JP2011099822A - Surface inspection method and surface inspection device - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a surface inspection method capable of detecting a focus abnormality distinctively from a dose amount abnormality. <P>SOLUTION: This method includes: an irradiation step (S104) for irradiating a wafer surface having a prescribed repeated pattern with linearly polarized light; a light receiving step (S105) for receiving reflected light from the wafer surface irradiated with the linearly polarized light; a detection step (S106) for detecting a polarization component vertical to a polarization direction of the linearly polarized light in the reflected light, on a conjugate surface to a pupil surface of an objective lens; and an operation step (S107) for determining a line width of the repeated pattern and a focus state during an exposure time from a gradation value of the detected polarization component. In the operation step, the line width is determined from a gradation value on a position in a pupil having high correlation with the line width on the pupil surface, and the focus state is determined from a gradation value on a position in the pupil having high correlation with the focus state. <P>COPYRIGHT: (C)2011,JPO&INPIT

Description

  The present invention relates to a surface inspection method and apparatus capable of detecting line width variation and cross-sectional shape variation in a repetitive pattern of a wafer in a semiconductor manufacturing process.

  In recent years, semiconductor devices tend to have finer patterns in order to increase processing speed, reduce power consumption, and increase storage capacity. At the same time, the detection performance of defects generated in the manufacturing process of semiconductor devices and the management requirement of line width (hereinafter referred to as CD (abbreviation of critical dimension)) are becoming stricter. If the wafer defect or CD value fluctuation that occurred in the exposure process exceeds the allowable value, the relevant wafer or lot is sent to the rework process, and the location where the problem occurs is corrected, and a good semiconductor wafer is obtained. To be able to produce.

  Problems that occur in the exposure process can be broadly classified into those caused by the focus during exposure and those caused by the illumination light quantity (in particular, the irradiation light quantity on the wafer is called the dose amount). When there is a problem with the focus accuracy of the exposure apparatus, the cross-sectional shape of the photoresist pattern formed on the wafer surface changes. For example, a rectangular cross section at the time of best focus expands the skirt of the cross section at the time of defocus. That is, the side wall angle of the pattern cross section (referred to as side wall angle) increases. As a result, it affects the CD value after etching, and in the case of a linear pattern (line and space pattern), it joins with an adjacent pattern after etching (referred to as a bridge, resulting in an electrical circuit short). In the case of a contact hole, a serious problem occurs such that the hole does not reach the lower layer after etching.

  On the other hand, the cause of fluctuations in dose is scan speed fluctuation in a scanning exposure apparatus. Due to the synchronization deviation of the movement speeds of both the wafer and the reticle, the dose amount decreases when the scan speed is relatively high, and the dose amount increases when the scan speed is relatively slow. As a result, if the dose amount is small, the photoresist pattern becomes thick, and if the dose amount is large, the photoresist pattern becomes thin, which directly affects the CD value of the photoresist pattern. Needless to say, the same CD value fluctuation occurs when the dose amount fluctuates due to fogging of the projection lens of the exposure apparatus, not limited to the scanning type exposure apparatus. As a result, after etching, the CD value becomes out of specification or causes disconnection. If these problems are discovered at an early stage and the problems cannot be resolved, many lots will become defective and suffer a large loss.

  Therefore, CD value management is important. In particular, the CD value management of the photoresist pattern after the exposure process is transferred to the rework process when the CD value is out of the standard, and after resist removal, resist coating and exposure are performed again. Is so important.

  For CD value management, a CD-SEM (scanning electron microscope) is widely used, and the CD value can be measured and output quantitatively. However, since the CD-SEM irradiates the wafer with an electron beam, the electron beam causes damage such as shrinking (shrinking) of the photoresist. The damage increases as the acceleration voltage of the electron beam increases to increase the measurement accuracy. In addition, since the measurement time requires several seconds per field of view, several sheets are sampled in one lot. Moreover, only a few points in the exposure shot are measured not for the entire surface of the extracted wafer but for several shots of each extracted wafer.

  As a defect inspection apparatus, there is an apparatus that irradiates a semiconductor wafer with illumination light and receives diffracted light from a repetitive pattern on the wafer. In this apparatus, the illumination light is irradiated to the wafer to receive the diffracted light from the repeated pattern on the wafer, and the diffracted light from the defect has a different CD value, so that the amount of diffracted light is changed, thereby detecting the defect. Do. Such a defect inspection apparatus called a macro inspection apparatus has a high throughput due to batch imaging of the entire surface of the wafer, and is therefore very active in the production line. However, since the diffraction angle of the diffracted light from the repetitive pattern becomes larger as the repetition period becomes smaller, the diffraction angle from the repetitive pattern smaller than the repetitive pitch of about 140 nm is limited by the device configuration such as the illumination wavelength and the optical system arrangement. It is virtually impossible to receive light.

  Therefore, as a new method, a defect inspection apparatus using a structural birefringence effect of a repeated pattern has been put into practical use (see, for example, Patent Document 1). In these macro inspection apparatuses, due to the nature of the apparatus, there is a limit to high resolution inspection of a minute region and quantitative output of CD value fluctuation amount. For example, when receiving diffracted light, the magnitude of the diffracted light intensity is not uniquely related to the CD value of the pattern, so it is impossible to convert the CD value from the diffracted light intensity. Further, when the structural birefringence effect is used, there is a restriction that the defect detection sensitivity may be inferior depending on the state of the pattern, the line width, and the like.

  Whether it is a CD-SEM or an automatic macro inspection device, the biggest problem is the causal retrospective power when an abnormality is detected. When the CD value is measured as a nonstandard value in the CD-SEM, the lot is held, the exposure apparatus that has been exposed is specified, what is the problem of the exposure apparatus, and whether the focus is abnormal. It is necessary to determine whether the dose is abnormal. However, it is extremely difficult to make a judgment based only on the measured CD value. This is because the CD-SEM is measured by irradiating an electron beam from above the pattern, so that the CD value hardly changes at the time of defocusing. Note that it is possible to visually determine the CD-SEM image obtained at the time of CD-SEM measurement and indirectly determine the pattern shape. In the case of defocusing abnormalities, the top and bottom surfaces of the photoresist appear as double circles in the contact holes, or the circles are crooked. In the linear pattern (line and space pattern), the linearity appears to be disturbed from the upper surface, and therefore this judgment is made on the basis of this disorder.

  In recent years, linearity disturbances have been quantified as LWR (Line Width Roughness) or LER (Line Edge Roughness). It became so. LWE and LER measure many CD values at intervals of about 10 nm and define them as standard deviations of the measured values, but it takes a lot of time to measure LWR and LER simultaneously with CD measurement. As described above, the photoresist pattern is damaged.

  For this reason, in CD-SEM, the focus abnormality is not found at the photoresist stage, and only the CD measurement and CD management are performed. For the focus abnormality, a defect is detected as an error if a hole is not opened in the case of a contact hole by using a microscope type micro inspection apparatus after etching, and as an error if it is short in the case of a linear pattern. However, since it is not a photoresist, it cannot be reworked. However, this is because it is better than detecting abnormalities until damage is caused to the photoresist, and rather than indirectly detecting abnormalities in focus from LWE, LER, etc. with high reliability.

JP 2006-343102 A

  Thus, it has become important to distinguish and detect a focus abnormality and a dose abnormality that occur in a composite manner at the stage of the photoresist after exposure.

  The present invention has been made in view of such a problem, and an object of the present invention is to provide a surface inspection method and apparatus capable of distinguishing and detecting a focus abnormality and a dose amount abnormality.

  In order to achieve such an object, a surface inspection method according to the present invention includes an irradiation step of irradiating a surface of a substrate having a predetermined repetitive pattern with linearly polarized light, and reflection from the surface of the substrate irradiated with the linearly polarized light. Light receiving step for receiving light by a light receiving optical system, and polarization substantially perpendicular to the polarization direction of the linearly polarized light in the reflected light received by the light receiving step on the pupil plane of the light receiving optical system or a plane conjugate with the pupil plane A detection step for detecting a component, and a calculation step for obtaining at least one of a line width of the repetitive pattern and a focus state at the time of exposure of the repetitive pattern from information on the polarization component detected in the detection step. In the calculation step, the first specific position having a high correlation with the line width in the pupil plane or a plane conjugate with the pupil plane is used. The line width is obtained from the information of the light component, and the polarization component at the second specific position different from the first specific position having a high correlation with the focus state on the pupil plane or a plane conjugate with the pupil plane. The focus state is obtained from the information.

  In the surface inspection method described above, the polarization component at a plurality of intra-pupil positions set in the pupil plane or a plane conjugate with the pupil plane using the first reference substrate having a known line width. The line width correlation calculating step for obtaining the correlation between the information on the line width and the line width respectively, and the information on the polarization component at the plurality of set positions in the pupil using the second reference substrate whose focus state is known And a focus correlation calculation step for obtaining a correlation between the focus state and the focus state, and among the plurality of pupil positions, the correlation between the polarization component information and the line width is high and the correlation with the focus state is low. A position is specified as the first specific position, and a position in the pupil having a low correlation between the polarization component information and the line width and a high correlation with the focus state is specified as the second specific position. And determining the line width from information of the polarization component at the first specific position specified in the specific step and calculating the second specific position specified in the specific step. It is preferable to obtain the focus state from the information of the polarization component at.

