Method and apparatus for correcting errors on a wafer processed by a photolithographic mask

Abstract

The invention relates to a method for correcting at least one error on wafers processed by at least one photolithographic mask, the method comprises: (a) measuring the at least one error on a wafer at a wafer processing site, and (b) modifying the at least one photolithographic mask by introducing at least one arrangement of local persistent modifications in the at least one photolithographic mask.

Claims

1. A method for correcting at least one error on wafers processed by at least one template for a nanoimprint lithography, the method comprising: a. measuring the at least one overlay error on a wafer at a wafer processing site; b. modifying a global overlay error of the at least one overlay error of the at least one template for the nanoimprint lithography by a linear transformation; and c. modifying a local overlay error of the at least one overlay error of the at least one template for the nanoimprint lithography by introducing at least one arrangement of local persistent modifications in the at least one template for the nanoimprint lithography.

2. The method according to claim 1, wherein the at least one overlay error comprises at least one error of at least one of at least two templates for the nanoimprint lithography used in a multiple patterning lithography process.

3. The method according to claim 1, wherein measuring of the at least one error comprises measuring of the at least one error in the active area of a chip (in-die).

4. The method according to claim 3, wherein in-die measuring of the at least one overlay error comprises measuring of at least one of: a shift of at least one two-dimensional structure on a wafer, a shift of at least one three-dimensional structure on a wafer, an ellipticity of at least one two-dimensional structure with an imaging-based or a model-based metrology method, or an ellipticity of at least one three-dimensional structure with an imaging based or a model-based metrology method.

5. The method according to claim 1, wherein the overlay error comprises at least one local overlay error of at least one first template for the nanoimprint lithography and at least one second template for the nanoimprint lithography, the method further comprising the step of: correcting the at least one local overlay error by introducing at least one arrangement of local persistent modifications in the at least one first template and/or the at least one second template so that the at least one overlay error is minimized.

6. The method according to claim 5, wherein correcting the at least one local overlay error comprises introducing at least one first arrangement of local persistent modifications in the at least one first template and/or introducing at least one second arrangement of local persistent modifications in the at least one second template.

7. The method according to claim 5, wherein the at least one overlay error comprises at least one critical dimension uniformity error of the first template and/or at least one pattern placement error of the second template in a multiple patterning lithography process.

8. The method according to claim 1, wherein measuring the at least one error on the wafer comprises: a. generating a test mask having a test pattern; b. printing and etching the test pattern of the test mask on the wafer; c. printing and etching a template pattern on the test pattern of the wafer; and d. determining the at least one error as a difference of at least one pattern element of the template for the nanoimprint lithography and the at least one respective test pattern element of the test mask.

9. The method according to claim 8, wherein printing and etching of the test mask and printing and etching of the template pattern comprises a single patterning or a multiple patterning lithography process.

10. The method according to claim 8, further comprising the step of determining an overlay error from a shift of a plurality of test pattern elements of the test mask relative to pattern elements of the template.

11. The method according to claim 1, further comprising the step of introducing the at least one arrangement of local persistent modifications in a sacrificial layer on the wafer used in a self-aligned double patterning process.

12. An apparatus for correcting at least one overlay error on wafers processed by at least one template for a nanoimprint lithography, comprising: a. at least one metrology system located in a wafer processing site and/or in a mask shop and adapted to measure the at least one overlay error on a wafer; b. at least one computing means adapted to calculate parameters for at least two error correcting means based on the at least one overlay error; c. at least one first error correcting means adapted for modifying a local overlay error of the at least one overlay error by introducing at least one arrangement of local persistent modifications in the template for the nanoimprint lithography by applying ultra-short light pulses; and d. at least one second error correcting means adapted for modifying a global overlay error of the at least one overlay error by performing a linear transformation.

13. The apparatus according to claim 12, wherein the apparatus is adapted to correct at least one error on wafers processed by at least one template for the nanoimprint lithography by: a. measuring the at least one error on the wafer at a wafer processing site; b. modifying a global overlay error of the at least one overlay error of the at least one template by a linear transformation; and c. modifying a local overlay error of the at least one overlay error or the at least one template for the nanoimprint lithography by introducing at least one arrangement of local persistent modifications in the at least one template for the nanoimprint lithography.

14. The apparatus according to claim 12, wherein the at least one metrology system comprises an ultrahigh-precision stage, at least one laser source or at least one other light source, at least one charge-coupled device camera operating in the ultraviolet wavelength range, a scanning electron microscope or a scatterometer, and an image-based or a model-based metrology system.

15. The apparatus according to claim 12, wherein the at least one overlay error comprises at least one error of at least one of at least two templates for the nanoimprint lithography used in a multiple patterning lithography process.

16. The apparatus according to claim 12, wherein the at least one overlay error comprises at least one error in the active area of a chip (in-die).

17. The apparatus according to claim 16, wherein the at least one metrology system is adapted to measure at least one of a shift of at least one two-dimensional structure on a wafer, a shift of at least one three-dimensional structure on a wafer, an ellipticity of at least one two-dimensional structure, or an ellipticity of at least one three-dimensional structure, with at least one of an imaging-based or a model-based metrology method.

18. The apparatus according to claim 12, wherein the overlay error comprises at least one local overlay error of at least one first template for nanoimprint lithography and at least one second template for nanoimprint lithography, and the apparatus is adapted to correct the at least one local overlay error by introducing at least one arrangement of local persistent modifications in the at least one first template and/or in the at least second template so that the at least one overlay error is minimized.

19. The apparatus according to claim 18, wherein the apparatus is adapted to correct the at least one local overlay error by introducing at least one first arrangement of local persistent modifications in the at least one first template and/or introducing at least one second arrangement of local persistent modifications in the at least one second template.

20. The apparatus according to claim 18, wherein the at least one overlay error comprises at least one critical dimension uniformity error of the first template and/or at least one pattern placement error of the second template in a multiple patterning lithography process.

21. The apparatus according to claim 12, wherein the at least one metrology system is adapted to measure the at least one overlay error on the wafer by determining the at least one overlay error as a difference of at least one pattern element of the template for the nanoimprint lithography and at least one respective test pattern element of a test mask.

22. The apparatus according to claim 21, wherein the at least one metrology system is adapted to determine an overlay error from a shift of a plurality of test pattern elements of the test mask relative to pattern elements of the template.

23. The apparatus according to claim 12, wherein at least one first error correcting means is adapted for introducing the at least one arrangement of local persistent modifications in a sacrificial layer on the wafer used in a self-aligned double patterning process.

Description

DESCRIPTION OF DRAWINGS

(1) In order to better understand the present invention and to appreciate its practical applications, the following Figures are provided and referenced hereafter. It should be noted that the Figures are given as examples only and in no way limit the scope of the invention.

