obtains, the control signal generating each micro-reflector controls it and carries out angle rotation, measure now light distribution on illumination target face, using the facula position difference of itself and required light distribution as evaluation index, system is carried out feedback control, until facula position difference meets design requirement, trimming process terminates.

As shown in Figure 2, the optical system being applicable to this free illumination iris implementation method by: excimer laser 1, single micro-compound eye array 2, micro reflector array 3, turnover speculum 4, Quasi-straight light-focusing mirror group 5, CCD camera 6, computer 7 and control unit 8 form. Wherein, before excimer laser 1 is positioned at single micro-compound eye array 2, turnover speculum 4 is between micro reflector array 3 and Quasi-straight light-focusing mirror group 5, CCD camera 6 is positioned at the central position in illumination target face, and be connected with computer 7, control unit 8, micro reflector array 3 is connected with control unit 8, and CCD camera 6 is fixed on an accurate translation stage and moves thereupon. The Gaussian beam that excimer laser 1 sends is incident on single micro-compound eye array 2, incoming beam is segmented and focuses on micro reflector array 3 by single micro-compound eye array 2, micro reflector array 3 is positioned on the front burnt face of Quasi-straight light-focusing mirror group 5, it comprises single micro-reflector unit 31 and substrate 32, light beam after reflection is by shining in Quasi-straight light-focusing mirror group 5 after turnover speculum 4, and focuses on by it in CCD camera 6 being positioned at its burnt position, face.In the present embodiment, the number of micro reflector array is 10000, and each mirror unit maximum two dimension angle of rotation is 5 ��.

Step (1) is measured before starting, whole micro-reflector unit is corrected by closed loop feedback system so that it is in step (1), (2), all can normal operation in (3) three processes. When energized, once reset first to micro reflector array, making each little micro-reflector unit 31 parallel with substrate 32, define now micro reflector array 3 and be operated in static state, the process that CCD camera 6 gathers the flare of single micro-reflector unit 31 under its static state is as follows:

Being located in micro reflector array 3, micro-reflector unit sum is M, and each mirror unit maximum two dimension angle of rotation is ��, with tested micro-reflector M_k(k is mirror unit subscript sequence number, and k=1,2 ..., M) it is example, provide micro reflector array 3 one angle of rotation matrix ��, make in this angle of rotation matrix except tested micro-reflector M_kAngle of rotation is outside 0, and other mirror units are controlled by control unit 8, and taking array center position as boundary line, the mirror unit comprising more than boundary line upward deflects �� angle, and following mirror unit deflects down �� angle, that is:

Along with the rotation of mirror unit, as shown in Figure 3, the flare on target illuminated area forms a long blank region for H, only has mirror unit M to be measured in the region_kFlare, when optical aberration and other alignment errors are less, have:

H=2 �� f_collimator��tan(2��)(1-2)

Wherein, f_collimatorFor collimating the focal length of condensor group 5, the miles of relative movement H of hot spot becomes, with the angle of rotation �� of micro-reflector unit, the fourier transformation relation being similar to. CCD camera 6, in the present embodiment its effective feeling light area is placed in the central position in illumination target faceThe position of CCD camera 6 is regulated so that the center of its effective feeling light area and the center superposition of target illuminated area, computer 7 gathers the light spot image in CCD camera 6 and it is carried out data processing, obtains tested micro-reflector M with accurate translation stage_kThe centroid position of corresponding flare in target face. Each micro-reflector unit is repeated above step, until obtaining the centroid position observed value matrix D of flare corresponding when each micro-reflector unit is in static state⁰(k)��

Fig. 4 is the schema of grid search ant group algorithm of the present invention, and wherein, the step of described grid search ant group algorithm is as follows:

Step S11: divide Searching Resolution Space region. By target illuminated area grid division, corresponding to a state residing for an ant on the net point in space, if every ant in ant group algorithm corresponds to the flare of each micro-reflector unit on target illuminated area, along with mirror unit two dimension rotates, ant is moved between each space networks lattice point, target function value according to each net point, leaves different information concentration, the travel direction of ant during to affect next iteration. In the process, every corresponding unknown variable of ant k is micro-reflector unit M_kTwo-dimentional angle of rotation, each solution all containing on x direction, angle of rotation Two Variables on y direction, according to the light distribution situation of light illumination mode required in target face, it is determined that the initial rotation angle degree scope of micro reflector array is as follows:

0 < | ��_x ¹(k) | < ��_upper

0 < | ��_y ¹(k) | < ��_upper

${\mathrm{\&theta;}}_{\mathrm{upper}}=\frac{1}{2}\arctan \left(\frac{{\mathrm{\&delta;}}_c{\mathrm{\&Phi;}}_{\mathrm{obj}}}{2f_{\mathrm{collimator}}}\right)---(1-3)$

Wherein, ��_x ¹(k), ��_y ¹(k) be respectively micro-reflector unit in the x direction with the initial rotation angle degree on y direction, ��_outerFor intending the maximum outer coherence factor of the target light illumination mode of generation, ��_objFor the area of target face illuminated area.