  In the surface inspection method described above, in at least one of the line width correlation calculation step and the focus correlation calculation step, the pupil plane or the entire plane conjugate with the pupil plane is divided into a plurality of sections, and the plurality of pupils are divided. It is preferable to set an inner position and obtain the correlation for each section.

  In the surface inspection method described above, it is preferable that at least one of the first specific position and the second specific position in the specifying step is an in-pupil position showing a correlation having linearity.

  In the surface inspection method described above, the first specific position in the specific step is a position in the pupil corresponding to an irradiation condition in which the irradiation direction of the linearly polarized light is inclined close to 45 degrees with respect to the polarization direction of the linearly polarized light. It is preferable that

  The surface inspection apparatus according to the present invention includes an irradiation unit that irradiates a surface of a substrate having a predetermined repetitive pattern with linearly polarized light, and a light receiving optical device that receives reflected light from the surface of the substrate irradiated with the linearly polarized light. And a detection unit for detecting a polarization component substantially perpendicular to the polarization direction of the linearly polarized light in the reflected light received by the light receiving optical system on a pupil plane of the light receiving optical system or a plane conjugate with the pupil plane. A calculation unit that obtains at least one of a line width of the repetitive pattern and a focus state at the time of exposure of the repetitive pattern from the information of the polarization component detected by the detection unit, The line width is obtained from the information of the polarization component at the first specific position where the correlation with the line width is high in the pupil plane or a plane conjugate with the pupil plane, and is shared with the pupil plane or the pupil plane. Determining the focus state from the information of the polarization component in the different second specific position to the high correlation of the first specific position of the focus state in a plane.

  In the above-described surface inspection apparatus, it is preferable that the imaging device is provided on a pupil plane of the light receiving optical system or a plane conjugate with the pupil plane.

  According to the present invention, the line width can be measured quantitatively, and the focus abnormality and the dose amount abnormality can be distinguished and detected.

It is a flowchart which shows a surface inspection method. It is a figure which shows the whole structure of a surface inspection apparatus. (A) is a perspective view of a repeating pattern, (b) is a top view of a repeating pattern. It is sectional drawing of the linear pattern which changed exposure conditions. (A), (b) is a figure which shows an example of a polarization state. It is a figure which shows the pupil image of an objective lens. It is a figure which shows the irradiation conditions which obtain the reflected light corresponding to the point A in FIG. It is a figure which shows the irradiation conditions which obtain the reflected light corresponding to the point B in FIG. It is a figure which shows the irradiation conditions which obtain the reflected light corresponding to the point C in FIG. It is a figure which shows the irradiation conditions which obtain the reflected light corresponding to the point D in FIG. It is a figure which shows the irradiation conditions which obtain the reflected light corresponding to the point E in FIG. It is a figure which shows the irradiation conditions which obtain the reflected light corresponding to the point F in FIG. It is a figure which shows the case where a focus abnormality generate | occur | produces in FIG. It is a figure which shows the case where a pattern becomes rough in FIG. It is a figure which shows the actual pupil image detected with the image pick-up element. It is a figure which shows the state which divided the pupil image into the area | region. It is a figure which shows a pupil CD map. It is a figure which shows CD reference | standard wafer. It is a figure which shows an example of the gradation value of CD reference | standard wafer measured with the surface inspection apparatus. It is a figure which shows the correlation with conversion CD value and CD measurement value. It is a figure which shows an example of the gradation value of CD reference | standard wafer measured with the surface inspection apparatus. It is a figure which shows a focus reference | standard wafer. It is a figure which shows a pupil focus map. It is the table | surface which put together the presence or absence of the sensitivity in the position in a pupil. It is a figure which shows an example of the gradation value of the focus reference | standard wafer measured with the surface inspection apparatus. It is a figure which shows an example of the gradation value of the focus reference | standard wafer measured with the surface inspection apparatus. It is a figure which shows the correlation with the measured value of a side wall angle, and the gradation value measured with the surface inspection apparatus. It is a figure which shows the histogram of the gradation value obtained from the surface inspection apparatus, and the histogram of the roughness degree obtained from the electron microscope. (A) is a figure which shows the pupil CD map at the time of changing illumination wavelength, (b) is a figure which shows the pupil focus map at the time of changing illumination wavelength. It is the table | surface which put together the presence or absence of the sensitivity in the position in a pupil at the time of changing illumination wavelength. (A) is a diagram showing a pupil CD map when the pattern azimuth is changed, (b) is a diagram showing a pupil focus map when the pattern azimuth is changed, and (c) is a pupil shown in FIG. It is the figure which cut off a part of CD map, (d) is the figure which cut off a part of pupil focus map of FIG. It is a figure which shows the modification of a surface inspection apparatus.

  Hereinafter, preferred embodiments of the present invention will be described with reference to the drawings. A surface inspection apparatus according to this embodiment is shown in FIG. 2, and the surface inspection apparatus 1 includes a stage 6, an objective lens 7, a prism 8, an illumination optical system 10, a detection optical system 15, and an arithmetic processing unit. 20 as a main component. A semiconductor wafer 5 (hereinafter simply referred to as a wafer 5 as appropriate) is carried from a wafer cassette or developing device (not shown) by a transport system (not shown) after exposure and development of the photoresist by an exposure device (not shown). The pattern (repetitive pattern) is placed on the stage 6 with the formation surface facing up.

  The stage 6 securely holds the wafer 5 placed on the stage 6 by vacuum suction or the like. The stage 6 is configured to be movable in the XY directions along the surface of the wafer 5, and is configured to be rotatable about the Z axis. The direction perpendicular to the paper surface of FIG. 2 is the X axis, the horizontal direction in FIG. 2 is the Y axis, and the vertical direction in FIG. 2 is the Z axis. Further, a rotation angle around the Z axis in the XY plane is referred to as an azimuth angle.

  The illumination optical system 10 includes a light source 11, a condensing lens 12, a wavelength selection filter 13, and a first polarizing filter 14 in order of arrangement from the right side to the left side in FIG. As the light source 11, a mercury lamp or the like is used. The mercury lamp has a plurality of wavelengths (hereinafter may be abbreviated as λ), for example, e-line (λ = 546 nm), g-line (λ = 436 nm), h-line (λ = 405 nm), j-line (λ = 313 nm), and light around λ = 250 nm is generated. The wavelength selection filter 13 is used to select only light having a specific wavelength from among the light having a plurality of wavelengths.

  The light source 11 may be a light source having a wide wavelength band such as a halogen lamp or a blue excitation type LED. Further, since the light from the halogen lamp or the blue excitation type LED extends over a wide wavelength range, the wavelength selection filter 13 may be, for example, a filter that transmits blue, green, and red. Therefore, it is preferable that the wavelength selection filter 13 has a configuration in which the wavelength and the wavelength band can be selectively changed by inserting and removing from the optical path as a plurality of filters having different transmission wavelength bands.

  The light emitted from the light source 11 passes through the condenser lens 12 and the wavelength selection filter 13 and then passes through the first polarizing filter 14. The first polarizing filter 14 is arranged so that the transmitted light is polarized in the X direction, that is, the polarization direction P of linearly polarized light obtained by transmitting through the first polarizing filter 14 is in the X direction.

  The light transmitted through the first polarizing filter 14 is reflected downward by the prism 8 to become parallel incident light I directed toward the wafer 5, and is irradiated onto the wafer 5 through the objective lens 7. Reflected light from the wafer 5 becomes parallel reflected light J through the objective lens 7, passes through the prism 8, and reaches the detection optical system 15. The detection optical system 15 includes a second polarizing filter 16, a relay lens 17, and an imaging element 18 such as a two-dimensional CCD in the order of arrangement from the lower side to the upper side in FIG. 2. The reflected light that has passed through the prism 8 passes through the second polarizing filter 16, is extended or enlarged (or reduced) by the relay lens 17, and then enters the image sensor 18. The reflected light incident on the image sensor 18 is photoelectrically converted into an electrical signal by the image sensor 18, and a detection signal of the reflected light is output to the arithmetic processing unit 20.

  The second polarizing filter 16 is arranged so that the transmitted light is polarized in the Y direction, that is, the polarization direction Q of linearly polarized light obtained by transmitting through the second polarizing filter 16 is in the Y direction. Such a state in which the polarization directions are orthogonal is called crossed Nicol, and the polarized light component perpendicular to the polarization direction of the incident light I (linearly polarized light) out of the reflected light J transmitted through the objective lens 7 and the prism 8 is shown. It is detected by the image sensor 18 (detection optical system 15). The first polarizing filter 14 on the incident side may be referred to as a polarizer, and the second polarizing filter 16 on the light receiving side may be referred to as an analyzer.