(2) FIG. 1 schematically represents a flow chart of a fabrication process of a set of photolithographic masks according to the prior art;

(3) FIG. 2 schematically shows a flow chart of a use case of a set of photolithographic masks at a wafer processing site according to the prior art;

(4) FIG. 3 schematically represents a block diagram of some of the major components of an apparatus used for measuring overlay errors on a wafer;

(5) FIG. 4 schematically shows a block diagram of an apparatus used for correcting overlay errors in the substrate of a photolithographic mask;

(6) FIG. 5 schematically represents a flow chart of a fabrication process of a set of photolithographic masks according to an embodiment of the inventive method;

(7) FIG. 6 schematically shows a flow chart of a use case of a set of photolithographic masks at a wafer processing site according to an embodiment of the inventive method;

(8) FIG. 7 schematically represents a displacement vector map measured at the scribe lines of dies;

(9) FIG. 8 schematically illustrates a displacement vector map measured at nodes of a regular grid, i.e. at the scribe lines of dies and in-die;

(10) FIG. 9 schematically presents process steps of a litho-etch-litho-etch (LELE) process;

(11) FIG. 10 schematically shows process steps of a litho-freeze-litho-etch (LFLE) process;

(12) FIG. 11 schematically depicts process steps of a self-aligned double patterning (SADP) process;

(13) FIG. 12 schematically illustrates critical dimension uniformity and overlay errors of a cut-out of a one-dimensional pattern generated with a LELE or LFLE process;

(14) FIG. 13 schematically shows a cut-out of a two-dimensional pattern generated with a LELE or LFLE process;

(15) FIG. 14 repeats FIG. 13, wherein the horizontal pattern has an overlay error;

(16) FIG. 15 illustrates FIG. 13, wherein the vertical pattern has a critical dimension uniformity error;

(17) FIG. 16 represents FIG. 13, wherein the horizontal pattern has an overlay error and the vertical pattern has a critical dimension uniformity error;

(18) FIG. 17 schematically shows enlarged the contact holes of FIG. 15 which are elliptically deformed due to the critical uniformity error;

(19) FIG. 18 schematically presents process steps and their variation of a self-aligned double patterning (SADP) process;

(20) FIG. 19 schematically depicts a test pattern of a test mask;

(21) FIG. 20A schematically shows a photo resist on a wafer after to the execution of second litho-etch steps in a LELE process having essentially a plane surface due to an introduction of a hard mask film with high optical contrast to the surface and photoresist layer to enable high resolution metrology in high accuracy, sensitivity and precision;

(22) FIG. 20B schematically shows a photo resist on a wafer after to the execution of second litho-etch steps in a LELE process having a pronounced surface topography;

(23) FIG. 21 schematically presents a LELE process of a test mask pattern and of a mask pattern having vertical line space features;

(24) FIG. 22 schematically depicts a LELE process of a test mask pattern and of contact holes of a mask used in the DPL process;

(25) FIG. 23 schematically shows a LELE process of a test mask pattern and of a mask pattern having horizontal line space features;

(26) FIG. 24 schematically illustrates a superposition of the test pattern of the test mask of FIG. 19 and of the mask patterns of FIGS. 21 to 23;

(27) FIG. 25 schematically depicts the superposition of the test pattern of the test mask of FIG. 19 and of the mask patterns of FIGS. 21 to 23, wherein the errors of the mask patterns of FIGS. 21 to 23 are corrected;

(28) FIG. 26 schematically presents in-field targets for the determination of an averaged in-field overlay error;

(29) FIG. 27 schematically shows a wafer covered with fields, wherein each field has the overlay targets of FIG. 26;

(30) FIG. 28 schematically reproduces an overlay error map across a wafer;

(31) FIG. 29 schematically illustrates an in-field error map resulting from an averaging of the in-field error distributions across the wafer; and

(32) FIG. 30 schematically depicts the averaged in-field error map of FIG. 29 after the systematic errors are corrected by introducing respective arrangements of local persistent modifications in the substrate of the photolithographic mask used for the printing of the wafer of FIG. 28.

DETAILED DESCRIPTION

(33) In the following, the present invention will be more fully described hereinafter with reference to the accompanying Figures, in which exemplary embodiments of the invention are illustrated. However, the present invention may be embodied in different forms and should not be construed as limited to the embodiments set forth herein. Rather, these embodiments are provided so that this disclosure will be thorough and will convey the scope of the invention to persons skilled in the art.

(34) This first section describes an embodiment of the inventive method. To illustrate the inventive principle, differences are highlighted in the fabrication of a set of photolithographic masks with respect to the fabrication according the prior art. Further, these differences are also exemplified for a use case of a mask set. The inventive method is explained for the utilization of transmissive photolithographic masks. However, the person skilled in the art will appreciate that this is just an example and that the above defined method can also be applied to reflective photolithographic masks. Moreover, the inventive method is also well suited to correct overlay errors of templates for the nanoimprint lithography at a wafer processing site or at a wafer fabrication site.

(35) In the following, the term integrated circuit (IC) is used for all devices fabricated on semiconducting wafers as for example memory or logic components, MEMS (micro-electromechanical systems) including sensors, detectors and displays and PICs (photonic integrated circuits) including lasers and photodiodes.

(36) The inventive method corrects errors detected on a wafer by applying femtosecond light pulses of a laser system to a photolithographic mask with which a wafer was illuminated. For the correction feature of the present invention, the specification refers to the US provisional with the No. 61/363,352. This document describes in detail how pattern placement errors can be corrected by the writing of a respective arrangement of pixels in the substrate of photolithographic masks. Some of the problems of photolithographic masks and of templates for the nanoimprint lithography are also briefly discussed in the above mentioned document.

(37) FIG. 1 outlines the fabrication process 100 for a mask set according to the prior art. The process begins at 105 with the writing of the pattern 115 of the first mask 110. Then, also at step 115, the excessive parts of the absorber layer are removed from the mask substrate, for example by etching. In the next step 120, the photolithographic mask is measured in order to determine the positioning errors of the mask pattern. At decision block 125, it is decided whether the positioning errors fulfil a predetermined specification. If this is correct, the pellicle is mounted on the mask in step 135. It is then decided at decision block 145 whether the mask is an overlay critical mask. The fabrication process is complete when mask m is not an overlay critical mask. If mask m is overlay critical (decision block 145), the mask is again measured at step 155 to check the effect of the pellicle mounting on the positioning errors. When it is determined at decision block 165 that mask m does not fulfil the specification, at step 160, the pellicle is removed and the pattern is rewritten in a new mask.

(38) When it is determined at decision block 125 that the positioning errors of mask m do not fulfil the predetermined specification, at step 130, the pattern placement errors are corrected by using a so called RegC (Registration Correction) process, which is described in detail in the U.S. 61/351,056 and U.S. 61/363,352. If the positioning errors cannot be reduced by a RegC process so that mask m meets the specification, the pattern is written on a new mask and the process proceeds to step 120 where the mask is measured.

(39) The process of FIG. 1 is repeated until all M masks of the mask set are fabricated (decision blocks 150 and 175). The complete mask set is then supplied to the wafer processing site at block 180.

(40) FIG. 2 briefly illustrates some steps of a mask illumination process for the fabrication of ICs at a wafer processing site. The process 200 begins at 205 with the illumination or the exposure of a wafer (step 215) with the first mask (step 210) of the mask set. The photo resist on the wafer is developed and the wafer is processed, for example by performing of an etching process. Then a second (generally m.sup.th) layer of photo resist is arranged on the wafer (not illustrated in FIG. 2). At step 220, the wafer is illuminated with the second, or generally with the (m+1).sup.th mask. The overlay errors between the first and the second mask, or in general between the m.sup.th and the (m+1).sup.th mask on the wafer are measured at step 230. If it is decided at decision block 240 that the overlay error is below the predetermined overlay budget, it is at decision block 265 checked whether the IC to be fabricated will operate according to its specification. If this is correct, the process proceeds via decision block 295 and step 225 to block 220 where the wafer is illuminated with the next (third or (m+2).sup.th) mask. When it is determined at decision block that m reaches M which is the number of masks in the respective mask set, the process ends at block 270.

