TWI285797B - Method for designing an overlay mark - Google Patents

Abstract

The present invention discloses a method for designing an overlay mark having an optimized pitch and line-to-pitch ratio. The present method comprising steps of irradiating an overlay mark with a probe beam; measuring a diffractive light from the interaction of the probe beam and the overlay mark; selecting parameters of the overlay mark to be optimized to increase the sensitivity of overlay measurement; optimizing the parameters using an optimization algorithm for increasing sensitivity of overlay measurement, and designing the overlay mark on a wafer according to the parameters with the highest sensitivity.

Description

1285797 IX. Description of the Invention: [Technical Field] The present invention relates to a method for designing a registration mark, and more particularly to a method for designing and aligning an alignment mark having an optimized grating period and a grating line width period ratio . [Prior Art] The pros and cons of process detection technology have always been an important factor in the yield of semiconductor processes. With the requirements of semiconductor process technology for stack-to-measurement measurement, it is becoming more and more stringent according to ITRS (International Technology R〇admap). For

Semiconductor) report on the development of semiconductor-related technologies; the stack-to-measure accuracy requirement for nanowire-line processing is 3.5 nm, 90 nm linewidth. The stack-to-measure accuracy requirement for the process is 3.2 nm, as the process line width is smaller and smaller, the 65 nm line width next generation semiconductor process stacking measurement accuracy requirements will reach 2.3 nm, the traditional optical image type (bright_field microscope) stack It is difficult for the measuring machine to meet the requirements of this measurement accuracy due to problems such as limitation of the diffraction limit and its own measurement accuracy. Recently, Lu was proposed to use the rigorous coupled wave theory (RCWT)-based scattering measurement method. Because the scatterometer machine itself has good repeatability and reproducibility, it is considered to be the next generation. The semiconductor stack is a powerful tool for measuring, so how to improve the accuracy of this scattering measurement method to meet the increasingly stringent requirements has become a very important topic. SUMMARY OF THE INVENTION The object of the present invention is to provide a design method for the quaternary (4) compliant filament period and the grating line width period ratio alignment mark. PD0086.doc 100210 PDoo86 004965308-1 1285797 To achieve the above object, the present invention discloses a method of designing an alignment mark with optimized grating period and grating line width period ratio. The design method of the alignment mark of the present invention firstly irradiates a pair of bit marks with a light beam, and then measures a diffracted light generated by the light beam passing through the alignment mark by a light detector. Next, a plurality of parameters of the alignment mark are selected, and the plurality of parameters are optimized to increase the sensitivity of the overlay measurement. Finally, the alignment mark of the wafer is designed using the alignment mark with the greatest sensitivity. [Embodiment]

Figure 1 is a schematic illustration of a pair of bit marks 10. The alignment mark 10 is composed of a first grating 14 and a second grating 16, and the pitches of the first grating 14 and the second grating 16 are the same. The first grating 14 is composed of photoresist, the second grating 16 is composed of silicium oxide, the intermediate layer 18 is a uniform polysilicon, and the bottom layer 12 is a germanium substrate ( Silic〇n substme). The thickness, refractive index and extinction coefficient of each layer are summarized in Table 1. Table I