By each ��_x ¹(k), ��_y ¹K () is divided into N decile, then total 2N the node of solution space, if the sequence number in certain solution space corresponding to micro-reflector unit is (p, q), when algorithm starts, each micro-reflector finite element rotation angle random is dispersed in solution space etc. somewhere, subregion, obtain angle of rotation as follows:

${{\mathrm{\&theta;}}_x}^1\left(k\right)=0+\frac{{\mathrm{\&theta;}}_{\mathrm{upper}}}{N}\&times;(p+\mathrm{rand})$

${{\mathrm{\&theta;}}_y}^1\left(k\right)=0+\frac{{\mathrm{\&theta;}}_{\mathrm{upper}}}{N}\&times;(q+\mathrm{rand}),(p,q=\mathrm{1,2},...,N)---(1-4)$

Wherein, rand is the randomized number between [0,1].

Step S12: initialize population and individual information element distribution. Iteration is initial, and setting ant sum is M, and in optimization, maximum cycle is NC_max, Information Meter volatilization parameter �� (0 < �� < 1). If after each ant k loads the initial rotation angle degree as shown in step S11, it is contemplated that to having measured the position measurements matrix D obtaining flare corresponding when each micro-reflector unit is in static state by CCD camera 6⁰(k), whereinWhen then iteration is initial, the position correction of each ant k on target illuminated area is:

${D_x}^1\left(k\right)=f_{\mathrm{collimator}}\&times;\tan \left(2{{\mathrm{\&theta;}}_x}^1\left(k\right)\right)+D_x^0\left(k\right)$

${D_y}^1\left(k\right)=f_{\mathrm{collimator}}\&times;\tan \left(2{{\mathrm{\&theta;}}_y}^1\left(k\right)\right)+D_y^0\left(k\right)---(1-5)$

The difference of the target light distribution that the light distribution in target face produces with plan after making objective function G equal the loaded and optimized gained angle of rotation matrix of micro reflector array, and optimize when working as | G |�ܦ� (�� is algorithm convergence accuracy value, and 0 < �� < 1) and terminate. The evaluation functional value of definition ant k is the objective function G of u point on the target illuminated area residing for it_uWith the objective function G of v point residing for ant l in its field_vSum, that is:

��G_uv=G_u+G_v(1-6)

Due in the process of the loaded and optimized angle of rotation obtained of micro reflector array, angle of rotation is more big, in target face, mobile spot distance is more big, the error of the facula position produced is more big, therefore the search direction setting ant not only with its mobile route (u, v) it is relevant that the pheromone concentration on evaluates function, distance dependent also and between ant, ant selects the optimization path nearest with it as far as possible, to reduce micro-reflector unit deviation in moving process of flare when carrying out two dimension angular and rotate. Kth ant is made to adopt the mode of roulette mobile to position v point by position u point according to state transition probability formula below:

${P_{\mathrm{uv}}}^k=\frac{{\mathrm{\&tau;}}_{\mathrm{uv}}/d_{\mathrm{uv}}}{\underset{v\&Element;{\mathrm{allowed}}_k}{\mathrm{\&Sigma;}}{\mathrm{\&tau;}}_{\mathrm{uv}}/d_{\mathrm{uv}}},u\&NotEqual;v---(1-7)$

Wherein, d_uvFor distance between position u point and position v point in target face, ��_uvFor the pheromone intensity of ant k on mobile route, allowed_kThe space networks lattice point set of paths that next step allows away for ant k, wherein, the initial pheromone concentration of every ant k position u point can be determined according to objective function, that is:

��_u ⁰=e^-G(u)(1-8)

Step S13: ant packet, determine elitist ants and general ant: select satisfied | G |�ܦ� elitist ants and by its currently on target illuminated area present position add to taboo list in, each ant is moved by formula (1-7) described optimizing probability, and adopt unidirectional search mechanisms, ant is moved to the position that objective function is little from the position that objective function is big, ensures that multiple target reaches its optimum value. When a loop ends, the pheromone intensity on ant k institute mobile route is pressed following formula and is upgraded:

${{\mathrm{\&tau;}}_{\mathrm{uv}}}^{\mathrm{new}}=\mathrm{\&rho;}\&times;{\mathrm{\&tau;}}_{\mathrm{uv}}^{\mathrm{old}}+e^{-G_{\mathrm{uv}}}---(1-9)$

The stagnation easily occurred in optimizing process due to ant group algorithm and diffusion problem are not easy to ignore, and in order to avoid algorithm Premature Convergence in non-overall situation optimum solution, possible pheromone strength range on each bar optimizing path are limited in [��_min,��_max] in, wherein, ��_minAlgorithm can be effectively avoided to stagnate, ��_maxThe pheromone concentration on certain paths can be avoided much larger than other paths, further, in order to avoid making multiple ant all focus on same paths, limit algorithm spreads, retaining optimum path after each loop ends, namely in a circulation, only the ant of shortest path just has weight update ��_uv ^new, on the basis of above-mentioned (1-9) formula, add following formula and carry out threshold decision selection, that is:

${{\mathrm{\&tau;}}_{\mathrm{uv}}}^{\mathrm{new}}=\left\{\begin{array}{c}{\mathrm{\&tau;}}_{\min },{{\mathrm{\&tau;}}_{\mathrm{uv}}}^{\mathrm{new}}\&le;{\mathrm{\&tau;}}_{\min }\\ {{\mathrm{\&tau;}}_{\mathrm{uv}}}^{\mathrm{new}},{\mathrm{\&tau;}}_{\min }\&le;{{\mathrm{\&tau;}}_{\mathrm{uv}}}^{\mathrm{new}}\&le;{\mathrm{\&tau;}}_{\max }\\ {\mathrm{\&tau;}}_{\max },{{\mathrm{\&tau;}}_{\mathrm{uv}}}^{\mathrm{new}}\&GreaterEqual;{\mathrm{\&tau;}}_{\max }\end{array}\right.---(1-10)$

Step S14: if cycle index NC > NC_max, then continue to optimize, otherwise go to step S15.Often taking turns after iteration terminates, recalculate the target function value G of ant group, the position of the elitist ants obtained is added in taboo list, if the target function value of every ant is less than the convergence precision value �� defined, then go to step S15, otherwise go to step S13, start next and take turns iterative search;

Step S15: terminate optimizing, obtains the location matrix D of each ant k current on target illuminated area^optimal(k)(D^optimal(k)=(D_x ^optimal(k), D_y ^optimal(k))), in conjunction with ant k on target illuminated area starting position two dimension matrix D¹K (), obtains the two-dimentional matrix h relative to starting position miles of relative movement^optimalK (), is defined as the best mobile location matrix after optimizing, and calculate corresponding two dimension angle of rotation matrix �� according to formula (1-11)^optimal(k), wherein,

$\begin{array}{cc}{h_x}^{\mathrm{optimal}}\left(k\right)=D_x^{\mathrm{optimal}}\left(k\right)-D_x^1\left(k\right)& {{\mathrm{\&theta;}}_x}^{\mathrm{optimal}}\left(k\right)=\frac{1}{2}\arctan \left(\frac{{h_x}^{\mathrm{optimal}}\left(k\right)}{f_{\mathrm{collimator}}}\right)\end{array}$

$\begin{array}{cc}{h_y}^{\mathrm{optimal}}\left(k\right)=D_y^{\mathrm{optimal}}\left(k\right)-D_y^1\left(k\right)& {{\mathrm{\&theta;}}_y}^{\mathrm{optimal}}\left(k\right)=\frac{1}{2}\arctan \left(\frac{{h_y}^{\mathrm{optimal}}\left(k\right)}{f_{\mathrm{collimator}}}\right)\end{array}---(1-11)$

Algorithm terminates.

In the present embodiment, as maximum cycle NC_maxGetting 2000, when convergence precision value �� is 0.05, algorithm is about 1000 reach convergence in cycle index.

Owing to above-mentioned ant colony optimization algorithm solves in process, do not consider the aberration due to fourier lense group in actual optical system, the position deviation of micro-reflector unit and the change of reflected spot size in two dimension rotary course thereof, now on target illuminated area, the miles of relative movement of flare does not meet the relation shown in formula (1-11), have impact on the positional accuracy of emergent light spot, further, in generation traditional lighting and off-axis illumination occasion, to ask the beam uniformity in target face more than 95%, then the hot spot deviation in above-mentioned situation must be corrected, namely compensated optimizing the micro-mirror angle of rotation matrix obtained.