(41) If it is decided at decision block 265 that the IC will not operate according to its specification, the root cause of the problem is analyzed at block 280 and the positioning specification for mask m+1 is tightened. In step 290, a new mask m+1 is then written at the mask processing site. The process ends at block 270. Then the new mask m+1 is supplied from the mask fabrication site to the wafer processing site, the process begins again at block 205 of FIG. 2.

(42) When it is decided at decision block 240 that the overlay error does not meet the specification, the projection device of the illumination system is readjusted at block 245. Then, at step 260, the overlay error measurement is repeated. If it is determined at decision block 275 that the overlay error does still not fulfil the predetermined error budget, the process proceeds to step 280 and the cause of the problem is analyzed. In case the overlay error meets the specification, it is at decision block 285 determined whether the IC to be fabricated will operate according to its specification. If this is true, the process proceeds via decision block 295 across step 225 to block 220 where the wafer is illuminated with the next (third or (m+2).sup.th) mask of the mask set. Alternatively, when the illuminated mask is the last mask of the mask set (m=M), the process ends at block 270.

(43) FIG. 3 shows a functional sketch of a registration metrology tool 300 with which pattern placement and overlay errors can be measured. A photolithographic mask 310 is supported by the high precision stage 320. The stage 320 is actively controlled in all six degrees of freedom and is the only moving part in the metrology system. As a light source an excimer laser 330 is used emitting light in the DUV (deep ultraviolet) wavelength range, at approximately 193 nm. This means that the inspection and the illumination of the photolithographic mask 310 occur at the same wavelength, as most masks are presently illuminated with a 193 nm light source. Hence, the registration and/or overlay metrology system 300 takes the effect of material properties properly into account.

(44) The imaging objective 340 has a numerical aperture (NA) of 0.6, but can be extended to a higher NA in order to gain even more resolving power. The short wavelength of the laser system 330 significantly improves the resolution, and at the same time permits a moderate NA, which is beneficial for the CD (critical dimension) metrology and enables a pellicle compatible free working distance of about 7.5 mm. The imaging objective 340 is firmly fixed to the optical tower and is unmovable. Focusing of the laser beam onto the photolithographic mask 310 is done by a stage movement in z direction.

(45) A CCD (charge-coupled device) camera 350 is used as a detector device which measures the light reflected from the photolithographic mask 310. The CCD camera 350 sends its signal to the signal processing unit 355 which calculates an image of the signal detected by the CCD camera 350.

(46) A computer system 360 can display the image calculated by the signal processing unit 355 and may store the measured data. Further, the computer system 360 may contain algorithms, realized in hardware, software or both, which allow to extract control signals from the experimental data. The control signals may control the writing of an arrangement of pixels in the substrate of the photolithographic mask 310 by a second laser system in order to correct the pattern placement errors of photolithographic mask 310 (cf. FIG. 4 below). Further, the computer system 360 may control the laser source 330 and/or the high-precision stage 320 and/or the objective 340 and/or the CCD camera 350 and/or the AF system 370.

(47) The surface of the photolithographic mask 310 may be slightly tilted, and in addition the bending of the mask 310 under its own weight leads to a variation of the best focal position. Therefore, the registration metrology tool 300 has an autofocus (AF) system 370 based on a tilted grating (not shown in FIG. 3) which supports the measurement process. The tilted mirrors 390 and the partially transmitting mirrors 395 direct the laser beam into the imaging objective 340.

(48) Furthermore, the registration metrology tool 300 comprises an auxiliary optical system 380 for a coarse alignment of the pattern placement elements on the photolithographic mask 310.

(49) FIG. 4 depicts a schematic block diagram of an apparatus 400 which can be used to correct errors on wafers by modifying the substrate of a photolithographic mask. Further, the apparatus 400 is also able to correct errors of templates used in the nanoimprint lithography. The apparatus 400 comprises a chuck 420 which may be movable in three dimensions. The photolithographic mask 420 or a template for the nanoimprint technique may be fixed to the chuck 420 by using various techniques as for example clamping.

(50) The apparatus 400 includes a pulse laser source 430 which produces a beam or a light beam 435 of pulses or light pulses. The laser source 430 generates light pulses of variable duration. The adjustable range of several import parameters of the laser source 430 is summarized in the following table. Table 1 represents an overview of laser beam parameters of a frequency-doubled Nd-YAG laser system which can be used in an embodiment of the inventive method.

(51) TABLE-US-00001 TABLE 1 Numerical values of selected laser beam parameters for a Nd-YAG laser system Overview Numerical Parameter value Unit Pulse energy 0.05-5 ?J Pulse length 0.05-100 ps Repetition rate 1-10000 kHz Pulse density 1000-10000000 mm-2 NA (numerical 0.1-0.9 aperture) Wavelength 532 nm

(52) In an alternative embodiment of the laser system the light pulses may be generated by a Ti: Sapphire laser operating at a wavelength of 800 nm. However, the correction of pattern placement errors is not limited to these laser types, principally all laser types may be used having a photon energy which is smaller than the band gap to the substrate of the photolithographic mask 410 and which are able to generate pulses with durations in the femtosecond range.

(53) The steering mirror 490 directs the pulsed laser beam 435 into the focusing objective 440. The objective 440 focuses the pulsed laser beam 435 onto the photolithographic mask 410. The NA (numerical aperture) of the applied objectives depends on the predetermined spot size of the focal point and the position of the focal point within the photolithographic mask 410 or of the template. As indicated in table 1, the NA of the objective 440 may be up to 0.9 which results in a focal point spot diameter of essentially 1 ?m and a maximum intensity of essentially 10.sup.20 W/cm.sup.2.

(54) The apparatus 400 also includes a controller 480 and a computer system 460 which manage the translations of the two-axis positioning stage of the sample holder 420 in the plane perpendicular to the laser beam (x and y directions). The controller 480 and the computer system 460 also control the translation of the objective 440 perpendicular to the plane of the chuck 420 (z direction) via the one-axis positioning stage 450 to which the objective 440 is fixed. It should be noted that in other embodiments of the apparatus 400 the chuck 420 may be equipped with a three-axis positioning system in order to move the photolithographic mask 410 to the target location and the objective 440 may be fixed, or the chuck 420 may be fixed and the objective 440 may be moveable in three dimensions. Although not economical, it is also conceivable to equip both the objective 440 and the chuck 420 with three-axis positioning systems. It should be noted that manual positioning stages can also be used for the movement of the mask 410 to the target location of the pulsed laser beam 435 in x, y and z directions and/or the objective 440 may have manual positioning stages for a movement in three dimensions.

(55) The computer system 460 may be a microprocessor, a general purpose processor, a special purpose processor, a CPU (central processing unit), a GPU (graphic processing unit), or the like. It may be arranged in the controller 480, or may be a separate unit such as a PC (personal computer), a workstation, a mainframe, etc. The computer 460 may further comprise I/O (input/output) units like a keyboard, a touchpad, a mouse, a video/graphic display, a printer, etc. In addition, the computer system 460 may also comprise a volatile and/or a non-volatile memory. The computer system 460 may be realized in hardware, software, firmware, or any combination thereof. Moreover, the computer 460 may control the laser source 430 (not indicated in FIG. 4). The computer systems 360 of FIG. 3 and 460 of FIG. 4 may be connected to exchange data. Moreover, the computer systems 360 and 460 may be combined in a single computer system.

(56) Further, the apparatus 400 may also provide a viewing system including a CCD (charge-coupled device) camera 465 which receives light from an illumination source arranged in the chuck 420 via the dichroic mirror 445. The viewing system facilitates navigation of the photolithographic mask 410 to the target position. Further, the viewing system may also be used to observe the formation of a modified area on the substrate material of the mask 410 by the pulsed laser beam 435 of the light source 430.