The t/reposition mark 10 stack-to-error measurement sensitivity factor is measured in addition to the machine's unique 1±, mechanism design, back-end detector quality and signal processing technology—PD0086.doc 1285797 The difference in structural parameters of the sample also changes the shape of the characteristic map of the reflected light and affects the degree of dispersion of the characteristic maps. For example, the refractive index and extinct〇n coefficient of different materials, thickness, geometry, and sidewall parameters of the sample after etching will affect the alignment mark overlap error. The measurement sensitivity. The material and thickness are determined by the characteristics of the component, while the angle of the sidewall is related to the sizing machine. Therefore, the parameters that can be adjusted by the designer are the geometry of the sample to be tested. As far as the grating alignment mark used by the scatterometer machine is concerned, its geometry includes two parameters of a grating period (phch) and a line-to-pitch ratio (LS ratio). 2 is a schematic diagram of the system architecture of an angle scatterometer 20. Although an angle scatterometer is exemplified below, the present invention is also applicable to a spectroscopic reflect 〇rmeter, a multi-wavelength spectroscopic ellipsometer. The angle scatterometer 2 is a single-wavelength laser light incident and multi-angle scanning (2 - phantom optical system architecture, the angle between the incident light 22 generated by the laser source 30 and the normal 26 and the angle between the reflected light 24 and the normal 26 The photodetector 32 only receives the zero-order reflected light. The incident light source can be selected from the currently used laser light, including an argon ion laser (wavelength of 488 nm and 514 nm); and a cadmium-doped laser (wavelength of 442) Nano); 氦氖 laser (wavelength 612 nm and 633 nm); 鈥-钇 aluminum 柘 laser (wavelength 532 nm). By changing the incident angle 0, an incident light angle and reflected light can be obtained. The characteristic map of the angle, that is, the characteristic map is a function of the incident angle 0. The incident light source is a laser light source, but a broadband source can also be used, and the obtained characteristic map is The wide PD0086.doc 100210 PDo〇86 004965308-1 1285797 is a function of the wavelength of the frequency source. The reference Fig. 1 'where the first grating 14 and the second grating 16 are both a grating period and a dimensional structure. Measure horizontal direction ) The amount of error in the vertical direction (Υ direction), so that the Ρ group has two one-dimensional light. The alignment mark of the grating structure, the grating vector vector (grating vector) is in the horizontal direction and the vertical direction. Then, scanning is performed separately by using the laser light source 3〇 as shown in Fig. 2. If the first grating 14 and the second grating 16 are both in a two-dimensional structure (refer to Fig. 3), the laser light source 3() only needs to be scanned. The alignment mark can simultaneously obtain the amount of overlap error in the horizontal and vertical directions, and the design method of the alignment mark of the present invention can also be applied to the two-dimensional alignment mark. The reflected light characteristic map of the error amount, wherein the horizontal axis is • the scanning angle (ie, the incident light angle), the vertical axis is the reflected light intensity, and the overlap error amount is bounded by 0 to 400 nm, and the pitch is 2 Meter. For the alignment marks of different geometric shapes, although the overlap of the feature maps is the same as the error amount, the degree of dispersion of the feature maps from each other will be different. In other words, φ If it is possible to find the alignment mark geometry of the feature map with the largest degree of dispersion, the easier the measurement system can distinguish the different stack error amounts, that is, the higher the sensitivity of the stack pair measurement. The intensity of reflected light in the feature map can be expressed as ··· R^\u{z2) X u(z2y\

t/W = exp[-(z2—Zi)M]t/(22) ”〇 VI

And 6 are respectively the incident position and the exit position coordinate of the light incident grating; M PD0086.doc 100210 PD0086 004965308-1 1285797 is a transformation matrix; A:. Is the wave number of the incident light between the incident light; 夂 is the wave number of the incident light on the optical axis of the grating region (A < z < a) (z axis); (iv) is the grating winding The number of shots; 丨 is the identity matrix. In the case of an angle scatterometer, Ar2(/-V) is a function of the grating period, the grating line width period ratio, the stacking error amount, and the incident light angle. Therefore, the reflected light intensity can be expressed as · i? = |ί7(ζ2) x JJ(z2 )* = R(pitch, LSratio, Θ., A0L ) If the grating period and the grating line width period ratio are fixed, Define an average standard deviation (ASD) formula as: where A is the starting angle of the incident laser beam, the heart is the final angle of the incident laser beam, Μ is the total number of sampling points of the scanning angle, and (4), △ %) is the reflected light characteristic map when the stacking error amount is Δ~. The phantom is the reflected light intensity with different stacking error amounts when the incident angle of the laser beam is Θ and (4), Δ~)| 12 / A standard deviation obtained. Therefore, the ASD value represents the degree of dispersion between the reflected light characteristic maps of different overlapping error amounts, and the larger the ASD value, the greater the degree of dispersion between the characteristic maps. As mentioned earlier, the greater the degree of dispersion, the easier it is for the measurement system to resolve the different amount of overlap error, that is, the higher the sensitivity of the overlay measurement. Conversely, the lower the sensitivity of the overlay measurement. For the reflectometer, #-ν)2 is a function of the grating period, the grating linewidth period ratio, the amount of overlap error, and the wavelength of the incident light. Therefore, the reflected light intensity can be expressed as PD0086.doc 100210 PDoo86 004965308-1 -10- 1285797 as: ^ U^y^Ripitch.LSratio,^,^) If the fixed grating period and the grating line width period ratio, then the average The standard deviation (ASD) is: ASDj|w,