But the position of each micro-reflector flare and size all with the position at micro-reflector unit place, it is angle of rotation in the two-dimensional direction, the centre deviation of single micro-compound eye array, the factors such as the aberration of fourier lense group are relevant, cannot express by concrete analytical function expression formula, in order to reduce computation complexity, the present invention adopts and first filters out the hot spot not meeting position offset requirement, then the mode these hot spots measured to the position of the angle of rotation of micro-reflector unit and corresponding measurement hot spot thereof and dimensional data by certain algorithm (such as method of least squares, genetic algorithm etc.) data fitting, draw the functional relation that the position of each flare and size change with angle of rotation ��. following step is adopted to correct gained angle of rotation after ant colony optimization algorithm:

Step S21: the micro reflector array angle of rotation obtained after grid ant group algorithm optimization be loaded into one by one on each micro-reflector unit, namely provides micro reflector array angle of rotation matrix ��, except tested micro-reflector M in this angle of rotation matrix_kTwo-dimentional angle of rotation be ��^optimalK () outward, other mirror units are controlled by control unit 8, taking array center position as boundary line, the mirror unit comprising more than boundary line upward deflects �� angle, and following mirror unit deflects down �� angle, and CCD camera 6 is placed on target illuminated area D^optimalK () position is measured, now flare only to be measured is positioned on the reception face of CCD camera 6, and wherein, matrix �� is:

Step S22: mobile accurate translation stage, is placed on target illuminated area D by CCD camera^optimalK () position accepts hot spot, to each micro-reflector unit, calculate its physical location measuring flare barycenter and the difference d of the upper pre-position of target illumination by step S21_diff, select d_diffIt is greater than the speculum set M to be corrected of maximum facula deviation amount �� L_s(s < k, and M_s��M_k), from micro-reflector angle of rotation 0����, get n point (n >=20), measure M respectively_sIn the flare location parameter of each micro-reflector unit at this n some place, it is designated as P ' (��, s), adopt method of least squares that the centroid position data of the flare of actual measurement under different rotary angle, �� is carried out the data fitting at discrete point place, it is used nonlinear function f_s(��) function representation, wherein:

$f_s\left(\mathrm{\&theta;}\right)=\underset{j=0}{\overset{n}{\mathrm{\&Sigma;}}}{a}_j{f}_j\left(\mathrm{\&theta;}\right)$

|f_s(��)-P��(��,s)|�ܦ�₁(1-13)

Wherein, a_jFor the coefficient of polynomial fitting each, ��₁For data fitting error, it is relevant with maximum facula deviation amount �� L.S function expression is obtained, the flare M to be corrected grid ant group algorithm optimization drawn from formula (1-13)_sLocation parameter substitute into f_s(��), in expression formula, obtain:

$D^{\mathrm{optimal}}\left(s\right)=\underset{j=0}{\overset{n}{\mathrm{\&Sigma;}}}{a}_j{f}_j\left(\mathrm{\&theta;}\right)---(1-14)$

Solve �� (s) from upper formula, replaced ��^optimalThe value of identical position in (k).

Step S23: repeating step S22, until the position difference of each mirror unit flare is all less than target illumination field allows maximum facula deviation amount �� L, export the micro-reflector two dimension angle of rotation matrix �� finally meeting actual optical system and requiring^finnal(k)��

Fig. 5 is micro reflector array flare correction front and back schematic diagram on target illuminated area, in the present embodiment, flare size mean value is 2mm, target illumination field allows maximum facula deviation amount �� L < 0.2mm, arrange at center one in the Y direction for micro reflector array, it comprises 100 micro mirror unit, and Fig. 5 (a) is the difference curve of light distribution and desirable light distribution before correction; Relatively big near the micro-reflector unit deviation of position, edge before correction as seen from the figure, the flare corresponding to these position micro-reflectors is screened, it is corrected. The difference curve of Fig. 5 (b) light distribution and desirable light distribution after the correction of row mirror unit for this reason. After correction, all hot spot deviation value is all less than maximum facula deviation amount as seen from the figure.