(57) FIG. 5 schematically depicts an example of a fabrication process of photolithographic masks of a mask set according to the inventive method. As already briefly explained in the second section of this specification, a pattern of absorbing elements is written on an absorbing layer on the substrate of a photolithographic mask with a pattern generator. In a subsequent etching process, the absorbing pattern elements are formed from the absorbing material (box 515). A material often used for the absorbing layer on photolithographic masks is chromium. Tungsten can be used as another absorber material on the surface of mask substrates. The usage of the inventive method is not restricted to these materials; rather any absorber material can be used.

(58) The positions of the generated absorbing pattern elements are measured with the registration metrology system of FIG. 3 in order to determine whether the pattern writing process was successful, i.e. the pattern elements have their predetermined size and form and are at the desired positions (box 520). If the determined positioning errors exceed a predetermined level (decision box 525), the positions of the pattern elements are modified by writing of an arrangement of pixels into the substrate of the photolithographic mask using the laser source 430 of the apparatus 400 of FIG. 4 (box 530 of FIG. 5). The arrangement of pixels locally changes the density of the mask substrate and thus shifts the pattern elements on the mask surface to the predetermined positions. Then, it is measured whether the repair of the mask was successful (box 520). If the measured positioning error is now below the predetermined threshold (decision box 525), the pellicle is mounted on the surface of the photolithographic mask carrying the absorbing pattern elements in order to protect them from being damaged. When all masks of the mask set are processed according to this scheme, the fabricated mask set is ready for the delivery to the wafer processing site.

(59) As can be seen from FIG. 5, an embodiment of the mask fabrication process according to the inventive method largely avoids rewriting of photolithographic masks. If the positioning errors of a mask exceed the tolerable level, the respective mask is corrected by using the RegC process.

(60) The mask fabrication process of FIG. 5 abandons of the measurement of positioning errors for overlay-critical photolithographic masks after mounting of the pellicle. This is the decisive distinction between the mask fabrication processes according to FIGS. 1 and 5.

(61) FIG. 6 schematically illustrates a use case of a set of photolithographic masks at the wafer processing site according to an embodiment of the inventive method. The process begins by illuminating of a wafer with a first mask using the projection device of the photolithographic illumination system (box 615). The first mask may be the first mask of a mask set, or in the general case, it may be any mask, but the last one of the mask set. The photo resist is developed, and the wafer is processed to generate a first layer, or generally an m.sup.th layer, respectively, of an integrated circuit. Then a new photo resist layer is dispensed on the wafer (not shown in FIG. 6).

(62) In the next step (box 620), similar to the first photolithographic mask, a second photolithographic mask is aligned with respect to alignment marks on the wafer. Then the second mask, or generally the (m+1).sup.th mask, is illuminated similar to the first mask in order to transfer the structure elements for the second layer, or general (m+1).sup.th, layer of the integrated circuit from the photolithographic mask to the wafer. The photo resist is then developed.

(63) The photolithographic masks have overlay targets which are used to determine the overlay of the second mask with respect to the first photolithographic mask. The standard overlay targets are BiB (box-in-box) targets, which allow the detection of shifts or of displacements of the second mask relative to the first mask. Since the BiB targets have a rather coarse structure, they are now more and more replaced by AIM (advanced imaging metrology) and micro AIM overlay targets.

(64) Up to now, the overlay targets are positioned in the scribe lines of the integrated circuits. FIG. 7 schematically presents a displacement vector map or a displacement vector field measured at overlay targets arranged in scribe lines of integrated circuits. The arrow tips of the individual vectors of the displacement vector field indicate the directions of the displacement of the respective positions of the second mask with respect to the first photolithographic mask. The lengths of the vectors denote the magnitude of the shift of the respective positions of the second mask relative to the first mask. It can be seen from FIG. 7 that the restriction of the placement of the overlay targets to the scribe lines results in an irregular distribution of the overlay measuring points across the wafer.

(65) With shrinking sizes of the structures of integrated circuits and, on the other hand, increasing sizes of integrated circuits, it will no longer be sufficient to determine the overlay at the scribe line, but not on the die on the integrated circuit itself (in-die measurement). FIG. 8 schematically shows a grid of regular nodes of overlay measuring points where the nodes are arranged in the scribe lines as well as on the die area of the integrated circuit itself. This dense grid of overlay measuring points allows the determination of the overlay error with a high spatial resolution. A dense grid of overlay measuring points is a prerequisite for a beneficial application of the inventive method.

(66) Now back to FIG. 6, the overlay error of the second ((m+1).sup.th) mask relative to the first (m.sup.th) mask can be measured by using the overlay metrology system 300 of FIG. 3 in order to determine a displacement vector field similar to FIG. 8. When the measured overlay error is below a predetermined threshold and the fabricated integrated circuit is at the end working properly, the first and the second masks can be used for the fabrication of the desired integrated circuit.

(67) If the measured overlay error does not fulfil the predetermined specification, the projection device is readjusted in order to reduce the overlay error (box 645). After removing the photo resist from the wafer, a new layer of photo resist material is dispensed on the wafer (not shown in FIG. 6). Then, the overlay measurement is repeated (box 660). When the measured overlay error fulfils the requirement with respect to the predetermined overlay error, the first and the second masks are ready for being used for the production of the respective integrated circuit.

(68) In case the overlay error is still too high, the overlay errors are analyzed based on measured displacement vectors as indicated in FIG. 8 in order to determine an arrangement of pixels for the second photolithographic mask (box 680). The writing of the arrangement of pixels with the laser source 435 of the apparatus 400 of FIG. 4 in the substrate of the second photolithographic mask shifts the pattern elements on the surface of the second photolithographic (box 690).

(69) The writing of pixels can be limited to the active area of the photolithographic mask. The correction of pattern placement errors in the active area of the photolithographic masks is very effective, since the correcting pixels can be placed close to the error positions. On the other hand, when the writing of pixels is not restricted to the active area, the flexibility of the error correction process is enhanced. If the writing of the pixels can be limited to the non-active area, the introduction of new errors in the active area of the substrate of the photolithographic mask by the pixel writing process can be avoided. Since the distance between the pattern placement error and the correcting pixels may be large, the effectiveness of the correction process may be lower. This may partly be compensated as the writing of pixels does not have to consider a variation of the optical transmission.

(70) After correcting of the second photolithographic mask, the wafer is prepared for a second illumination or exposure with the corrected second mask as described above. At the second illumination of the corrected second mask, its overlay error with respect to the first photolithographic mask is significantly reduced, so that the mask combination fulfils the predetermined overlay error.

(71) In contrast to FIG. 2, the use case of a set of photolithographic masks according to FIG. 6 eliminates the necessity to send defective photolithographic masks from the wafer processing site to the mask fabrication site. Moreover, since the same metrology tool is used for the measurement of the overlay error and of the corrected photolithographic mask, tool related impacts on the measured data are avoided.

(72) The following second section describes examples of the application of the inventive principle to double patterning lithography (DPL) processes. As already mentioned, in DPL the pitch size, which typically limits the patterning resolution, can be doubled for a pattern on a wafer compared to a single illumination or exposure. The inventive principle is in the following described in the context of DPL processes. However, it is recognized that the inventive principle can also be applied to lithography processes which use more than two photolithographic masks or more than two patterning processes for the generation of a pattern of a single layer on a wafer.