  The image sensor 18 is provided at a position conjugate with the pupil plane (not shown) of the objective lens 7. In addition, it is also possible to make the wavelength selection filter 13 unnecessary by using a color CCD for the image sensor 18 and using respective outputs of red (R), green (G), and blue (B).

  Further, although the incident light I and the reflected light J are on the same optical axis, if attention is paid to a part of the incident light beam I1 of the incident light I, the incident light I is incident on the wafer 5 at an incident angle K and is imaged as a reflected light beam J1. The element 18 is reached. In other words, the position 19 on the image sensor 18 receives only the reflected light information from the incident light beam I1.

  Here, the structural birefringence effect in the repeated pattern will be described with reference to FIG. Details of the structural birefringence are described in detail in "Principle of optics" by M, Boron & E. Wolf, but here it is cited from JP-A-9-145921. On the surface of the wafer 5, as shown in FIG. 3A, a repetitive pattern 30 that is a linear pattern (line and space pattern) is formed.

In the repetitive pattern 30 as shown in FIG. 3A, the line width of the line portion 31 is t ₁ , the refractive index of the line portion 31 is n ₁ , the line width of the space portion 32 is t ₂ , and the space portion 32 The refractive index of n is n ₂ . Then, the pattern pitch is t ₁ + t ₂ . This repeated pattern 30 can be regarded as a uniaxial crystal having a first material (line portion) 31 and a second material (space portion) 32 that are different from each other, and as shown in FIG. the refractive index and n _0, the refractive index for the light vibrating perpendicular to the boundary surface as n _e, can handle birefringence for light vibrating parallel to. Light 33 enters perpendicularly to the repetitive pattern 30 of FIG. 3A, and the polarization direction (vibration direction) of the light 33 is the direction R (parallel to the X axis) in FIG. 3B according to the configuration of FIG. When the repetitive pattern 30 is inclined at a pattern azimuth angle θ (the azimuth angle θ is an angle formed by the X axis and the line portion 31) with respect to the X axis, the direction R is the boundary plane parallel direction component R ₀ . it is possible to decompose the interface vertical component R _e, respectively the refractive index n ₀ and the refractive index n _e, it is given by the following equations (1) and (2) below.

However, f ₁ and f ₂ in the expressions (1) and (2) are given by the following expressions (3) and (4), respectively.

Further, the phase difference of the reflected light to the incident light due to the form birefringence and [delta], when the height of the line portion 31 and is h, the phase difference [delta], of the following with reference to the refractive index n ₀ and the refractive index n _e It is given by equation (5).

  In the exposed wafer 5, the line portion 31 is a photoresist, and the space portion 32 is a base (for example, an oxide film). Although the thickness of the photoresist depends on the light source wavelength of the exposure apparatus, the thickness of the resist for the ArF light source used in the most advanced exposure apparatus is 100 nm or less.

  When linearly polarized light is incident, when the phase difference δ satisfies the condition of (m + 1/4) λ (m = 0, 1, 2,...), It acts like a quarter wave plate, so that the reflected light is circular. If the phase difference δ satisfies the condition of (m + 1/2) λ (m = 0, 1, 2,...), It acts like a half-wave plate, so that the reflected light has a polarization direction of 90 degrees. It becomes a rotated linearly polarized light. In addition, under the intermediate condition between the two cases, the reflected light is elliptically polarized light.

  FIG. 4A shows a cross section of the repeated pattern 40 exposed under normal conditions. As shown in FIG. 4B, the line width of the line portion 43 when the dose amount is large is narrower than that of the normal line portion 41. On the other hand, as shown in FIG. 4C, the line width of the line portion 45 when the dose amount is small is thicker than that of the normal line portion 41. As shown in FIG. 4D, the shape of the line portion 47 when exposed in the defocused state is a cross-sectional shape with a skirt (trapezoid), and is the inclination angle of the side wall of the line portion 47. Side wall angle α increases.

According to the above equation (5), it is the line width and height of the repetitive pattern that determines the polarization state of the reflected light. In the semiconductor wafer exposure process, the period (pitch) of the repetitive pattern is strictly constant. As shown in FIG. 4B, when the dose amount is excessive, the pattern height does not change, the line width t ₁ of the line portion 43 increases, and the line width t ₂ of the space portion 44 decreases. As a result, the refractive index n ₀ and the refractive index n _e, respectively changed, the phase difference is generated in the reflected light. On the other hand, in the pattern cross-sectional state at the time of defocus exposure as shown in FIG. 4D, the line width is not constant in the height direction, and the line width decreases in proportion to the height of the line portion 47. may be considered to have a (5) in the formula, a function of line width t ₂ is the height h of the line width t ₁ and the space portion 44 of the line section 43.

  When the change in the pattern cross-sectional state with respect to the normal pattern in FIG. 4A is as shown in FIG. 4B, FIG. 4C, or FIG. The change in the polarization state of the reflected light is extracted as a change in only the polarization component in the Y direction by the second polarization filter 16 (see FIG. 2) arranged in crossed Nicols so that only the polarization direction Q in the Y direction is transmitted. The For example, as shown in FIG. 5A, it is assumed that the polarization state of the reflected light accompanying the pattern cross-sectional state change is the first polarization state 50, the second polarization state 51, and the third polarization state 52. . Each of the polarization states 50, 51, and 52 is elliptically polarized light, and the direction of the major axis and the ellipticity change, but the transmitted light obtained by transmitting through the second polarizing filter 16 is in the Y direction. Only the first light amount 50a, the second light amount 51b, and the third light amount 52c of the component are received by the image sensor 18 (accurately, since the Y direction component in FIG. 5A is an amplitude, the light amount is It corresponds to the square of the amplitude, but is shown here in a simplified diagram).

  Further, for example, as shown in FIG. 5B, the polarization state of the reflected light accompanying the pattern cross-sectional state change is the fourth polarization state 53, the fifth polarization state 54, and the sixth polarization state 55. Suppose it was a thing. Each of the polarization states 53, 54, and 55 is elliptically polarized light that is close to circularly polarized light. The result is that the repetitive pattern 30 (see FIG. 3) is close to circularly polarized light by giving a phase difference δ to the incident linearly polarized light like a quarter wave plate. Although the axial direction is greatly different, the transmitted light obtained by transmitting through the second polarizing filter 16 is the same as the fourth light amount 53a of the Y direction component in any of the polarization states 53, 54, and 55. There is no big change, and the polarization state does not seem to change.

  5A and 5B are merely examples, and the polarization state of the reflected light from the repetitive pattern 30 when the pattern cross-sectional state such as line width, height, shape, etc. changes is elliptically polarized light. The ellipticity, major axis direction, and size of the light beam change, and the way the polarization state changes depends on the irradiation conditions. The description given with reference to FIGS. 3 to 5 is a case where light is incident on the repetitive pattern 30 perpendicularly, but a general description will be given next, including the case where it is incident obliquely.

  As shown in FIG. 2, the incident light I irradiated through the objective lens 7 is a bundle of light incident from various incident angles and directions, and thus is provided at a position conjugate with the pupil plane of the objective lens 7. The image sensor 18 detects a collection of reflected light based on these various incident conditions. FIG. 6 shows a pupil image 60 of the objective lens 7 detected by the image sensor 18. The coordinates Px and Py on the pupil plane are the same as the XYZ coordinate system of FIGS. An intersection point O between the Px axis and the Py axis is the optical axis of the objective lens 7.

  Here, the irradiation conditions in the present embodiment will be described. Irradiation conditions include the incident angle of incident light, the polarization direction with respect to the incident surface, the azimuth angle of the incident surface, the azimuth angle formed between the incident surface and the repeated pattern direction, and the like. In the pupil plane coordinate system (Px, Py) shown in FIG. 6, the points B, C, D, E, and F on the circumference of a circle 61 having a radius r centered on the optical axis O are all The reflected light by the light which entered with the same incident angle (this is made into the incident angle i) is shown.

  FIG. 7 shows irradiation conditions for obtaining reflected light corresponding to the point A on the optical axis in the pupil image 60 of FIG. 7 to 14, the same members as those shown in FIG. 3 are denoted by the same reference numerals, and the repetitive pattern 30 that is a linear pattern (line and space pattern) is a line that is a photoresist. It consists of a portion 31 and a space portion 32 which is a base layer such as an oxide film. Further, the line portion 31 (and the space portion 32) is inclined by the azimuth angle θ with respect to the X axis, as in FIG. In FIG. 7, incident light 70 a that is linearly polarized light is incident with a polarization direction 71 a perpendicular to the repetitive pattern 30 (Z-axis direction). On the other hand, the reflected light 72a is reflected perpendicularly to the repetitive pattern 30 with, for example, a polarization state 73a.