Fig. 6 is the embodiment of several illumination iris distribution that application the present invention produces, Fig. 6 (a), 6 (b) be respectively in illumination analysis software Lighttools simulation free illumination iris distribution plan, Fig. 6 (c), 6 (d) corresponds respectively to Fig. 6 (a), with Fig. 6 (b), when quantizing for target illuminated area being got the sampling number of 150 �� 150, X-direction crosses the difference of the distribution of the illumination intensity in its center cross-sectional with the desirable light distribution intending producing, get its each light intensity difference quantized in grid and do normalization method RMS graphic representation, as can be seen from the figure, what application the inventive method was formed free illumination iris is less with desirable illumination iris difference, RMS value is no more than 0.1%.

Foregoing description is only the description to the better embodiment of the present invention, and not to any restriction of the present invention, any change that the those of ordinary skill in field of the present invention does according to above-mentioned disclosure, discloses, all belong to the scope of claim book.

Claims (6)

1. one kind realizes the method for variable free illumination iris based on micro reflector array, it is characterised in that: described method comprises the steps:

Step (1), the location parameter of flare on illumination target face when measure each micro-reflector unit by CCD camera static;

When in described step (1), each micro-reflector unit is in static state, the detailed process measuring its flare location parameter on illumination target face by CCD camera is as follows:

If micro reflector array unit sum is M, provide micro reflector array angle of rotation matrix ��, except micro-reflector unit M to be measured in this angle of rotation matrix_k(k is mirror unit subscript sequence number, and k=1,2, ..., M) angle of rotation be outside 0, controlling other mirror units taking array center by control unit is boundary line, the mirror unit comprising more than boundary line upward deflects �� angle, and following mirror unit deflects down �� angle, and namely �� is:

Wherein �� is the maximum angle of rotation of each mirror unit, target illuminated area is formed one long be the blank region of H, and:

H=2 �� f_collimator��tan(2��)(1-2)

Wherein, f_collimatorFor collimating the focal length of condensor group;

The center position that CCD camera is placed on target illuminated area accepts the emergent light spot of Quasi-straight light-focusing mirror group, it is also carried out data processing by light spot image in computer acquisition CCD camera, each micro-reflector unit is repeated above step, until obtaining the centroid position observed value matrix D of flare corresponding when each micro-reflector unit is in static state⁰(k);

Step (2), the location parameter that step (1) is measured gained flare are brought in grid search ant group algorithm, and optimization produces the micro reflector array angle of rotation matrix of required light distribution;

Step (3), utilize the angle of rotation matrix that step (2) obtains, generate the control signal of each micro-reflector driving mechanism and control micro-reflector and carry out angle rotation, measure now light distribution on illumination target face, using the facula position difference of itself and required light distribution as evaluation index, system is carried out feedback control, until facula position difference meets design requirement.

2. the method realizing variable free illumination iris based on micro reflector array according to claim 1, it is characterized in that: the method is applicable to utilize micro reflector array to realize the system of any illumination profile, wherein each micro-reflector unit can two dimension continuously rotate, and micro reflector array is imaged onto on target illuminated area by a collimation condensor group, micro reflector array position becomes Fourier's relation with target illuminated area.

3. the method realizing variable free illumination iris based on micro reflector array according to claim 1, it is characterized in that: by closed loop feedback system, micro reflector array was demarcated before carrying out the measurement of flare position, make each mirror unit of micro reflector array in above-mentioned steps (1), (2), all can normal operation in (3) three processes.

4. the method realizing variable free illumination iris based on micro reflector array according to claim 1, it is characterized in that: mirror unit deflection angle more than described micro reflector array center can also get other values being less than ��, mirror unit deflection angle below center can also get other values being greater than-�� so that only has flare to be measured on effective reception face of CCD.

5. the method realizing variable free illumination iris based on micro reflector array according to claim 1, it is characterised in that: in described step (2), the step of grid search ant group algorithm is:

Step S11: divide Searching Resolution Space region;

Using each micro-reflector unit two dimension angle of rotation as solution space set, by illumination target face grid division, according to the target light intensity distribution situation of required generation, it is determined that the two-dimentional initial rotation angle degree �� of each micro-reflector unit¹(k)=(��_x ¹(k), ��_y ¹(k)) scope as follows:

0<|��_x ¹(k)|<��_upper

0<|��_y ¹(k)|<��_upper(1-3)

Wherein, ��_x ¹(k), ��_y ¹(k) be respectively micro-reflector unit along the x-axis direction with the initial rotation angle degree in y-axis direction,��_cFor the maximum outer coherence factor produced under required target light illumination mode, ��_objFor the area in illumination target face;