(73) A first example describes the critical dimension uniformity (CDU) and overlay process control in litho-etch-litho-etch (LELE) and litho-freeze-litho-etch (LFLE) DPL processes. FIG. 12 schematically depicts a cut of a cut-out of one-dimensional pattern generated with a LELE of LFLE process. As briefly discussed at the description of FIGS. 9 and 10, the first lithography and etch step generates the lines L1 (indicated in dark grey in FIG. 12). The second lines L2 (light grey rectangles in FIG. 12) are fabricated a in second litho-etch process. The combined DPL pattern results in lines L1 and L2 which are separated by a first space S1 between lines L1 and L2 and by a second space S2 between lines L2 and L1.

(74) The upper row of lines and stripes of FIG. 12 shows a perfect form of the lines L1 and L2. Moreover, lines L1 and L2 are at their predetermined positions, and thus also automatically generating spaces S1 and S2 having an identical stripe width.

(75) The middle part of FIG. 12 illustrates a situation in which the lines L2 have a perfect form and are located exactly at their predetermined position. On the other hand, the lines L1 also have their predetermined position but their width decreases in the example of FIG. 12 from left to right. The decreasing width of the lines L1 leads to a corresponding increase in the widths of the spaces S1 and S2. The decreasing width of the lines L1 results in CDU errors of lines L1 as well as of the spaces S1 and S2. These CDU errors can be corrected by introducing an arrangement of local persistent modifications in mask 1 which generates the pattern of lines L1 of the upper part of FIG. 12.

(76) The CDU problem can be compensated by varying the optical transmission of mask 1 in the portion not having the CDU variation a in the range of up to some percent, i.e. by artificially decreasing the optical transmission of the defect free portions of mask 1. The portions of mask 1 having a local CDU variation are compensated by locally varying the portions of mask 1 with respect to the portion having the maximum CDU error (which is the left part or the middle row of FIG. 12). A respective increase the exposure dose of mask 1 by the projection system of the photolithographic exposure system corrects the CDU variation of mask 1 to a large extent.

(77) The lower part of FIG. 12 presents perfect forms or shapes of lines L1 and also lines L2. However, in the cut-out of FIG. 12 the lines L2 are shifted with respect to their predetermined position which is indicated by the vertical dash dotted lines. The shift of lines L2 leads a smaller width of spaces S1 and to a broader width of spaces S2 compared the predetermined space widths illustrated in the upper part of FIG. 12. This results in CDU errors of spaces S1 and S2.

(78) The CDU errors of spaces S1 and S2 are corrected by introducing at least one arrangement of local persistent modifications in mask 2 of the DPL process which compensates the local pattern placement error of the lines L2 of mask 2 in the cut-out of the line space pattern presented in FIG. 12.

(79) If a line shape pattern shows the combined errors of the middle and the lower part of FIG. 12, these errors can also be corrected by introducing at least one respective arrangement local persistent modifications in mask 1 as well as in mask 2.

(80) The CDU variation of mask 1 in the middle part of FIG. 2 and the local shift problem or the local overlay error of the lines L2 of mask 2 manifest in an overlay error in the DPL process. This means that a DPL process entangles local errors of the individual masks forming the combined pattern of a layer on a wafer to overlay errors of the pattern of the single layer. It is therefore mandatory to correct local errors of the individual masks of a DPL process in order to meet the tight overlay budget of technology nodes smaller than 32 nm.

(81) As already mentioned, a prerequisite for the error correction or error compensation in DPL processes or generally in multiple patterning lithography (MPUL) processes is a determination of the local errors with high resolution. In FIG. 3 a high resolution metrology tool has been presented which uses photons for the defect detection. In addition, a scanning electron microscope (SEM) as well as a scatterometry, a proliferometry, an atomic force microscope (AFM) or other metrology tool types can also be applied in order to detect local defects of photolithographic masks which manifest in printing errors on a wafer. Electrons in an electron beam can be focussed to a small spot, thus an SEM can resolve structures down to the nanometer range.

(82) There are additional metrology technologies available which have a resolution in a sub-nanometer range. Most high resolution imaging technologies have a trade-off due to a high numerical aperture (NA), and are hence very limited in the depth of focus (DOF) and have limited imaging capabilities closed to the wafer surface. Consequently, such tools may have difficulties to perform the imaging of the top layer and of previous layer(s) that are expected to be aligned. The test mask method (serving both single and multiple patterning) is solving this issue in multiple surfaced layers while the other methods are limited to multiple patterning, where all layers of interest for the specific process step are placed in the same topographical surface.

(83) As already mentioned, MPL processes comprise double, triple, quadruple, quintuple, etc. patterning lithography processes. Presently, DPL processes are preferred.

(84) In order to provide the high resolution required for the fabrication of patterns of technology nodes smaller than 32 nm, in-die measurements are required on the wafer. The metrology tools discussed above can be utilized for this purpose. Furthermore, the high-resolution error determination by in-die measurements simultaneously provides an indication or sets a local flag at the position of the error correction by introducing at least one arrangement of local persistent modifications in the substrate of the respective photolithographic mask.

(85) At the moment, the pitch size of a single exposure lithography process is limited to about 60 nm for lithography processes having a high uniform pattern. However, the fabrication of state of the art ICs requires a pitch size of less than 40 nm. FIG. 13 shows a two-dimensional (2D) pattern 1300 generated in a DPL process which can achieve pitch sizes of less than 40 nm in two vertical directions. The horizontal lines 1310 are a regular arrangement of lines separated by regular spaces 1320 which are generated by a first mask. The vertical lines 1330 are a regular arrangement of lines separated by regular spaces 1340 which are generated by a second mask in a LELE or an LFLE process. The circles 1350 symbolize schematically the designed shape and the predetermined location of an arbitrary pattern element which is used as a reference pattern element. Reference pattern elements 1355 having a form similar to the element 1350 are for example a contact hole (CH), a via, a memory capacitor, micro-electro mechanical system (MEMS) elements, or any other 2D and/or three-dimensional (3D) electronic device.

(86) As will be seen in the following Figures, any deviation from the CDU specification will cause a distortion of the printed 2D pattern. Moreover, any local shift or displacement of the lines 1310 or 1330 of one of the first or the second mask also results in a distortion of the printed 2D pattern.

(87) The diagram 1400 of FIG. 14 schematically illustrates the 2D pattern 1300 of FIG. 13, wherein the first mask which prints the horizontal lines 1310 has a local pattern placement problem or local overlay error with respect to a predetermined line position of the first mask. The displacement problem of FIG. 14 leads to the printing of lines 1415 instead of lines 1310. The displacement caused by a shift of lines 1415 is represented in the cut-out 1400 of FIG. 14 by a shift of the reference pattern elements 1350 of FIG. 13 in the vertical direction to a new position 1455 in FIG. 14. In essence, the overlay of printed (horizontal) lines 1415 will cause an overlay issue of the contact holes or any other 2D features, and the same goes for any shift in the vertical lines 1330.

(88) The local shift of the reference pattern elements 1455 of FIG. 14 can be compensated by introducing local arrangement(s) of local persistent modifications in the first mask generating the horizontal lines 1415, and thus compensating the local shift of the horizontal lines 1415. The defect compensation will bring the lines 1415 in close agreement with the lines 1310 of FIG. 13, whereby the reference pattern elements 1455 are shifted to the positions of the reference pattern elements 1350. A global shift of all feature elements of the first mask can be compensated by a scaling and/or orthogonality correction of the scanner of the photolithographic exposure system.