FIG. 8 shows irradiation conditions for obtaining reflected light corresponding to the point B in the pupil image 60 of FIG. The azimuth angle θ with respect to the X axis of the line portion 31 (and the space portion 32) is the same as that in FIG. Incident light 70b in FIG. 8 is incident at an incident angle i. An incident surface (a plane between the normal line Z of the repetitive pattern 30 and the incident light 70b) 74b coincides with the YZ plane. The polarization direction 71b of the incident light 70b is inclined by an angle φ _B (hereinafter referred to as the polarization angle φ in the present embodiment) with respect to a straight line 75b included in the incident surface 74b and perpendicular to the incident light 70b. The size of φ _B is 90 degrees. That is, it is incident in the S polarization state. This can be easily understood from the fact that the polarization direction 71b is parallel to the X axis. Needless to say, the reflected light 72b is also included in the incident surface 74b.

FIG. 9 shows irradiation conditions for obtaining reflected light corresponding to the point C in the pupil image 60 of FIG. The azimuth angle θ with respect to the X axis of the line portion 31 (and the space portion 32) is the same as that in FIG. The incident light 70c in FIG. 9 is incident at an incident angle i, and forms an incident surface (a plane between the normal line Z of the repetitive pattern 30 and the incident light 70c) 74c. The polarization direction 71c of the incident light 70c is inclined by an angle φ _{C with} respect to a straight line 75c included in the incident surface 74c and perpendicular to the incident light 70c, and the magnitude of the polarization angle φ _C is between 0 degree and 90 degrees. It is.

FIG. 10 shows irradiation conditions for obtaining reflected light corresponding to the point D in the pupil image 60 of FIG. The azimuth angle θ with respect to the X axis of the line portion 31 (and the space portion 32) is the same as that in FIG. The incident light 70d in FIG. 10 is incident at an incident angle i, and forms an incident surface (a plane between the normal line Z of the repetitive pattern 30 and the incident light 70d) 74d. The polarization direction 71d of the incident light 70d is inclined by an angle φ _{D with} respect to a straight line 75d included in the incident surface 74d and perpendicular to the incident light 70d, and the magnitude of the polarization angle φ _D is between 0 and 90 degrees. It is. Note that the polarization angle φ _D in FIG. 10 is closer to 0 degrees compared to the polarization angle φ _C in FIG. 9, and 0 degrees <φ _D <φ _C <90 degrees.

FIG. 11 shows irradiation conditions for obtaining reflected light corresponding to the point E in the pupil image 60 of FIG. The azimuth angle θ with respect to the X axis of the line portion 31 (and the space portion 32) is the same as that in FIG. Incident light 70e in FIG. 11 is incident at an incident angle i. An incident surface (a plane on which the normal line Z of the repetitive pattern 30 and the incident light 70e come off) 74e coincides with the XZ plane. The polarization direction 71e of the incident light 70e is a direction in which the polarization angle φ _E with respect to the straight line 75e included in the incident surface 74e and perpendicular to the incident light 70e is 0 degree. That is, it is incident in the P-polarized state. This can be easily understood from the fact that the polarization direction 71e is parallel to the X axis.

FIG. 12 shows irradiation conditions for obtaining reflected light corresponding to the point F in the pupil image 60 of FIG. The azimuth angle θ with respect to the X axis of the line portion 31 (and the space portion 32) is the same as that in FIG. The incident light 70f in FIG. 11 is incident at an incident angle i, and forms an incident surface (a plane between the normal line Z of the repetitive pattern 30 and the incident light 70f) 74f. The polarization direction 71f of the incident light 70f is inclined by an angle φ _{F with} respect to a straight line 75f included in the incident surface 74f and perpendicular to the incident light 70f, and the magnitude of the polarization angle φ _F is between 90 degrees and 180 degrees. It is.

The points A, B, C, D, E, and F in the pupil image 60 of FIG. 6 are respectively reflected light 72a, 72b, 72c, 72d, 72e, and 72f obtained under the irradiation conditions shown in FIGS. It is. Therefore, the irradiation conditions shown in FIG. 9 will be described in more detail as an example. Even if the incident angle i is the same, if the azimuth angle formed by the incident surface 74c and the XZ plane (this is referred to as the incident surface azimuth angle ω) is different, the polarization angle φ _C of the incident light 70c incident as linearly polarized light is different. . Even if the deflection angle φ _C is the same, if the azimuth angle θ (referred to as the pattern azimuth angle θ) of the line portion 31 (and the space portion 32) of the repeated pattern 30 with respect to the X axis is different, the line portion 31 ( And the angle of the polarization direction 71c with respect to the space portion 32) is different.

  The height h in the equation (5) is the height of the line portion 31 in the light traveling direction. At normal incidence, the height h shown in FIG. 7 is the same as the height h in the equation (5). However, under the oblique incidence conditions shown in FIGS. 8 to 12, the direction of each incident light 70b, 70c, 70d, 70e, and 70f due to the oblique incidence is the light traveling direction. For example, in FIG. 9, the height corresponds to the vector 79c when the incident light 70c is a vector. In the present embodiment, the height corresponding to the vector 79c is referred to as a pattern effective height hi. That is, the pattern effective height hi corresponds to the height h in the equation (5). Therefore, even if the height h is the same in the Z-axis direction, if the incident angle i, the incident surface azimuth angle ω, and the pattern azimuth angle θ are different, the phase difference δ expressed by the equation (5) is different.

In general, since the refractive index depends on the wavelength, the refractive index n ₁ of the line portion 31 and the refractive index n ₂ of the space portion 32 in the equations (1) and (2) differ depending on the wavelength λ of the incident light. That is, when a linear pattern (line and space pattern) is irradiated with light, the phase difference δ given by the equations (1) to (5) is equal to the incident angle i and the incident surface due to structural birefringence. It is determined by the azimuth angle ω and the pattern azimuth angle θ. That is, the phase difference δ is a line determined by the polarization direction represented by the polarization angle φ, the pattern effective height hi determined by the incident angle i, the incident surface azimuth angle ω, and the pattern azimuth angle θ, and the wavelength λ. It varies depending on the refractive index n ₂ of the refractive index n ₁ and the space portion 32 of the part 31. These parameters are related to the equation (5).

Next, the phase state in the pattern cross-sectional state in which the side wall angle α is increased as shown in FIG. 4D will be considered with reference to FIG. In FIG. 13, parameters such as the incident angle i are the same as those in FIG. 9, and the same members as those in FIG. As shown in FIG. 13, when the line section 31 is trapezoidal cross section, at normal incidence, the line width t ₂ of a line width t ₁ and the space portion 32 of the line section 31 is a function of the height h just as was considered, it handled the line width t ₂ of a line width t ₁ and the space portion 32 of the line section 31 as a function of the pattern effective height hi.

  When a focus abnormality occurs in the exposure process, a skirted (trapezoidal) cross-sectional shape is formed as described above, the side wall angle α is increased, and the pattern shape is as shown in FIG. May be rough. The polarization state of the reflected light 72c from the rough line portion 31 changes depending on the defocus amount. In FIG. 14, parameters such as the incident angle i are the same as those in FIG. 9, and the same members as those in FIG.

As described above, when the linear pattern (line and space pattern) is irradiated with light, the polarization state of the reflected light is 1) the incident angle i, the incident surface azimuth angle ω, and the pattern azimuth angle θ. Illumination, such as the polarization direction of the incident light determined and represented by the polarization angle φ, 2) the incident angle i, the incident plane azimuth angle ω, the pattern effective height hi determined by the pattern azimuth angle θ, and 3) the wavelength λ. Varies depending on conditions. Furthermore, 4) a line width t _2, 6 of the refractive index n _2, 5) the line width of the line portion 31 t ₁ and the space portion 32 of the refractive index n ₁ and the space portion 32 of the line section 31 which is determined by the wavelength lambda) It depends on the pattern condition and material condition such as side wall, angle α, etc.

Therefore, by selecting the irradiation condition, it is possible to detect the fluctuation of the line width t ₁ of the line portion 31 and the fluctuation of the side wall angle α, and further, if the irradiation condition is determined optimally, the line portion It is possible to detect the change in the line width t ₁ of 31 and the change in the side wall angle α separately. Since the line width t ₁ and the side wall angle α of the line portion 31 are directly related to the dose amount change and the focus change at the time of wafer exposure, according to the present embodiment, the dose amount change and the focus change are changed. It becomes possible to detect by dividing.

  Selection and determination of the optimum irradiation condition is performed by analysis in the pupil. FIG. 15 shows an actual pupil image 150 detected by the image sensor 18 in the pupil coordinate system. FIG. 15 shows the intensity distribution of light on the pupil plane, and shows the amount of light transmitted through the second polarizing filter 16. In FIG. 15, a strong portion (high gradation value portion) of the optical signal detected by the image sensor 18 is set to a dark color (black), and a weak light signal portion (low gradation value portion) is set to a light color. It was set as (white) and displayed as the gradation legend 151. The pupil coordinate system in FIG. 15 is the same pupil coordinate system (Px, Py) as in FIG. 6, and the intersection O between the coordinate axis Px and the coordinate axis Py is the optical axis. A point C in the pupil image 60 of FIG. 6, that is, a place where the reflected light is obtained at the pupil position that gives the irradiation state and pattern state shown in FIGS. 9, 13, and 14 is the pupil image 150 (pupil plane) of FIG. This is the intra-pupil position 150c.

