JP7848272B2 - Semiconductor device manufacturing methods - Google Patents

Description

One aspect of this disclosure relates to a semiconductor device manufacturing method and a measuring apparatus.

Patent Document 1 discloses a technique for separating light from an object using a dichroic mirror whose transmittance and reflectance change depending on the wavelength, determining the wavelength centroid by imaging each of the separated light streams, and estimating the film thickness of a semiconductor device based on the wavelength centroid.

International Publication No. 2021/161986

In recent years, with the increasing integration of semiconductor devices, ensuring uniformity of the thickness of each layer during the manufacturing process has become crucial. If the thickness of each layer is not uniform during integration, it can lead to failures such as wiring defects and void formation. As semiconductor devices become smaller, the thickness of each layer is extremely thin, for example, less than 100 nm, requiring highly accurate thickness measurement. One known method for measuring thickness involves detecting reflected interference light from the semiconductor device using a spectrometer to obtain a spectral distribution and estimate the thickness of each layer. However, this measurement method only measures thickness at specific points, and when it is necessary to derive the thickness distribution across the entire wafer with high accuracy, the measurement time becomes extremely long.

This disclosure is made in view of the above circumstances and aims to provide a semiconductor device manufacturing method that can derive the wafer thickness distribution quickly and with high accuracy.

(1) A semiconductor device manufacturing method according to one aspect of the present disclosure comprises: a first step of irradiating a wafer in a planar manner with light while a first layer is formed on the surface side of the wafer, imaging the light from the wafer, and obtaining the thickness distribution of the first layer in the plane of the wafer based on the imaging signal; a second step of forming a second layer by performing a predetermined processing on the first layer; a third step of irradiating the wafer in a planar manner with light after the second step, imaging the light from the wafer, and deriving the distribution of measurement parameters in the plane of the wafer based on the imaging signal; and a fourth step of deriving the thickness distribution of the second layer based on the thickness distribution of the first layer obtained in the first step and the distribution of measurement parameters derived in the third step.

In a semiconductor device manufacturing method according to one aspect of this disclosure, after the formation of the second layer, light is irradiated onto the wafer in a planar manner, and the light from the wafer is imaged, thereby deriving the distribution of measurement parameters within the wafer plane. Then, the thickness distribution of the second layer is derived based on the thickness distribution of the first layer obtained before the formation of the second layer and the distribution of measurement parameters within the wafer plane derived after the formation of the second layer. In this way, because the thickness distribution of the first layer is obtained in advance before the formation of the second layer, the thickness distribution of the second layer, which is the surface layer of the wafer, can be appropriately (with high accuracy) derived based on the distribution of measurement parameters within the wafer plane derived after the formation of the second layer and the thickness distribution of the first layer. Furthermore, since the thickness distribution of the second layer is derived taking into account the image results of light irradiated onto the wafer in a planar manner, the thickness distribution of the second layer within the entire wafer plane can be derived all at once (in a short time). As a result, the measurement time required to derive the thickness distribution within the entire wafer plane can be significantly reduced compared to, for example, when the thickness of the second layer is measured at points using a spectrometer or the like. As described above, according to one aspect of this disclosure, the semiconductor device manufacturing method allows for the rapid and highly accurate determination of the wafer thickness distribution.

(2) The semiconductor device manufacturing method described in (1) above further comprises a relationship information derivation step, which derives relationship information showing the relationship between measurement parameters and the thickness of the second layer for each predetermined region, based on the thickness distribution of the first layer obtained in the first step. In the fourth step, the thickness distribution of the second layer may be derived based on the relationship information derived in the relationship information derivation step and the distribution of measurement parameters derived in the third step. In this way, by deriving and using relationship information showing the relationship between measurement parameters and the thickness distribution of the second layer for each predetermined spatial region under the conditions of the obtained thickness distribution of the first layer, the thickness distribution of the wafer surface layer can be derived quickly and with high accuracy from the distribution of measurement parameters derived in the third step.