咐)'t (4 recognizes lJ - like, v)) 2 A VAOL / where 4 is the starting wavelength of the incident laser beam; 1 〃 is the final wavelength of the incident laser beam, Μ is the total number of incident wavelength sampling points . For the ellipsometer, #-ν)2 is a function of the grating period, the grating linewidth period ratio, the overlapping error amount, and the wavelength of the incident light. The intensity of the reflected light can be expressed as:

R^U(z2) X U(z2y^Rp X R

R. x R where \ and the foot are respectively the amplitude of the p-polarized light and the s_polarized light of the reflected light, and are a function of the grating period, the grating linewidth period ratio, the stacking error amount, and the incident light wavelength.

R

~ = tan(^)eiA R where V and A are the parameters of the ellipsometer, which are also a function of the grating period, the grating linewidth period ratio, the stacking error amount, and the incident light wavelength. ψ = y/(pitch, LSratio, λ., A0L) Δ = A(pitch, LSratio, λί, A0L) If the grating period and the grating line width period ratio are fixed, the average standard deviation (ASD) is: PD0086.doc -11 - 1285797 Lake factory to frw, ^(Λ) = J)-ΚΛ, Δ-))2 Ν AOL, ο 咐)=Ji> "·, ν) - /ν U〇L: where; ^ is incident The starting wavelength of the laser beam, 4 is the final wavelength of the incident laser beam, and Μ is the total number of sampling points of the incident wavelength.

5(a) and 5(b) are flowcharts showing an optimization procedure of the design method of the alignment mark of the present invention, wherein p and r respectively represent the grating period and the grating line width period ratio; and X represents the pr plane. Position vector; m, u represent the step size and direction vector respectively; Ν, e represent the maximum number of iterations and the minimum step size, respectively. Referring to Fig. 5(a), the flow is based on a specific structural parameter (here, the grating period p and the grating line width period ratio r are taken as an example), and an average standard deviation ADS is first calculated (step S41). Then, an optimization algorithm (step S42) is used to determine an average standard deviation maximum value, wherein the step S42 included in step S42 is to calculate the detailed process of the ASD (refer to FIG. 5(b)), and finally the utilization occurs. The structural parameters of the mean standard deviation maximum are used to design the alignment mark. The present invention utilizes a numerical algorithm to optimize the period of the grating and the line width period. In addition, since the wavelength of the incident light or the incident angle of the incident light also affects the measurement sensitivity, the present invention can also select the optimum incident light wavelength or incident light angle in a numerical simulation manner. In addition, 'optimal algorithm is a random walk method PD0086.doc 100210 PD〇〇86 0〇49653〇8-i -12- 1285797