  The pupil image 160 in FIG. 16 is the same as the pupil image 150 in FIG. In FIG. 16, one section when the pupil image 160 is divided into M columns × N rows is referred to as an intra-pupil position, and FIG. 16 shows an intra-pupil position 161 as an example. M and N are natural numbers, preferably about 10 to 100. Usually, M = N. The gradation value at the position in the pupil can be extracted and obtained by selecting a pixel of the image sensor 18.

  FIG. 17F is a map called a pupil CD map. A pupil image corresponding to the pupil image 160 of FIG. The horizontal and vertical directions of the pupil CD map correspond to the pupil coordinates (Px, Py), respectively, and the center of the circle 170 corresponds to the pupil coordinates (Px, Py) = (0, 0). The pupil CD map does not indicate the gradation in the pupil. In each pupil position divided into M columns × N rows in FIG. 16, the gradation value at the position in the pupil is separately measured along the vertical axis. It is an aggregate of bivariate correlation diagrams with the CD value represented on the horizontal axis. For example, enlarged bivariate correlation diagrams at five positions 171a, 171b, 171c, 171d, and 171e in the pupil CD map are shown in FIGS. 17 (a), 17 (b), and 17 (c). ), FIG. 17 (d), and FIG. 17 (e). 17 (a), 17 (b), 17 (c), 17 (d), and 17 (e) indicate the in-pupil positions 161a, 161b, 161c, 161d, and 161e in FIG. 16, respectively. Is the gradation value of the image sensor 18. The pupil CD map shown in FIG. 17F is obtained by measuring a CD reference wafer whose CD value is known. The method will be described next.

  First, CD value measurement will be described. 17 (a), 17 (b), 17 (c), 17 (d), and 17 (e), the horizontal axis (that is, the measured CD value) represents a different CD value (CD The value can be obtained by measuring a wafer having a known value (CD reference wafer).

  FIG. 18 shows a CD reference wafer 180 for obtaining a pupil CD map. A total of 40 shots of 5 rows and 8 columns formed on the surface of the CD reference wafer 180 are obtained by exposing a linear pattern (line and space pattern) while changing the dose amount. The numbers shown in each shot in FIG. 18 represent the magnitude of the dose amount, the number “0” is a shot exposed with the standard dose amount, the number “+1” is one step higher, and the number “−1” is 1. This is a shot exposed with a small amount of dose. For example, the number “+3” shot 181 on the right end is a shot that is exposed with a dose amount that is three steps higher, and therefore, when a negative type resist is used, the line width of the line portion of the pattern should be the narrowest. Thus, the CD reference wafer 180 is a wafer including various CD values.

  In the CD value measurement, the CD value of the linear pattern (line and space pattern) on the CD reference wafer 180 is measured by a CD measuring machine such as a CD-SEM (not shown) separately from the surface inspection apparatus 1 of the present embodiment. Measure. In order to obtain a preferable number of data for statistical processing, it is preferable to measure a large number of points in a shot in a CD-SEM or the like. For example, by measuring 25 points in a shot, 25 points × 40 shots = 1000 Get the measurement data.

  Next, the gradation value measurement will be described. The same location as the measurement point of the CD reference wafer 180 measured by a CD-SEM or the like is measured using the surface inspection apparatus 1 of the present embodiment. In the above example, 1000 pupil images are captured and acquired by the image sensor 18 for 25 points in shot × 40 shots = 1000 points. From the 1000 pupil images obtained in this way, the arithmetic processing unit 20 is used to obtain gradation values for each position in the pupil divided into M columns × N rows.

  Then, with respect to all measurement points (1000 points in the above example) of the CD reference wafer 180, for each position (section) in the pupil of M columns × N rows (47 columns × 47 columns in the present embodiment), Bivariate correlation diagrams (horizontal axis = measured CD value, vertical axis = tone value at the position in the pupil) are arranged in an array according to pupil coordinates (Px, Py). Thereby, a pupil CD map (a map showing the relationship between the CD change and the luminance change for each position of the pupil image) as shown in FIG. 17 can be obtained.

  The scale of the horizontal axis of Fig.17 (a)-(e) is a common scale. On the other hand, the vertical axis of FIGS. 17 (a) to 17 (e) is displayed with an auto scale (automatically scales from the minimum to the maximum value range) because the gradation value varies depending on the position in the pupil. Yes. Therefore, in FIG. 17C, the gradation value has a mountain-shaped tendency with respect to the CD value, but the change amount of the gradation value with respect to the CD variation is small, and it can be said that the gradation value is hardly sensitive to the CD value. . In the pupil CD map of FIG. 17F, all the bivariate correlation diagrams for each position in the pupil are displayed in an auto scale. At the position in the pupil where the gradation change with respect to the CD change is small, as shown in FIG. 17C, the plot of the bivariate correlation diagram is enlarged in the vertical axis direction (a graph having a width in the vertical axis direction). . It can be seen from the pupil CD map in FIG. 17F that such places are distributed in the vicinity of the Px axis, the Py axis, and the pupil center in the pupil. On the other hand, at the upper left pupil positions 171a and 171b, the gradation value tends to increase as the CD value increases. Conversely, at the lower left pupil positions 171d and 171e, the gradation value increases as the CD value increases. Shows a tendency to decrease.

  As described with reference to FIGS. 7 to 12, the irradiation conditions differ depending on the position in the pupil. Therefore, as described in FIGS. 5A and 5B, the polarization state of the reflected light varies depending on the position in the pupil. For example, polarized light whose gradation value (corresponding to the Y-direction component of elliptically polarized light (a component that passes through the second polarizing filter 16)) detected by the image sensor 18 increases when the CD value increases. If there is an in-pupil position that gives a state change (the in-pupil position is a part of the irradiation condition), there is also an in-pupil position that gives a change in polarization state such that the gradation value becomes small. In addition, a position in the pupil that gives a change in polarization state that increases the gradation value when the CD value increases (this is referred to as a positive change position), and a change in polarization state that decreases the gradation value. Between the intra-pupil position (referred to as a minus change position), there is an area where the gradation value does not change even if the CD value changes.

  As described above, the pupil image 150 in FIG. 15, that is, the pupil image 160 in FIG. 16, includes locations that are sensitive to CD value changes such as the left and upper left pupil positions 171 a, 171 b, 171 d, and 171 e It is clear from the pupil CD map in FIG. 17F that there is a place where the sensitivity to the CD value change is weak, such as the in-pupil position 171c (near the Px axis). Theoretically, when a high NA objective lens is used, there are positive change positions (positive correlation) and negative change positions (negative correlation) in the pupil position of the objective lens. In the meantime, the position in the pupil is weakly sensitive to changes in the CD value. In the present embodiment, the NA (numerical aperture) of the objective lens 7 is NA = 0.9 or a high NA close thereto. Note that light is received at NA = 0.01 to 0.1 at one intra-pupil position divided into M columns × N rows.

  The gradation value (corresponding to the Y-direction component of elliptically polarized light (the component transmitted through the second polarizing filter 16)) does not always change linearly with respect to the CD value change (this is illustrated in FIG. 5). (It is clear from the explanation in (a)). At intra-pupil positions 171a and 171e in the pupil CD map of FIG. 17 (f), as shown in FIGS. 17 (a) and 17 (e), the gradation value changes nonlinearly with respect to the CD value change. . Many of the locations in the pupil that obtain the positive change position and the negative change position show non-linear changes, and fewer show linear changes. In the pupil positions 171b and 171d in the pupil CD map of FIG. 17 (f), as shown in FIGS. 17 (b) and 17 (d), the gradation value (vertical axis) with respect to the CD value change (horizontal axis). ) Changes almost linearly.

  FIG. 19 shows the gradation value at the in-pupil position 171b of FIG. 17F when the CD reference wafer 180 of FIG. 18 is measured by the surface inspection apparatus 1 of this embodiment, and is the same as FIG. The coordinates on the wafer surface are shown. In FIG. 19, a portion with a high gradation value is set to a dark color (black), and a portion with a low gradation value is set to a light color (white), and displayed as a gradation legend 191. As can be seen from FIG. 19, shots that are exposed by changing the dose amount on the CD reference wafer 180, that is, shots having different line widths, are clearly detected as shots having a gradation difference.

As can be seen from FIG. 17B, the CD measurement value and the gradation value at the upper left pupil position 171b are highly correlated. When the gradation value of the in-pupil position 171b is GL _B and the converted CD value converted from the gradation value GL _B is CD _GLB , the converted CD value CD _GLB is expressed by the following equation (6).

As described above, since the relationship between the CD measurement value and the gradation value is basically non-linear, the equation (6) is a cubic polynomial. FIG. 20 shows the correlation between the converted CD value CD _GLB obtained using the equation (6) and the CD measurement value (represented as CD _{S in} FIG. 20) measured by a CD measuring machine such as a CD-SEM. Is. Since the error of the converted CD value CD _GLB in FIG. 20 is 0.01 and the standard value of the CD value of the line portion is normalized to 1, this indicates that the error of 0.01 is an error of 1%.