(3) In the semiconductor device manufacturing method described in (2) above, the relational information derivation step may include: deriving the theoretical reflectance for each predetermined region of the second layer thickness distribution based on the thickness distribution of the first layer obtained in the first step and the thin-film information of the second layer obtained in advance; and deriving measurement parameters for each predetermined region of the second layer thickness distribution based on the theoretical reflectance for each thickness distribution of the second layer and the spectral characteristics obtained in advance, and deriving relational information that shows the relationship between the measurement parameters and the thickness of the second layer for each predetermined region. In this way, by specifying the thickness distribution of the first layer and the thin-film information of the second layer, the theoretical reflectance for each predetermined region of the second layer thickness distribution can be derived with high accuracy. Furthermore, by specifying the theoretical reflectance and spectral characteristics for each thickness distribution of the second layer, measurement parameters for each predetermined region can be derived for each thickness distribution of the second layer, and the aforementioned relational information can be derived with high accuracy. With this configuration, the thickness distribution of the wafer surface layer can be derived with high accuracy from the distribution of measurement parameters derived in the third step using the relational information.

(4) In the semiconductor device manufacturing method described in (2) or (3) above, the related information derivation step may be performed during the execution of the second step. This allows for the rapid derivation of the thickness distribution of the wafer surface layer after the execution of the second step.

(5) In the semiconductor device manufacturing method described in any one of the above items (2) to (4), the relational information may be a relational expression or table obtained by fitting for each predetermined region. By using such relational information, the thickness distribution of the wafer surface layer can be derived quickly and with high accuracy.

(6) In the semiconductor device manufacturing method described in any one of the above items (1) to (5), the measurement parameter may be the wavelength centroid. With this configuration, the thickness distribution of the wafer surface layer can be derived with high accuracy by using the wavelength centroid, which has a high correlation with the film thickness, as the measurement parameter.

(7) In the semiconductor device manufacturing method described in any one of items (1) to (6) above, the processing treatment may be any of the following treatments on the wafer: film deposition, photoresist coating, etching, planarization, electrode formation, pattern formation, insulating film formation, contact hole formation, contact formation, trench formation, or wiring formation. The thickness distribution of the surface layer of the wafer formed after these treatments can be derived quickly and with high accuracy using the method described above. However, after the pattern formation treatment, conventional point measurement for film thickness derivation is difficult due to difficulties in alignment, etc. In this regard, the method according to this disclosure eliminates the need for the above-mentioned alignment, etc., after the pattern formation treatment, making it easy to derive the thickness distribution of the surface layer of the wafer.

This disclosure provides a semiconductor device manufacturing method that can shorten measurement time and perform thickness measurement with high accuracy.

Figure 1 is a schematic diagram showing the film thickness measuring device according to this embodiment. Figure 2 illustrates the relationship between the characteristics of a dichroic mirror and the wavelength of light emitted from a light source. Figure 3 illustrates the characteristics of the light spectrum and the tilted dichroic mirror. Figure 4 illustrates the wavelength shift corresponding to the amount of transmitted and reflected light. Figure 5 shows the relationship between wavelength and film thickness. Figure 6 is a schematic diagram showing an example of a configuration for estimating spectral characteristics. Figure 7 illustrates an example of a method for estimating spectral characteristics using the configuration shown in Figure 6. Figure 8 schematically shows an example of another configuration for estimating spectral characteristics. Figure 9 illustrates the derivation of the expected values of the measurement parameters. Figure 10 illustrates the derivation of the relational equation using curve fitting. Figure 11 illustrates the derivation of film thickness using a relational equation. Figures 12(a) to 12(c) illustrate the derivation of relational equations related to measurement parameters other than the wavelength centroid. Figure 13 is a schematic diagram showing a film thickness measuring device according to a modified example. Figure 14 is a schematic cross-sectional view showing a wafer with a multilayer film structure according to a modified example. Figure 15(a) shows the distribution of the wavelength centroid measured for the first layer, and Figure 15(b) shows the distribution of the film thickness for the first layer derived based on the wavelength centroid shown in Figure 15(a). Figure 16 shows the distribution of the wavelength centroid measured after the deposition of the second layer. Figure 17 illustrates the derivation of the second layer thickness based on the wavelength centroid measured after the second layer deposition and the first layer thickness. Figure 18 illustrates the relationship between the second layer thickness and the wavelength centroid for each first layer thickness. Figure 19 is a diagram illustrating the manufacturing process of semiconductor devices. Figure 20 is a diagram illustrating the manufacturing process of semiconductor devices. Figure 21 illustrates the procedure for deriving the thickness distribution of the second layer. Figure 22 illustrates the derivation of the theoretical reflectance for each candidate second layer thickness. Figure 23 illustrates the derivation of the measurement parameters. Figure 24 illustrates the derivation of relationship information between measurement parameters and the thickness distribution of the second layer. Figure 25 illustrates the derivation of the thickness distribution of the second layer. Figure 26 illustrates the problems of a semiconductor device relating to a comparative example. Figure 27 illustrates the effects of deriving the thickness distribution according to this embodiment.