By each ��_x ¹(k), ��_y ¹K () is divided into N decile, if the sequence number in certain solution space corresponding to micro-reflector unit is (p, q), when algorithm starts, each micro-reflector finite element rotation angle random is dispersed in solution space etc. somewhere, subregion, obtain initial rotation angle degree as follows:

Wherein, rand is the randomized number between [0,1];

Step S12: initialize population and individual information element distribution;

Iteration is initial, and setting ant sum is M, and in optimization, maximum cycle is NC_maxInformation Meter volatilization parameter is �� (0 < �� < 1), convergence precision is �� (0 < �� < 1), the difference of the target light distribution that target illumination field light distribution produces with plan after making objective function G equal the loaded and optimized gained angle of rotation matrix of micro reflector array, by the position calculation matrix of each flare of gained in step (1)When correction iteration is initial, each ant k in the position of target illuminated area is:

The evaluation function of definition ant k is the objective function G of u point on the target illuminated area residing for it_uWith the objective function G of v point residing for ant l in its field_vSum, that is:

��G_uv=G_u+G_v(1-6)

By position u point to the mobile state transition probability formula of position v point it is:

Wherein, d_uvFor distance between u point and v point in target face, ��_uvFor the pheromone intensity of ant k on mobile route, allowed_kThe space networks lattice point set of paths that next step allows away for ant k, the initial pheromone concentration of every ant k position u point is:

��_u ⁰=e^-G(u)(1-8)

Step S13: elitist ants and general ant are determined in grouping;

After loading the micro reflector array initial rotation angle degree matrix Ru shown in (1-4) formula, select the elitist ants of satisfied | G |�ܦ� and its present position on target illuminated area is added in taboo list, by the mobile each ant of formula (1-7), unidirectional search mechanisms is adopted to be moved from the position that objective function is big to the position that objective function is little by ant, when a loop ends, the pheromone intensity on ant k institute mobile route is pressed following formula and is upgraded:

Possible pheromone strength range on each bar optimizing path is limited in [��_min,��_max]:

Step S14: if NC > NC_max, then continue to optimize, otherwise go to step S15, often taking turns after iteration terminates, upgrade target function value G and the taboo list of ant group, if the objective function of every ant is less than the convergence precision value �� of definition, then go to step S15, otherwise go to step S13, start next and take turns iterative search;

Step S15: terminate optimizing, according to the two-dimensional position matrix D of current ant k on target illuminated area^optimalK () calculates best miles of relative movement matrix h^optimal(k) and corresponding two-dimentional angle of rotation matrix ��^optimal(k), wherein,

6. the method for variable free illumination iris is realized according to claim 1 based on micro reflector array, it is characterised in that: angle of rotation matrix �� after correction grid search ant group algorithm optimization in described step (3)^optimalK the step of () is:

Step S21: provide micro reflector array angle of rotation matrix ��, except tested micro-reflector M in this angle matrix_kTwo-dimentional angle of rotation be ��^optimalK () outward, other mirror units are by control unit control, and taking array center position as boundary line, the mirror unit comprising more than boundary line upward deflects �� angle, and following mirror unit deflects down �� angle, and matrix �� is:

Step S22: CCD camera is placed on target illuminated area Doptiaml (k) position, measures each micro-reflector M_kThe physical location of flare barycenter, calculates itself and the upper predetermined position D of target illumination^optimalThe position difference d at (k) place_diff, select d_diffIt is greater than the speculum set M to be corrected that target illumination field allows maximum facula deviation amount �� L_s(s < k, and M_s��M_k), in these speculums, its angle of rotation �� is got from 0���� n point (n >=20), measure micro-reflector unit M respectively_sIn the facula position parameter at this n some place, it is designated as P ' (��, s), is used nonlinear function f_s(��) it is carried out data fitting, represents and be:

|f_s(��)-P��(��,s)|�ܦ�₁(1-13)

Wherein, a_jFor the coefficient of polynomial fitting, ��₁For data fitting low precision, obtain S function expression from formula (1-13), the flare M to be corrected grid ant group algorithm optimization drawn_sPosition D^optimalS () substitutes into f_s(��) in expression formula, that is:

From upper formula, solve �� (s), replaced ��^optimalThe value of identical position in (k);

Step S23: repeating step S22, until the position difference of each mirror unit flare is all less than target illumination field allows maximum facula deviation amount, export the last micro-reflector two dimension angle of rotation matrix �� meeting actual optical system requirement^finnal(k)��