  FIG. 21 shows the gradation value at the in-pupil position 171c of FIG. 17 (f) when the CD reference wafer 180 of FIG. 18 is measured by the surface inspection apparatus 1 of this embodiment, and is the same as FIG. The coordinates on the wafer surface are shown. In FIG. 21, a portion with a high gradation value is set to a dark color (black), and a portion with a low gradation value is set to a light color (white), and displayed as a gradation legend 211. As can be seen from FIG. 21, shots that are exposed by changing the dose amount on the CD reference wafer 180, that is, shots having different line widths, cannot be detected as shots having gradation differences. That is, it is hardly sensitive to the CD value.

  Next, the tendency of the position in the pupil with respect to the pattern cross-sectional state represented by the side wall angle α will be examined using the focus reference wafer exposed by changing the focus condition for each shot.

  FIG. 22 shows the focus reference wafer 220. A total of 40 shots of 5 rows and 8 columns formed on the surface of the focus reference wafer 220 are obtained by exposing a linear pattern (line and space pattern) while changing the focus amount. The numbers shown in each shot in FIG. 22 represent the amount of focus, the number “0” is a shot exposed at the central focus value near the best focus, and the number “+1” is decremented by one step plus side. It is a shot that is exposed with focus (deliberately defocusing), and the number “−1” is a shot that is defocused to the negative side by one step and exposed. For example, since the number “+3” shot 221 at the right end is a shot that is defocused and exposed to the three-step plus side, the cross-sectional shape of the line portion of the pattern is in the state where the skirt is pulled out the most (see FIG. 4D). The side wall angle α should be large). At the same time, the pattern shape is often rough as shown in FIG. As described above, the focus reference wafer 220 is a wafer including cross-sectional shapes and pattern states of various line portions. The focus reference wafer 220 has the same shot arrangement as the CD reference wafer 180 shown in FIG. 18, but there is no particular reason for the arrangement, and the same process (for example, the gate line exposure process or the wiring exposure process). Etc.).

  FIG. 23F is a map called a pupil focus map. A pupil image corresponding to the pupil image 160 of FIG. The horizontal direction and vertical direction of the pupil focus map respectively correspond to the pupil coordinates (Px, Py), and the center of the circle 230 corresponds to the pupil coordinates (Px, Py) = (0, 0). The pupil focus map does not indicate the gradation in the pupil. In each pupil position divided into M columns × N rows in FIG. 16, the gradation value at the position in the pupil is plotted on the vertical axis during exposure. Is a collection of bivariate correlation diagrams with the amount of focus on the horizontal axis. For example, enlarged bivariate correlation diagrams at five intra-pupil positions 231a, 231b, 231c, 231d, and 231e of the pupil focus map are shown in FIGS. 23 (a), 23 (b), and 23 (c). ), FIG. 23 (d), and FIG. 23 (e). 23 (a), FIG. 23 (b), FIG. 23 (c), FIG. 23 (d), and FIG. 23 (e) are respectively in-pupil positions 161a, 161b, 161c, 161d, and 161e in FIG. Is the gradation value of the image sensor 18. The pupil focus map shown in FIG. 23F is obtained by measuring the focus reference wafer 220 with the focus amount varied during exposure. The pattern azimuth angle θ is the same azimuth as when the pupil CD map of FIG.

  The scales on the horizontal axis in FIGS. 23A to 23E are a common scale. On the other hand, the vertical axis in FIGS. 23 (a) to 23 (e) is displayed with an auto scale (automatically scales from the minimum to the maximum value range) because the gradation value varies depending on the position in the pupil. Yes. Therefore, in FIG. 23B and FIG. 23D, it can be said that the change amount of the gradation value with respect to the focus amount change is small, and is hardly sensitive to the focus amount. In the pupil focus map of FIG. 23 (f), all the bivariate correlation diagrams for each position in the pupil are displayed in an auto scale. At the intra-pupil positions 231b and 231d where the gradation change with respect to the focus amount change is small, as shown in FIGS. 23B and 23D, the plot of the bivariate correlation diagram is expanded in the vertical axis direction (vertical axis Graph with width in the direction). It can be seen from the pupil focus map in FIG. 23 (f) that such places are distributed in a catenary manner on the outer periphery of the pupil. At the intra-pupil position 231c inside the catenary distribution, the maximum gradation value is obtained when the focus amount = 0, and the gradation value tends to decrease with defocusing regardless of whether the focus amount is positive or negative (focus amount). It tends to be a mountain-shaped change in gradation value graph). On the other hand, at the intra-pupil positions 231a and 231e outside the catenary distribution, the minimum gradation value is obtained when the focus amount = 0, and the gradation value increases with defocusing regardless of whether the focus amount is positive or negative. A tendency (a tendency to become a valley-shaped gradation value change graph with respect to the focus amount change).

  In-pupil positions 231a, 231b, 231c, 231d, and 231e in the pupil focus map of FIG. 23 (f) and in-pupil positions 171a, 171b, 171c, 171d, and 171e in the pupil CD map of FIG. 16 is a gradation value of the image sensor 18 at the intra-pupil positions 161a, 161b, 161c, 161d, 161e of the pupil image 160 in FIG. The sensitivity to the CD reference wafer and the sensitivity to the focus reference wafer are different as shown in FIG. 17 (f) and FIG. This tendency is summarized in FIG. According to FIG. 24, the intra-pupil positions 161b and 161d in the pupil image 160 in FIG. 16 are irradiation conditions that are sensitive to the CD reference wafer 180 but not (or hardly sensitive) to the focus reference wafer 220. The intra-pupil position 161c is an irradiation condition that is sensitive to the focus reference wafer 220 but not sensitive to the CD reference wafer 180 (almost insensitive). Note that “+” in FIG. 24 indicates that the gradation value increases when the CD value increases, and “−” in FIG. 24 indicates that the gradation value decreases when the CD value increases. To express.

  FIG. 25 shows the gradation value at the in-pupil position 161b in FIG. 16 when the focus reference wafer 220 in FIG. 22 is measured by the surface inspection apparatus 1 of the present embodiment. The coordinates are shown. In FIG. 25, a portion having a high gradation value is set to a dark color (black), and a portion having a low gradation value is set to a light color (white), and displayed as a gradation legend 251. As can be seen from FIG. 25, a shot exposed by changing the focus amount on the focus reference wafer 220 cannot be clearly detected as a shot having a gradation difference. That is, it is hardly sensitive to the change in the cross-sectional shape of the line portion of the pattern accompanying the change in focus. 25 is the same irradiation condition as the in-pupil position 161b when the gradation value of the CD reference wafer 180 is measured as shown in FIG. 19, the gradation legend 251 in FIG. The same as the legend 191. Therefore, the tendency difference at the intra-pupil position 161b summarized in FIG. 24 can be easily seen.

  FIG. 26 shows the gradation value at the in-pupil position 161c in FIG. 16 when the focus reference wafer 220 in FIG. 22 is measured by the surface inspection apparatus 1 of the present embodiment. The coordinates are shown. In FIG. 25, a portion having a high gradation value is set to a dark color (black), and a portion having a low gradation value is set to a light color (white), and displayed as a gradation legend 261. As can be seen from FIG. 25, the shot exposed by changing the focus amount on the focus reference wafer 220, that is, the cross-sectional shape of the line portion of the pattern is different and the skirt is pulled out (the side wall in FIG. 4D). A shot having a large angle α is clearly detected as a shot having a gradation difference. 26 is the same irradiation condition as the in-pupil position 161c when the gradation value of the CD reference wafer 180 is measured as shown in FIG. 21, the gradation legend 261 in FIG. The same as the legend 211. Therefore, the difference in tendency at the in-pupil position 161c summarized in FIG. 24 can be easily seen.

  Although it is not easy to quantify the change in physical quantity when the focus amount is changed, the result of actually verifying the effect of the present invention will be shown by some examples.

  Measurement of the side wall angle is extremely difficult, and is impossible with a CD-SEM for CD value measurement. In practice, a method is used in which a cross section is cut off with a focus ion beam or the like, and the cross section is reviewed and measured with an SEM. This is a destructive measurement and unsuitable for measuring many points. By this method, a shot with the focus amount of the focus reference wafer 220 changed was selected, and several side wall angles were measured. The surface inspection of this embodiment is performed on the same location on the vertical axis (displayed as an arbitrary scale from a predetermined angle, indicating that the hem is wider as the number is larger). FIG. 27 shows the gradation value at the in-pupil position 161c as measured by the apparatus 1 on the horizontal axis. From FIG. 27, it can be seen that the change in the side wall angle is captured as the change in the gradation value with good correlation at the in-pupil position 161c. That is, according to the present embodiment, it is possible to easily know the sidewall angle that has been measured over a long period of time due to the fracture measurement.