(random walk method)^ - ^ ^ ^ ^(simplex method) 〇 T Describe in detail the design method of the alignment mark of the present invention, which is applied to the process of measuring the sample by the angle scatterometer. Step (4), selecting - at least one parameter of the alignment mark, the parameter is used in the - alignment mark optimization process. The parameter in the basin contains the material of the sample, the thickness of the sample, and the sidewall angle of the pattern; The 豸 parameter can be determined by a constant in a comparison table (for example, Table 1). Step (b), selecting a first grating having a -th-raster feature. The first grating can be selected as a known standard target or a best empirical guess alignment mark. The first grating feature may be a grating period or a raster line width period ratio in this embodiment. Step (c), performing the alignment mark optimization process based on the first grating feature, which determines a maximum average standard deviation of one of the diffracted lights at a particular range of incident angles. The diffracted light system M is diffracted by a mathematical mode alignment mark of the first optical thumb feature. In performing the alignment mark optimization process, it includes the step of increasing a particular step (e.g., 5 nm) to the parameter (raster period or raster line width period ratio) in the first grating feature. After the parameter is increased by the specific level, the average standard deviation is recalculated until the maximum number of iterations is reached, and the maximum average standard deviation is determined. The above steps (a), 0) and (c) use a numerical simulation to analyze the software execution without measuring the actual alignment mark. Step (d), designing an alignment mark of the first grating feature corresponds to the maximum average standard deviation. Figure 6 and Figure 7 are three-dimensional plots of ASD values and their contour plots for different grating period and linewidth period ratio simulation results. When the incident laser wavelength is 633 nm, the X-axis is the grating period (the range is bounded by 〇 1 micron to 2 micron), PD0086.doc -13- 100210 PDoo86 004965308-1 1285797 The Y axis is the line width period ratio (range 1:9 to 9:1), and the Z axis is the ASD value. Table 2 shows the simulation results when using the six different laser incident light sources described above. According to the structural parameters of the sample to be tested, when the incident laser light wavelength is 612 nm, and the grating period to the line width period ratio are 0.4 μm and 48:52, respectively, the measurement system has an ASD maximum value of 0.015581, which is The total average value is 21.47 times that of 0.000726. Table 2 Wavelength (nano) ASD Minimum ASD Intermediate value ASD Maximum ASD maximum value of the structure parameter grating period (micron) grating line width period ratio 442 1.02 xl (T05 0.000144 0.002481 0.24 54:46 488 1·36χ10· 05 0.000786 0.007731 0.28 44:56 514 1.77 xl〇·06 0.000866 0.010951 0.26 48:52 532 7'43 xl(T07 0.000933 0.010542 0.28 58:42 612 2.55 xlO-08 0.001998 0.015581 0.40 48:52 633 L42xl〇·08 0.001853 0.010765 0.46 48:52 ASD total average value 0.000726 (in the grating period 0.1 to 2 microns, grating line width period ratio 1:9 to 9:1 range) It can be seen that after the optimization analysis of the present invention, it can be greatly The degree of dispersion between the characteristic maps of different stacking error amounts is increased, thereby effectively improving the sensitivity of the stacking measurement. # The technical content and technical features of the present invention have been disclosed above, but those familiar with the technology may still be based on this The invention is not limited to the embodiment of the invention, and the scope of protection of the present invention should not be limited to those disclosed in the embodiments. The replacement and modification of the present invention are covered by the following patent application. [Simplified Schematic] FIG. 1 is a schematic diagram of a dimensional alignment mark; FIG. 2 is a schematic diagram of a system architecture of an angle scatterometer; PD0086. Doc -14- 100210 PDoo86 004965308-1 1285797 Figure 3 is a top view of a two-dimensional alignment mark; Figure 4 shows a reflected light characteristic map of different stacking error amounts; Figure 5 (a) and Figure 5 (b) are FIG. 6 and FIG. 7 are respectively a three-dimensional diagram and a contour map of ASD values of different grating period and line width period ratio simulation results. [Main component symbol description] 10 Alignment mark 12 Bottom layer 14, IV First grating 16, 16' Second grating 18 Intermediate layer 20 Angle scatterer 22 Incident light 24 Reflected light 26 Normal line 30 Laser source 32 Light detector S41 ' S42 - S421 Step PD0086 .doc -15- 100210 PD〇〇86 004965308-1