The embodiments of this disclosure will be described in detail below with reference to the drawings. In each drawing, the same or corresponding parts are denoted by the same reference numerals, and redundant descriptions are omitted.

Figure 1 is a schematic diagram of the film thickness measuring device 1 according to this embodiment. The film thickness measuring device 1 is a device that irradiates a wafer 100 with light in a planar manner and measures the thickness of a film formed on the wafer 100, specifically the thickness distribution of the film, based on the reflected light from the wafer 100. Examples of wafers include logic devices, memory devices, analog devices, and mixed-signal devices that combine the same, such as integrated circuits (ICs) or large-scale integrated circuits (LSIs) having PN junctions, such as transistors; power semiconductor devices (power devices) such as high-current/high-voltage MOS transistors, bipolar transistors, and IGBTs; and light-emitting devices such as LEDs and semiconductor lasers. The wafer 100 has a film 100b formed on the surface of a substrate 100a. Here, it is explained that only one layer of film 100b is formed on the surface of the substrate 100a of the wafer 100. The film 100b is, for example, an oxide film or a nitride film, but other films may also be used.

As shown in Figure 1, the film thickness measuring device 1 comprises a light source 10, a half mirror 11, a field lens 12, a camera system 20, and a control device 30.

The light source 10 irradiates light onto the wafer 100 in a planar manner. For example, the light source 10 irradiates light onto substantially the entire surface of the wafer 100 in a planar manner. The light source 10 is, for example, a light source capable of uniformly irradiating the surface of the wafer 100, and irradiates the wafer 100 with diffused light. The light source 10 may be a surface illumination unit using a white LED, SC light source, halogen lamp, or Xe lamp, etc. Alternatively, it may be a surface illumination unit using a monochromatic LD, multichromatic LD, monochromatic LED, or multichromatic LED. The light emitted from the light source 10 passes through the half-mirror 11 and the field lens 12, and is then irradiated onto the wafer 100 in a planar manner.

The light source 10 irradiates the wafer 100 with light of a wavelength included in a predetermined wavelength range of the tilted dichroic mirror 22 (details described later) of the camera system 20. As will be described in detail later, the tilted dichroic mirror 22 is an optical element that separates light from the wafer 100 by transmitting and reflecting it according to its wavelength. Within the predetermined wavelength range described above, the transmittance and reflectance of the tilted dichroic mirror 22 change according to the wavelength.

Figure 2 illustrates the relationship between the characteristics of the tilted dichroic mirror 22 and the wavelength of light emitted from the light source 10. In Figure 2, the horizontal axis represents wavelength, and the vertical axis represents the transmittance of the tilted dichroic mirror 22. As shown in the characteristic X4 of the tilted dichroic mirror 22 in Figure 2, in the tilted dichroic mirror 22, the transmittance (and reflectance) of light changes gradually (monotonically or linearly) in a predetermined wavelength range X10 according to the change in wavelength, while in wavelength ranges other than this specific wavelength range, the transmittance (and reflectance) of light remains constant regardless of the change in wavelength. As shown in Figure 2, the light X20 output from the light source 10 includes light with wavelengths included in the predetermined wavelength range X10 described above. That is, the light source 10 outputs light with a broad spectrum including the predetermined wavelength range X10. The wavelength range (interference peak wavelength) related to measurement is determined by the material of the film formed on the wafer 100 and the measurement film thickness range.

Returning to Figure 1, the half-mirror 11 is a mirror that reflects light emitted from the light source 10 towards the wafer 100 (specifically, towards the field lens 12 that guides the light to the wafer 100) and transmits light from the irradiated wafer 100 (specifically, light that has passed from the wafer 100 through the field lens 12). The field lens 12 is a lens that aligns the direction of light propagation.