  When defocused and exposed, the side wall angle becomes large and the pattern shape is often rough as shown in FIG. The vertical axis of FIG. 28 shows the degree of roughness obtained by digitizing the degree of pattern roughness from data obtained by an electron microscope such as SEM for the focus reference wafer 220. The degree of roughness is a numerical value of the deviation when the line part is a straight line for each minute point of the line part observed from above, and the larger the number, the more the pattern linearity is more disturbed and rougher I reckon. On the other hand, the horizontal axis of FIG. 28 shows the gradation value at the in-pupil position 161c when the surface roughness measuring apparatus 1 measures the roughness level. In FIG. 28, different markers (for example,..., X, +, Y, □, Z, etc.) represent different focus amounts shown in FIG. The horizontal histogram 280 shown in FIG. 28 shows the distribution of gradation values at the in-pupil position 161c, and the vertical histogram 281 shows a rough distribution. In the focus reference wafer 220 exposed by changing the focus amount stepwise, the degree of roughness should change stepwise from the focus amount = 0. However, the histogram 281 showing the distribution of the degree of roughness shows stepwise change. Not. On the other hand, in the histogram 280 showing the distribution of gradation values, the gradation values show a stepwise distribution such as the first distribution 280a, the second distribution 280b, the third distribution 280c, and the fourth distribution 280d. ing. That is, according to the present embodiment, it is possible to detect the difference in the degree of roughness of the pattern with a higher discrimination ability than ever before.

  When a defocus error occurs during exposure, a phenomenon called “film slippage” is also induced. Film slip is a reduction in photoresist thickness. As described above, in the structural birefringence, since the pattern height is included in the equation for determining the polarization state, the variation in height cannot be ignored. Since the defocus shot of the focus reference wafer 220 includes a combination of tail, roughness, film slip, etc., when calculating the side wall angle from the relationship shown in FIG. 27, the calculation accuracy is limited. is there. The effect of the present embodiment is maximized not by obtaining a relational expression with an existing physical quantity (such as a side wall, an angle, or a rough condition represented by LER, LWR, etc.), but by defocusing. A total physical quantity change that sometimes occurs is detected as a gradation value change at the in-pupil position 161c, and the change quantity is made to correspond to the focus amount of the exposure apparatus. Thereby, the focus error at the time of exposure can be discovered at an early stage.

  The results summarized in FIG. 24 and the results shown in FIG. 19, FIG. 21, FIG. 25, and FIG. 26 are results under optimum irradiation conditions. The optimum irradiation condition is an irradiation condition that is sensitive to one of the two variables of the line width of the pattern and the cross-sectional state of the pattern and not sensitive to the other at the position in the pupil. That is, there exists an in-pupil position where the line width can be measured without being affected by the cross-sectional state by the specific wavelength λ and the pattern azimuth angle θ, as in the in-pupil position 161b, and a line in the in-pupil position 161c. There is an in-pupil position where the cross-sectional state can be known without being affected by the width. In other words, when the wavelength λ and the pattern azimuth angle θ are changed, the optimal irradiation condition does not exist in the position in the pupil, or exists only in a region limited to the position in the pupil.

  FIG. 29A shows a pupil CD map when the CD reference wafer 180 is irradiated with light having a wavelength λ2 (for example, about 600 to 700 nm). FIG. 29B shows a pupil focus map when the focus reference wafer 220 is irradiated with light of wavelength λ2. When the wavelength when obtaining the pupil CD map of FIG. 17 (f) and the pupil focus map of FIG. 23 (f) (that is, the wavelength under the optimal irradiation condition) is the wavelength λ1, the wavelength λ1 is the wavelength λ2. Is a different wavelength (for example, about 400 to 500 nm). The pattern azimuth angle θ is the pattern azimuth angle θ1 under the same conditions in FIGS. 17 (f), 23 (f), 29 (a), and 29 (b).

  Sensitivity to the CD reference wafer 180 and the focus reference wafer 220 at the intra-pupil positions 290a, 290b, 290c, 290d, 290e, and 290f shown in FIGS. 29A and 29B is summarized including correlation errors. This is shown in FIG. The position in the pupil where the line width can be measured without being affected by the cross-sectional state and the position within the pupil where the cross-sectional state can be known without being affected by the line width are both in a considerably narrow range. The intra-pupil position 290e can be an intra-pupil position where the line width can be measured without being affected by the cross-sectional state, and the intra-pupil position 290d can be an intra-pupil position where the cross-sectional state can be known without being affected by the line width. The intra-pupil position 290b is a so-called dead zone position that not only has no cross-sectional stress, but also has a weak stress with respect to the line width. In FIG. 29 (a), the line width is affected by conditions other than the line width, such as the background of the disturbance and the pattern cross-sectional state, and conditions other than the side wall angle, such as the in-pupil position 290a. Although there is sensitivity to, there is a position with a large error in the pupil.

  In this way, depending on the wavelength, the position in the pupil that can measure the line width without being affected by the cross-sectional state is lost or narrowed, and the cross-sectional state is known without being affected by the line width. The selection of the wavelength is important because there may be a loss of the position within the pupil that can be reduced. This is because the wavelength is included in the equation for determining the polarization state change due to structural birefringence. That is, the optimum irradiation conditions including the wavelength should be determined.

  FIG. 31A shows a pupil CD map when the CD reference wafer 180 is irradiated with light having a wavelength λ1 at a pattern azimuth angle θ2 (for example, about 20 degrees). FIG. 31B shows a pupil focus map when the focus reference wafer 220 is irradiated with light having the wavelength λ1 at the pattern azimuth angle θ2. The pattern azimuth angle θ1 (that is, the pattern azimuth angle under optimum irradiation conditions) when the pupil CD map of FIG. 17F and the pupil focus map of FIG. 23F are obtained is the pattern azimuth angle θ2. Are different pattern azimuth angles (for example, about 45 degrees). The wavelength λ1 is under the same conditions in FIGS. 17 (f), 23 (f), 29 (a), and 29 (b).

  For comparison, FIGS. 31 (c) and 31 (d) show a pupil CD map and a pupil focus map, respectively, when light of wavelength λ1 is irradiated at the pattern azimuth angle θ1. FIGS. 31 (c) and 31 (d) show the same region in the pupil of FIGS. 17 (f) and 23 (f) cut out. The intra-pupil position 171b in FIG. 31 (c) and the intra-pupil position 231b in FIG. 31 (d) have the same pupil coordinates as the intra-pupil position 310b in FIGS. 31 (a) and 31 (b). Further, the intra-pupil position 171c in FIG. 31 (c) and the intra-pupil position 231c in FIG. 31 (d) have the same pupil coordinates as the intra-pupil position 310c in FIGS. 31 (a) and 31 (b). Comparing FIG. 31 (a) with the pattern azimuth angle θ2 and FIG. 31 (c) with the pattern azimuth angle θ1, in the pupil CD map, even if the pattern azimuth angle changes, the positions in the pupil that are insensitive to the line width are almost the same. You can see that it is a place.

  On the other hand, comparing FIG. 31 (b) with the pattern azimuth angle θ2 and FIG. 31 (d) with the pattern azimuth angle θ1, in the pupil focus map, the position in the pupil that is not sensitive to focus changes when the pattern azimuth angle changes. I understand that. In FIG. 31 (b), the positions in the pupil not sensitive to the focus are indicated by lines 311 and 312, and in FIG. 31 (d), the positions in the pupil not sensitive to the focus are indicated by lines 313 and 314. Then, the change in the position in the pupil that does not respond to the focus becomes easier to understand, and the region that is not sensitive to the focus in FIG. 31 (b) with the pattern azimuth angle θ2 than the FIG. 31 (d) with the pattern azimuth angle θ1. (That is, the lines 311 and 312 in FIG. 31B are moved outside the pupil from the lines 313 and 314 in FIG. 31D). As a result, as is apparent from FIG. 31B of the pattern azimuth angle θ2, the in-pupil position 310b is no longer an irradiation condition that is not sensitive to the focus. In this case, in order to obtain the optimum irradiation condition, it may be optimum if the intra-pupil position 310b is shifted, but it is necessary to determine whether or not it is optimum from an error from the CD in the pupil CD map.

  As described above, since a good position in the pupil where the line width can be measured without being influenced by the cross-sectional state may move depending on the pattern azimuth, selection of the pattern azimuth is important. From these facts, the optimum irradiation condition should be determined including not only the wavelength but also the pattern azimuth angle. As described above, by optimally selecting the pattern azimuth angle, wavelength, and position in the pupil, pattern state inspection can be performed simultaneously with CD measurement with high accuracy.

  Here, an example of the surface inspection method according to the present embodiment will be described with reference to the flowchart shown in FIG. First, using the CD reference wafer 180, the pupil CD map shown in FIG. 17F is obtained by the above-described method (step S101). Further, using the focus reference wafer 220, the pupil focus map shown in FIG. 23F is obtained by the method described above (step S102).