(89) FIG. 15 schematically represents a cut-out 1500 of the 2D pattern 1300 of FIG. 13, wherein the line width of the vertical lines 1535 increases from right to left. This is accompanied by a respective decrease of spaces 1545 between the vertical lines 1535. The inhomogeneous line width 1535 is caused by a critical dimension uniformity (CDU) variation of the pattern of the second mask used for the generation of the vertical lines 1535. As shown in FIG. 15, the reference pattern elements 1350 of FIG. 13 are deformed from cycles to ellipses 1555 and 1560, whereby their vertical half axes represents the varying space 1545 between the vertical lines 1535. It has here to be stressed that the elliptical deformation is just an example and that any deviation of a 2D or 3D structure from its original form can be used to detect a CDU variation.

(90) The ellipses 1555 and 1560 schematically represent the impact of a local CDU variation on the finally printed 2D pattern distortion. The major and minor axes of the ellipses 1555 and 1560 can for example be measured with a SEM and this metric can be used to calculate the overlay error.

(91) The overlay problem of FIG. 15 can be compensated by writing at least one arrangement of pixels in the substrate of the second mask which introduces a variation of the optical transmission in the second mask as discussed above in the context of FIG. 12.

(92) The pattern cut-out 1600 of FIG. 16 shows the 2D pattern 1300 of FIG. 13 having a local shift in the horizontal lines 1615 of FIG. 14 relative to the ideal lines 1310 of FIG. 13. Furthermore, the vertical lines 1625 have the CDU error of FIG. 15. The shift and deformation of the reference pattern elements 1350 of FIG. 13 are indicated in FIG. 16 by the shifted ellipses 1655 and 1660. FIG. 17 depicts the enlarged ellipses 1655 and 1660 of FIG. 16.

(93) The errors of the two masks are corrected as explained during the discussion of FIGS. 14 and 15. The shift as well as the dimensions of the major and the minor axes of the ellipses 1655 and 1660 can again be measured with a high-resolution metrology tool, as for example a SEM. Similar to FIG. 14, the local shift of the vertical lines 1615 can cause placement errors of the reference pattern elements 1350 which are for example contact holes (CHs). Additionally, it may lead to some distortion issues. The critical dimension (CD) non-uniformity will cause a CD distortion. Again analog to FIG. 15, the CDU error of the second mask used for the generation of the vertical pattern 1635 typically results in a distortion of the reference pattern elements 1350 as schematically indicated by the ellipses 1655 and 1660.

(94) FIG. 18, which is taken from the paper Towards 3 nm overlay and critical dimension uniformity: an integrated error budget for double patterning lithography of W. A. Arnold mentioned in the second part, illustrates schematically some important process steps of a self-aligned double patterning (SADP) process. A layer 1820 to be etched or patterned is arranged on a wafer 1810. On top of the layer 1820 a so-called hard material photo resist of a sacrificial layer 1830 is deposited. The sacrificial layer 1830 is a photo resist or a hard mask layer and may comprise silicon nitride, silicon oxide, or any other appropriate material for this purpose or function. A photo resist layer 1840 is arranged on the sacrificial layer 1830. The wafer 1810 is then exposed to a photolithographic mask which forms dense pattern elements 1850 and isolated pattern elements 1870. In the example of FIG. 18, the pattern elements 1850 and 1870 have a nominal width of 1855 or 1875, respectively. The actual widths 1860 and 1880 of the pattern elements 1850 and 1870 on the wafer 1810 are smaller than the nominal or predetermined width 1855 or 1875, respectively. The small widths 1860 and 1880 generated in the lithography step cause a CDU error in the layer 1820 to be etched.

(95) Further, the spacer deposition and the subsequent etching process (as indicated in FIG. 11) introduce a further variation in the spacer elements 1890 formed by the etching process. After the removal of the remaining photo resist 1840 and of the sacrificial layer 1830, the pattern formed by the spacer elements 1890 is etched in the layer 1820, wherein the spacer elements 1890 act as etch stop elements. Finally, the spacer elements 1890 and the sacrificial layer portions below the spacer elements 1890 are removed from the generated pattern elements of the layer 1820.

(96) In the etching process a variation of the lithography pattern 1850, 1870 or a CD error transforms in an error of the widths of the stripes S2 as can be seen in the lower part of FIG. 18. Furthermore, the etching process converts a variation of the spacer elements 1890 in a variation of the lines L1 and L2. A variation of the lines L1 and L2 also results in a variation of the spaces S1. Consequently, the pattern generated with the SADP process shows an overlay error; an effect which is known as pitch walking.

(97) The errors in the one-dimensional (1D) line space pattern of the SADP process can again be measured by using the metrology tool described in FIG. 3, by using for example a SEM and/or a scatterometer.

(98) The defined method enables to correct the defects of the SADP pattern of FIG. 18. In a first step at least one arrangement of pixels is introduced or written in the photolithographic mask used for the generation of the pattern 1850 and 1870 in a lithography process. The arrangement(s) of pixels compensate the CDU variation by introducing a respective variation of the optical transmission in the substrate of the photolithographic mask. This process brings the spaces S2 in FIG. 18 in close agreement with the predetermined space.

(99) In a second step, an arrangement of local persistent modifications is introduced in the sacrificial layer 1830. The local persistent modifications or the pixels in the sacrificial layer 1830 induce a structural change of the material of the sacrificial layer 1830 and thus avoiding a variation of the lines during the etching of the layer 1820. The modified sacrificial layer 1830 prevents pitch walking effects during the etching step of the layer 1820. As a consequence the lines L1 and L2 are generated having a uniform width. The pixels are written in the sacrificial layer 1830 prior to the deposition of the photo resist 1840. In case the sacrificial layer 1830 is placed wrong, it is also possible to apply a RegC process in order to correct the local as well as the wrong placement of the sacrificial layer 1830, while the scanner can correct only global (low frequency) overlay errors.

(100) This means that a critical dimension non-uniformity (CDNU) in a sacrificial layer 1830 leads to overlay error in a SADP process. This error can be measured by conventional wafer fab metrology tools or systems such as but not limited to: CD SEM (critical dimension scanning electron microscope), AFM, scatterometry and/or profilometry. Then, the identified overlay error can be fixed by applying at least one arrangement of local persistent modifications in the sacrificial layer 1830 similar than described above for the substrate of a photolithographic mask.

(101) In the following, a further example for the application of the inventive principle is presented. In this example, the at least one error on a wafer is measured with the aid of a test mask. For this purpose, a test mask is designed and created which has test features which are aligned with at least one mask used for the fabrication of ICs on a wafer. This mask is in the following also called a production mask. The test features are in-die test features in order to secure a high resolution at the error detection.

(102) FIG. 19 schematically presents a test mask 1900 having a test pattern 1910. The actual test pattern 1910 of the test mask 1900 is designed by IC or device designers. It is also beneficial to consult optical proximity correction (OPC) designers at the determination of the layout of the test pattern 1910 of the test mask 1900. The test pattern 1910 of FIG. 19 comprises dense pattern elements 1920, semi dense (or denso) pattern elements 1930, and isolated pattern elements 1940.

(103) The test mask 1900 is investigated in detail on mask level. Further, the pattern printed by the test mask 1900 on a wafer is also investigated with respect to CDU and pattern placement errors. Thus, the test mask 1900 can be regarded as a reference test mask which prints its pattern elements 1910 in a minimal and yet well mapped and documented CDU and pattern placement error.

(104) The test mask 1900 is presently preferably applied in combination with a second mask or a production mask in a double patterning process. It is however appreciated that the test mask concept can also be applied in the fabrication process of a wafer using a single exposure lithography technology. In the following, the concept is discussed in the context of a DPL process.