The camera system 20 comprises a lens 21, an inclined dichroic mirror 22, an area sensor 23, and an area sensor 24. The camera system 20 may also include a linear image sensor instead of the area sensors.

Lens 21 is a lens that focuses light from the wafer 100 that has been incident via the field lens 12 and the half mirror 11. Lens 21 may be positioned upstream of the inclined dichroic mirror 22, or it may be positioned in the region between the inclined dichroic mirror 22 and the area sensors 23 and 24. In this embodiment, it is described that lens 21 is positioned upstream of the inclined dichroic mirror 22. Lens 21 may be a finite-focus lens or an infinite-focus lens. If lens 21 is a finite-focus lens, the distance from lens 21 to the area sensors 23 and 24 is set to a predetermined value. If lens 21 is an infinite-focus lens, lens 21 is a collimator lens that converts light from the wafer 100 into parallel light, and is aberration-corrected to obtain parallel light. The light output from lens 21 is incident on the inclined dichroic mirror 22.

The tilted dichroic mirror 22 is a mirror made using a special optical material, and is an optical element that separates light from the wafer 100 by transmitting and reflecting it according to its wavelength. The tilted dichroic mirror 22 is configured such that its light transmittance and reflectance change according to wavelength within a predetermined wavelength range.

Figure 3 illustrates the characteristics of the light spectrum and the tilted dichroic mirror 22. In Figure 3, the horizontal axis represents wavelength, and the vertical axis represents spectral intensity (in the case of the light spectrum) and transmittance (in the case of the tilted dichroic mirror 22). As shown in the characteristic X4 of the tilted dichroic mirror 22 in Figure 3, in the tilted dichroic mirror 22, the transmittance (and reflectance) of light changes gradually in accordance with the change in wavelength in a predetermined wavelength range (wavelength range λ1 to λ2). On the other hand, in wavelength ranges other than the predetermined wavelength range (i.e., wavelengths lower than λ1 and wavelengths higher than λ2), the transmittance (and reflectance) of light may be considered constant regardless of the change in wavelength. In other words, in a specific wavelength band (wavelength range λ1 to λ2), the transmittance of light increases monotonically (reflectance decreases monotonically) in accordance with the change in wavelength. Transmittance and reflectance have a negative correlation; when one increases, the other decreases. Therefore, below, the term "transmittance (and reflectance)" may be used instead of simply "transmittance." Note that "the transmittance of light is constant regardless of the change in wavelength" includes not only cases where it is perfectly constant, but also cases where, for example, the change in transmittance for a change of 1 nm in wavelength is 0.1% or less. At wavelengths lower than λ1, the transmittance of light may be approximately 0% regardless of the change in wavelength, and at wavelengths higher than λ2, the transmittance of light may be approximately 100% regardless of the change in wavelength. Note that "the transmittance of light is approximately 0%" includes transmittances of approximately 0% + 10%, and "the transmittance of light is approximately 100%" includes transmittances of approximately 100% - 10%. In Figure 3, waveform X1 shows the waveform of light output from light source 10. As shown in waveform X1 of Figure 3, the light output from the light source 10 includes light with wavelengths within a predetermined wavelength range (wavelength λ1 to λ2) of the tilted dichroic mirror 22. This predetermined wavelength range is, for example, the visible wavelength range, and one example is the wavelength range from 400 nm to 700 nm.

Returning to Figure 1, the area sensors 23 and 24 image the light from the wafer 100 (imaging step). The area sensors 23 and 24 image the light separated by the tilted dichroic mirror 22. Area sensor 23 images the light transmitted through the tilted dichroic mirror 22 and outputs a second signal. Area sensor 24 images the light reflected by the tilted dichroic mirror 22 and outputs a first signal. The wavelength range to which area sensors 23 and 24 are sensitive corresponds to a predetermined wavelength range in the tilted dichroic mirror 22 where the transmittance (and reflectance) of light changes according to the change in wavelength. Area sensors 23 and 24 are, for example, monochrome sensors or color sensors. The imaging results (images) from area sensors 23 and 24 are output to the control device 30 via the first and second signals described above.