Claims (1)

1285797 X. Patent Application Range: 1 · A method for designing a registration mark, comprising the steps of: illuminating a pair of bit marks with a light beam; measuring a diffracted light by a photodetector, the diffracted light system passing the light beam Produced by a registration mark; selecting a plurality of parameters of the alignment mark; and optimizing the plurality of parameters to increase the sensitivity of the overlay measurement. = The design method of the alignment mark according to Item 1, wherein the alignment mark includes a first grating and a second grating. 3. The design method of the alignment mark according to the second aspect, wherein the first grating and the first grating are each a one-dimensional structure of a grating period. 4. The method of designing an alignment mark according to claim 2, wherein the first grating and the second grating are each a two-dimensional structure having a grating period. 5. The design method of the alignment mark according to the item 1, wherein the beam is generated by a source and the measurement signal generated by the measurement is a function of the incident angle of the beam. 6. The method of designing the alignment mark according to item 1, wherein the beam is generated by a -2 source and the measurement signal generated by the photodetector is a function of the wavelength of the beam. 7. The design method of the alignment mark 'where the plural parameter 8' 9 · The design method of the alignment mark according to the request item 1 ' is applied to an angle PD0086.doc 100210 PDo〇86 004965308-1 1285797 scatterometer . 10. The design method of the alignment mark according to claim 1, which is applied to a reflector. 11. According to the design method of the alignment mark of the request item i, it is applied to a polarization meter. 12. The method of designing a registration mark according to claim 1, wherein the optimizing step comprises the steps of: calculating an average standard deviation; and • determining an average standard deviation maximum value by the optimization algorithm. The time stack has the best sensitivity to error measurement. 13. The method of designing an alignment mark according to claim 12, wherein the average standard deviation is defined by the following equation in a one-angle scatterometer system: 士 I·, 邶, where Θ1 is the starting angle of the beam, ~ The final angle of the beam, M is the total number of scanning angle sampling points, paper V) is the reflected light characteristic map when the overlap error amount is •, # represents the maximum number of iterations, 岣) is at the incident angle of the beam (4) A standard deviation obtained by the intensity of the reflected light with different stacking errors and (4), ν. 14. The method of designing a registration mark according to claim 12, wherein the average standard deviation is defined by the following equation in a reflectometer system: PD0086.doc -2- 1285797 wherein 4 is the starting wavelength of the beam, & For the final wavelength of the beam, M is the total number of incident wavelength sampling points, which is the reflected light characteristic map when the overlap error amount is , # represents the maximum number of iterations, and & is at the wavelength of the beam; , a standard deviation obtained by the intensity of the reflected light of different stacking errors and ^, ΔΟΙ7)|>12 〆. 15. The method of designing a registration mark according to claim 12, wherein the average standard deviation is defined by the following two equations in an ellipsometer system: w=Yanshi;)-post^)2/# and de-I) , W) = Yang, where 4 is the starting wavelength of the beam, 4 is the final wavelength of the beam, from the total number of incident wavelength sampling points, # represents the maximum number of iterations, y and △ are functions of the complex parameters And satisfy the contention =tan(w#, where \ and the ruler are the amplitudes of the P-polarized light and the s-polarized light of the diffracted light, y, △%) and Δ(& Δ%) are in the overlapping The reflected light characteristic map with the error amount is, the column (乂/) and the illusion) are the reflected light intensity with different stacking error amounts when the wavelength of the beam is 4, respectively. | Wind Δ (Α人; )|The wind has 2 standard deviations. 16. The method of designing a registration mark according to claim 12, wherein the optimization algorithm is a random walk method. 17. The method of designing a registration mark according to claim 12, wherein the optimization algorithm is a simplex method. PD0086.doc Ι〇〇2ΐ〇 PD〇〇86 °°49653〇8-i