(105) In a first step, the test mask 1900 is exposed to a wafer. After the development of the exposed photo resist, the test pattern 1910 of the test mask 1900 is etched in the wafer. The combination of the exposure and etching process is also called printing of the test mask 1900. In the second step, the photolithographic mask used in the fabrication of a DPL layer on the wafer or a production mask is also printed on the wafer as described above for the test mask 1900.

(106) Standard very large scale integration (VLSI) processes require a surface planarization in order to avoid a problem with the depth of focus (DOF) due to a non-planar surface topography of the photo resist on the wafer. The proposed method overcomes this challenge. In order to minimize deviations from a standard fabrication process, it is however advantageous to print the test mask 1900 on a thin layer with a high optical contrast to ease the metrology process, as this layer can be treated like a hard mask, for example a silicon nitride layer, and to perform a full lithography and etch process on the test mask followed by a printing process of the production mask used in the DPL process and then performing the overlay and CD measurements in-die on the wafer.

(107) FIG. 20A schematically shows a test pattern 2010 arranged on the nitride layer 2020 of the wafer 2000. The test pattern 2010 has been generated in a first litho-etch (LE) process (right part of FIG. 20A). After performing an etching step and after applying a second photo resist 2030, the hard mask layer 2020 allows a minimal topography, i.e. is essentially flat. This is indicated in the left part of FIG. 20A. On the other hand, FIG. 20B illustrates an example wherein the wafer 2005 does not have a hard mask layer 2020, and the surface of the photo resist 2035 shows a distinct topography 2045. The topology can negatively affect both, the photolithography process and the metrology resolution, accuracy and precision. Therefore, it is recommended to us the method of FIG. 20A.

(108) In the next step, the overlay error between the test mask 1900 and the production mask is measured. This can be performed by using the metrology tool described in FIG. 3, by using a SEM, and/or by using a scatterometer and/or by using additional image base or modeling base metrology methods. The overlay error map of the overall wafer is determined by stacking the relevant sampling fields of the individual exposures of the test mask 1900 and the production mask across the wafer.

(109) The overlay errors on wafer level are then compensated by writing one or more arrangements of pixels in the substrate of the production mask. The process of printing and measuring of the test mask 1900 and the corrected production mask is then repeated in order to check whether the correction process by the writing of pixel arrangements has been successful.

(110) The described procedure is performed for all overlay critical production masks in order to minimize the overall overlay error budget. It is in this process beneficial to include in the test pattern 1910 of the test mask 1900 pattern elements that allow checking of the patterns of as many as possible production masks. This procedure minimizes the cost of the test mask concept as only one or a very limited number of test masks 1900 are required. On the other hand, this procedure also reduces error introduced at the transition from a first test mask to a second test mask. This is shortly explaining with the aid of the following examples.

(111) The first example describes the application of a single test mask 1900 for the control of several patterns of several layers of a wafer. Two layers, called A and B, are supposed to the overlay critical layers. The patterns P of the layers A and B at the position i are called PA.sub.i and PB.sub.i. A test pattern 1910 of a test mask 1900 has test pattern TA.sub.i, and TB.sub.i, wherein the test pattern TA.sub.i and TB.sub.i are close to each other. This allows the assumption that the placement error of the test pattern TA.sub.i and TB.sub.i are identical. This means that one test pattern T.sub.i can actually serve for the test of both pattern PA.sub.i and PB.sub.i:
TA.sub.i=TB.sub.i=T.sub.i(1)

(112) In a first step, the test mask 1900 and the mask or production mask for the layer A are printed by using a LELE or a LFLE process and the displacements of the pattern elements relative to the test pattern are measured:
?A.sub.i=PA.sub.i?TA.sub.i=PA.sub.i?T.sub.i(2)

(113) Then, the test mask 1900 and the production mask for the layer B of the wafer are printed in a LELE of LFLE process, and the resulting displacements between the test pattern 1910 and the pattern of mask B are determined:
?B.sub.i=PB.sub.i?TB.sub.i=PB.sub.i?T.sub.i(3)

(114) From the two measurements presented in equations 2 and 3 the relative displacements of the pattern elements PA.sub.i and PB.sub.i can be determined:
?AB.sub.i=PA.sub.i?PB.sub.i=?A.sub.i?T.sub.i?(?B.sub.i?T.sub.i)=?A.sub.i??B.sub.i(4)

(115) As already mentioned, the relative location of the test pattern TA.sub.i and TB.sub.i can additionally be obtained from a detailed analysis of the test mask 1900 using one or more of the metrology tools described above.

(116) In the following example, two test masks A and B are designed, wherein the test mask A comprises the test pattern TA.sub.i and the second test mask B comprises the test pattern TB.sub.i. The test pattern of each test mask contains an alignment element AL, as for example an Archer alignment mask. It is again assumed that the test patterns TA.sub.i and TB.sub.i are close to the alignment AL, so that the placement error of the test patterns TA.sub.i and TB.sub.i relative to the respective alignment element AL can be neglected.

(117) The test masks A and B are applied in order to check two overlay critical layers A and B of a wafer. For this purpose, a test print of the mask A is performed and the locations of the test pattern TA.sub.i are determined by using the alignment marks ALA.sub.i:
TA.sub.i=ALA.sub.i??TA.sub.i(5)

(118) Then, a test print of the mask B is performed and the locations of the test pattern TB.sub.i are determined by using the alignment marks ALB.sub.i:
TB.sub.i=ALB.sub.i??TB.sub.i(6)

(119) In a first LELE or LFLE process the test pattern TA.sub.i of the test mask A and the pattern PA.sub.i of the production mask A are printed. This results in displacements of the pattern elements PA.sub.i relative to the test pattern TA.sub.i:
?A.sub.i=PA.sub.i?TA.sub.i=PA.sub.i?(ALA.sub.i+?TA.sub.i)=PA.sub.i?ALA.sub.i??TA.sub.i(7)

(120) In a second print process the test pattern TB.sub.i of the test mask and the pattern P.sub.i of the production mask B are printed which leads to displacements of the pattern elements PB.sub.i relative to the test pattern TB.sub.i of test mask B:
?B.sub.i=PB.sub.i?TB.sub.i=PB.sub.i?(ALB.sub.i+?TB.sub.i)=PB.sub.i?ALB.sub.i??TB.sub.i(8)

(121) From the two measurements the relative displacements PA.sub.i and PB.sub.i are known according to:
?AB.sub.i=PA.sub.i?PB.sub.i=?A.sub.i+(ALA.sub.i+?TA.sub.i)?(?B.sub.i+(ALA.sub.i+?TA.sub.i))(9)

(122) wherein the terms in the brackets are determined from the test prints of the test masks A and B as described above.

(123) A possible variation of the presented procedure is the design of the test pattern TA.sub.i and TB.sub.i that allow a measurement of their relative position. In this case, the alignment marks AL are redundant and can serve for further qualification purposes.

(124) FIG. 21 schematically illustrates a field of a wafer layout after the test pattern 1910 of the test mask 1900 of FIG. 19 (dark grey elements) and a vertical pattern 2110 (light grey elements) of a production mask are printed in a LELE of an LFLE process. The production mask comprises dense vertical lines 2120, medium dense lines 2130 and one isolated line 2140. For the production mask of FIG. 21 only vertical lines of the test pattern 1910 of the test mask 1900 are relevant. The dense lines 2120 and the isolated line 2140 of the production mask pattern and of the test mask pattern 1910 are essentially aligned to each other, whereas the medium dense line 2130 has a local line space error 2150 with respect to the respective test pattern element of the test mask 1900. The detected local overlay error 2150 between the test pattern element and the vertical line pattern of FIG. 21 is caused by an in-die placement or registration error. It can be corrected by writing an arrangement of pixels in the range of the medium dense lines 2130 of the mask of FIG. 21 in order to compensate the defect of the production mask of FIG. 21.