  When the pupil CD map of FIG. 17 (f) and the pupil focus map of FIG. 23 (f) are obtained, for example, from among a plurality of intra-pupil positions set for the pupil image 160 of FIG. A position 161b (first specific position) in the pupil that is sensitive to the CD reference wafer 180 but not to the focus reference wafer 220 is specified and sensitive to the focus reference wafer 220, but sensitive to the CD reference wafer 180. The in-pupil position 161c (second specific position) not to be specified is specified (step S103). At this time, other appropriate irradiation conditions (wavelength λ1 and pattern azimuth angle θ1) are also determined. Note that the position in the pupil that is sensitive to the CD reference wafer 180 but not sensitive to the focus reference wafer 220 is 45 degrees from the upper left to the lower right in the pupil image 160 in FIG. 16, as in the above-described pupil position 161b. The irradiation direction of the linearly polarized light is tilted in the vicinity of 45 degrees (counterclockwise in FIG. 16) with respect to the position in the pupil lined up on the straight line extending to The position in the pupil corresponding to the irradiation condition is preferable. Since this intra-pupil position is a position that is not affected by the background of the repetitive pattern 30, the CD value can be accurately obtained from the gradation value at the intra-pupil position.

  When the irradiation conditions are obtained in this way, the surface inspection of the wafer 5 can be accurately performed using the surface inspection apparatus 1 under the irradiation conditions. Therefore, the surface of the wafer 5 is irradiated with linearly polarized light using the illumination optical system 10 under the obtained irradiation conditions (step S104). At this time, as shown in FIG. 2, the light emitted from the light source 11 passes through the condenser lens 12, the wavelength selection filter 13, and the first polarizing filter 14, and then is reflected downward by the prism 8 to be parallel. Incident light I (linearly polarized light) is applied to the wafer 5 through the objective lens 7.

  The reflected light from the wafer 5 is received by the objective lens 7 (step S105). At this time, the reflected light from the wafer 5 becomes parallel reflected light J through the objective lens 7, passes through the prism 8, the second polarizing filter 16, and the relay lens 17 and enters the image sensor 18. Therefore, a polarization component perpendicular to the polarization direction of the incident light I (linearly polarized light) is detected using the image sensor 18 provided at a position conjugate with the pupil plane (not shown) of the objective lens 7 (step S106). At this time, the reflected light incident on the image sensor 18 is photoelectrically converted into an electric signal by the image sensor 18, and a detection signal of the reflected light transmitted through the second polarizing filter 16 is output to the arithmetic processing unit 20.

  Then, the arithmetic processing unit 20 obtains the CD value of the line portion 31 of the repetitive pattern 30 and the focus amount at the time of exposure based on the detection signal input from the image sensor 18 (step S107). At this time, the arithmetic processing unit 20 uses the pupil CD map of FIG. 17 (f) and the pupil focus map of FIG. 23 (f) to calculate the floor at the in-pupil position 161b (first specific position) specified in step S103. The CD value is obtained from the tone value, and the focus amount is obtained from the gradation value at the in-pupil position 161c (second specific position). When the obtained CD value deviates from a predetermined threshold value, it can be determined that the dose amount is abnormal and can be displayed on an image display device (not shown) or the like. Further, when the obtained focus amount deviates from a predetermined threshold value, it can be determined that the focus is abnormal and can be displayed on the image display device or the like.

  As described above, according to the present embodiment, not only can the CD value (line width) of the wafer 5 be accurately measured, but also the change in the cross-sectional state of the pattern can be detected simultaneously with the CD measurement. Therefore, if applied to the wafer 5 after exposure, it is possible to distinguish and detect the focus abnormality and the dose amount abnormality that occur in the exposure process, so that the problem of the exposure apparatus is solved, and the defective wafer is transferred to the rework process and regenerated. can do. If focus abnormalities, which were previously inspected for a long time after etching, can be inspected at the photoresist stage, it will have a tremendous effect on the early detection and resolution of problems, and the unit cost per chip is increasing. For cutting-edge semiconductor devices, it can also be expected to reduce the manufacturing cost per chip.

  In the above-described embodiment, illumination light and reflected light are received using the objective lens 7 having a high NA, and a specific irradiation condition is set by selectively receiving the position of the objective lens 7 in the pupil plane. Although selected, the specific irradiation condition is as small as about NA = 0.01 to 0.1 as described above. For this reason, after determining a specific irradiation condition, it is possible to irradiate and receive light using an optical system with a low NA set to the specific irradiation condition without using an objective lens with a high NA. For example, as shown in FIG. 32, it is determined by the incident angle i1, the incident plane azimuth angle ω1, and the pattern azimuth angle θ1 so that the line width can be measured without being affected by the cross-sectional state, and is represented by the polarization angle φ1. The incident angle at which the first optical system 321 constituting the first irradiation condition determined by the polarization direction of the incident light and the wavelength λ1 and the sectional state can be known without being affected by the line width The second optical system that is determined by i2, the incident surface azimuth angle ω2, and the pattern azimuth angle θ2, and that constitutes the second irradiation condition determined by the polarization direction of the incident light typified by the polarization angle φ2 and the wavelength λ2. 322 may be included. Therefore, the present invention is not limited to the configuration using the same pupil of the same objective lens.

  Further, in the above-described embodiment, the CD reference wafer 180 and the focus reference wafer 220 used for determining the optimum irradiation condition do not need to be separate wafers. The configuration of the CD reference wafer 180 and the configuration of the focus reference wafer 220 are the same. It is good also as one wafer of the composition which also had.

DESCRIPTION OF SYMBOLS 1 Surface inspection apparatus 5 Wafer (substrate) 7 Objective lens (light-receiving optical system)
10 Illumination optical system (irradiation unit) 15 Detection optical system (detection unit)
20 Arithmetic processing part (calculation part)
30 Repeat pattern (linear pattern)
161b In-pupil position (first specific position)
161c In-pupil position (second specific position)
180 CD reference wafer (first reference substrate)
220 Focus reference wafer (second reference substrate)

Claims (7)

An irradiation step of irradiating the surface of the substrate having a predetermined repeating pattern with linearly polarized light;
A light receiving step of receiving reflected light from the surface of the substrate irradiated with the linearly polarized light by a light receiving optical system;
A detection step for detecting a polarization component substantially perpendicular to a polarization direction of the linearly polarized light in the reflected light received in the light receiving step on a pupil plane of the light receiving optical system or a plane conjugate with the pupil plane;
From the information of the polarization component detected in the detection step, a calculation step for obtaining at least one of a line width of the repetitive pattern and a focus state at the time of exposure of the repetitive pattern,
In the calculating step, the line width is obtained from information on the polarization component at a first specific position having a high correlation with the line width on the pupil plane or a plane conjugate with the pupil plane, and the pupil plane or the pupil A surface inspection method, wherein the focus state is obtained from information on the polarization component at a second specific position different from the first specific position having a high correlation with the focus state on a plane conjugate with a surface. Correlation between information on the polarization component and the line width at a plurality of intra-pupil positions set in the pupil plane or a plane conjugate with the pupil plane using the first reference substrate having a known line width Line width correlation calculating step
A focus correlation calculating step for obtaining a correlation between the information of the polarization component and the focus state at the plurality of set positions in the pupil, using a second reference substrate whose focus state is known;
Among the plurality of intra-pupil positions, the intra-pupil position having a high correlation between the polarization component information and the line width and a low correlation with the focus state is specified as the first specific position, and the polarization component A specifying step of specifying, as the second specific position, a position in the pupil having a low correlation between information and the line width and a high correlation with the focus state;
In the calculation step, the line width is obtained from the information on the polarization component at the first specific position specified in the specification step, and from the information on the polarization component at the second specific position specified in the specification step. The surface inspection method according to claim 1, wherein the focus state is obtained.   In at least one of the line width correlation calculation step and the focus correlation calculation step, the pupil plane or the entire plane conjugate with the pupil plane is divided into a plurality of sections, and the plurality of pupil positions are set. The surface inspection method according to claim 1, wherein the correlation is obtained respectively.   4. The at least one of the first specific position and the second specific position in the specifying step is an intra-pupil position showing a correlation having linearity. 5. Surface inspection method.   The first specific position in the specifying step is a position in a pupil corresponding to an irradiation condition in which an irradiation direction of the linearly polarized light is inclined close to 45 degrees with respect to a polarization direction of the linearly polarized light. The surface inspection method according to any one of 1 to 4. An irradiation unit for irradiating the surface of the substrate having a predetermined repeating pattern with linearly polarized light;
A light receiving optical system that receives reflected light from the surface of the substrate irradiated with the linearly polarized light;
A detection unit that detects a polarization component substantially perpendicular to a polarization direction of the linearly polarized light in the reflected light received by the light receiving optical system in a pupil plane of the light receiving optical system or a plane conjugate with the pupil plane;
From the information on the polarization component detected by the detection unit, a calculation unit for obtaining at least one of a line width of the repetitive pattern and a focus state at the time of exposure of the repetitive pattern,
The calculation unit obtains the line width from information on the polarization component at a first specific position having a high correlation with the line width on the pupil plane or a plane conjugate with the pupil plane, and the pupil plane or the pupil. A surface inspection apparatus, wherein the focus state is obtained from information on the polarization component at a second specific position different from the first specific position having a high correlation with the focus state on a plane conjugate with a surface.   The surface inspection apparatus according to claim 6, further comprising an imaging device disposed on a pupil plane of the light receiving optical system or a plane conjugate with the pupil plane.