(125) FIG. 22 schematically presents the situation in which the test mask pattern 1910 of the test mask 1900 is printed in a litho-etch (LE) process. In a first step, a LE process of the LELE process the contact holes of a production mask are printed and etched. The contact holes (CHs) are indicated in FIG. 22 by cycles. Many thousands CHs are etched in a field of a layer of a wafer which forms a layer of a modern IC. Only five of them are important for a test of the CH arrangement by the test pattern 1910 of the test mask 1900. These are depicted in FIG. 22 by dark grey circles 2220, 2230, 2240, 2250, and 2260. Four small squares 1915 of the test pattern 1910 are used to check the placement of the CH arrangement.

(126) As can be seen from FIG. 22, the four test CHs 2220, 2230, 2240, and 2250 are essentially at their predetermined position, whereas the CH 2260 is shifted with respect to its predetermined location. The placement error of the contact hole 2260 can be corrected by writing an arrangement of pixels in the range of the erroneous CH 2260, which shifts the CH 2260 essentially to its predetermined position. The in-die error determination provides an indication where to position the at least one arrangement of local persistent modifications used for the defect compensation. The expression essentially means here as well as on other positions within this description a positioning within the resolution limit of a state-of-the-art metrology tool.

(127) Similar to FIGS. 21 and 22, FIG. 23 depicts a wafer pattern printed and etched with the test pattern 1910 of the test mask 1900 of FIG. 19 and a production mask having a horizontal line space pattern. For this use case, only the horizontal bars 1925, 1935 and 1945 of the test pattern 1910 are relevant and used for a check of the horizontal lines 2320 and 2330 of the production mask of FIG. 23. As can be seen from FIG. 23, the lines 2320 and 2330 of the production mask are simply vertically shifted relative to the test pattern elements 1925, 1935 and 1945. This means that the detected error is a global overlay error. As already explained at the discussion of FIG. 19, the test pattern 1910 of the test mask 1900 is investigated in great detail, so that the test pattern can be taken as a reference pattern. Thus, the shift detected between the test pattern elements 1925 and 1935 and the line 2320 and the test pattern element 1945 and the line 2330 can be attributed to a shift of the lines 2320 and 2330 of the production mask. Therefore, the test mask pattern 1910 can not only be used to detect local errors of a production mask, but can also be used analyze global overlay errors of a production mask.

(128) The displacement error of the horizontal lines 2320 and 2330 of the mask of FIG. 23 detected by in-die measurements is compensated by a linear transformation (scaling and orthogonality correction) of the scanner used in the DPL process. This means that the test mask concept, which is based on in-die measurements, can also be used to correct global overlay errors, and thus minimizing the overlay error of overlay error tight DPL processes as well errors of single patterning processes where the overlay is measured between layers in different topographic planes on a wafer.

(129) Alternatively, it is also possible to write arrangements of pixels in the mask of FIG. 23 in order to bring lines 2320 and 2330 in line with the test pattern elements 1925, 1935 and 1945.

(130) FIG. 24 schematically presents a superposition of a field of a wafer layout of the three various layers of FIGS. 21 to 23. The corrective actions of the individual masks in order to minimize the overlay error of the various masks have been indicated at the discussion of the various layers. FIG. 25 schematically represents the superposition of the various layers of FIG. 24 after the errors of the individual errors have been corrected with respect to the corresponding pattern elements of the test pattern 1910 of the test mask 1900. As can be seen from FIG. 25, the overlay error of the layers of FIGS. 21 to 23 is minimized by the corrective actions of each mask.

(131) The examples presented up to now refer to a single exposure of a mask on a wafer. However, a wafer layout comprises the subsequent exposure of a mask many times in order to cover the overall wafer area with the pattern of a mask in order to print a layer of an IC to be fabricated many times on the wafer. A single exposure of a mask in on a wafer is called a field. Thus, the error discussed up to now are in-field errors.

(132) In the following, a data analysis for an overlay error determination and compensation on the wafer level is discussed. To be more precise, a method for the determination of a route mean square (RMS) overlay error across a wafer and its compensation is presented. FIG. 26 schematically represents a field 2600 of a single exposure or a multiple exposure as for example a double exposure of a mask. The field comprises three in-field metrology sites 2610, 2620 and 2630 at which the in-field overlay error is determined. It is important to emphasize that the above methods allow measurements of an in-die overlay metrology on real product features, and thus providing a unique opportunity to control the overlay of product features rather than test features that may or may not be correlative to the actual device features' overlay.

(133) FIG. 27 schematically shows a wafer 2700 which is covered by fields 2720 of subsequent exposures of a photolithographic mask. In a LELE or LFLE process two masks are required for a single in-field exposure. Each of the fields 2600, 2720 on the wafer 2700 has the overlay in-field targets 2610, 2620 and 2630 of FIG. 26. As is indicated by the crossed overlay targets in FIG. 27, an area from a sub-millimeter range up to several millimeters from the edge 2710 of the wafer is excluded from the determination of the overlay error across the wafer 2700. This measure excludes wafer edge bias effects at the determination of the overlay error at wafer level. Based on wafer fab experience, the edged die effect cannot be compensated by lithography processes and should therefore not taken into account while averaging metrology data of all fields as an input of RegC and CDC processes.

(134) In the next step, the overlay error on field level measured at multiple points per field, three positions 2610, 2620 and 2630 are shown in FIG. 26. For the determination of the in-field overlay error one of the methods presented above can be used. The in-field measurement focuses on hot spots in the exposure area of a mask in order to have sufficient experimental data for a statistically significant in-field error determination. A full wafer or a field sampling can be used to create an average field-level error based on averaging of each of the targets at the sampled fields.

(135) The in-field overlay error map of the individual fields 2600, 2720 are then stacked. The mask is then corrected by writing at least one arrangement of pixels in the substrate of the mask. A new wafer is printed with the corrected mask based on average metrology of all wafer fields except for edge-affected fields and the correction process is applied on this average in order to achieve a minimal error across the wafer. The overlay error resulting from the corrected mask is measured on both, the in-field level and the wafer level in order to check the error compensation improvement.

(136) It is important to note that the number of measurement points per field should be defined by the manufacturer of the device fabricated from the process wafer in order to optimize the trade-off between the correction efficiency and the metrology productivity as more measurement sites means more metrology time and higher cost of the end-product. The discussed methods are relevant to work with any given sampling plan of the device manufacturer.

(137) FIG. 28, which is taken from an internet publication, presents an example of an overlay error map 2810 across a wafer 2800. In order to obtain the error contributions which cannot be compensated by a scaling or orthogonality (S/O) correction, these errors are removed from the overlay error map 2810 of FIG. 28. The resulting overlay error map 2900 is presented in FIG. 29. In the in-field error map 2900 of FIG. 29 three regions 2910, 2920 and 2930 can be identified as examples of areas in the field which have a systematic error.

(138) FIG. 30 illustrates the remaining in-field error distribution 3010 of FIG. 29 after the systematic errors 2910, 2920 and 2930 have been corrected by introducing respective arrangements of local persistent modifications in the substrate of the photolithographic mask. The systematic errors 2910, 2920, and 2930 of the combined mask and lithography process, which result in an increase of the respective overlay error, are essentially removed.