A bandpass filter (not shown) may be placed upstream of the area sensors 23 and 24. Such a bandpass filter (not shown) may, for example, be a filter that removes light in wavelengths other than the predetermined wavelength range described above (the wavelength range in the tilted dichroic mirror 22 where the transmittance and reflectance of light change depending on the wavelength).

The control device 30 is a computer, and physically comprises memory such as RAM and ROM, a processor (arithmetic circuit) such as a CPU, a communication interface, and a storage unit such as a hard disk. The control device 30 functions by executing programs stored in memory using the CPU of the computer system. The control device 30 may also be composed of a microcontroller or an FPGA.

The control device 30 derives the film thickness of the wafer 100 (the thickness distribution within the plane of the wafer 100) based on the first and second signals, which are signals from the area sensors 23 and 24 that capture light. The control device 30 performs two processes related to the derivation of film thickness: a measurement parameter derivation process based on the signals from the area sensors 23 and 24, and a film thickness derivation process based on the measurement parameters. Furthermore, as a prerequisite for performing the film thickness derivation process, the control device 30 stores relational information showing the relationship between film thickness and measurement parameters for each predetermined region. The control device 30 has a functional configuration to realize the above-described processes and storage, comprising a calculation unit 31, an analysis unit 32, and a storage unit 33. The calculation unit 31 has the function of performing the measurement parameter derivation process. The analysis unit 32 has the function of performing the film thickness derivation process. The storage unit 33 stores the above-described relational information. Each function will be described in detail below.

The calculation unit 31 derives measurement parameters for the wafer 100 for each predetermined region based on the signals from the area sensors 23 and 24 that image the light. The signals from the area sensors 23 and 24 indicate the transmitted light distribution and reflected light distribution in the imaging region. The measurement parameters can be any parameters correlated with the film thickness, such as the wavelength centroid of the light from the wafer 100, the intensity of the light transmitted through the tilted dichroic mirror 22, the intensity of the light reflected from the tilted dichroic mirror 22, or the ratio of the intensity of the light transmitted through the tilted dichroic mirror 22 to the intensity of the reflected light (IT/IR or IR/IT). In the following explanation, the measurement parameter will be described assuming it is the wavelength centroid of the light from the wafer 100.

The calculation unit 31 may derive the wavelength centroid of each light as a measurement parameter for each predetermined region, based on the spatial distribution of transmitted light (intensity of light transmitted through the tilted dichroic mirror 22) identified based on the second signal from the area sensor 23 indicating the imaging result at the area sensor 23, and the spatial distribution of reflected light (intensity of light reflected from the tilted dichroic mirror 22) identified based on the first signal from the area sensor 24 indicating the imaging result at the area sensor 24. The predetermined region may be, for example, a region corresponding to a pixel of the area sensors 23 and 24, or a region corresponding to multiple adjacent pixels. In the following, a pixel will be described as a predetermined region. Specifically, in deriving the wavelength centroid of the light, the calculation unit 31 derives the wavelength centroid of each pixel based on the following equation (1). In the following equation (1), x' represents the wavelength centroid, IT' represents the transmitted light, and IR' represents the reflected light.
x' = (IT' - IR') / 2(IT' + IR') (1)

Furthermore, the calculation unit 31 may derive the wavelength centroid of light for each pixel by considering the center wavelength of the tilted dichroic mirror 22 (the center wavelength of a predetermined wavelength range) and the width of the tilted dichroic mirror 22. The width of the tilted dichroic mirror 22 is, for example, the wavelength range from the wavelength at which the transmittance in the tilted dichroic mirror 22 is 0% to the wavelength at which the transmittance is 100%. In this case, the calculation unit 31 may derive the wavelength centroid of each pixel based on the following equation (2). In the following equation (2), x' is the wavelength centroid, IT' is the amount of transmitted light, IR' is the amount of reflected light, λ0 is the center wavelength of the tilted dichroic mirror 22, and A is the width of the tilted dichroic mirror 22.
x'=λ0+A(IT'-IR')/2(IT'+IR') (2)

Figure 4 illustrates the wavelength shift depending on the amount of transmitted and reflected light. When deriving x' (wavelength centroid) using equation (1) or (2) above, as shown in Figure 4, for pixels where IT' (transmitted light) = IR' (reflected light), x' = λ0 (the center wavelength of the tilted dichroic mirror 22). For pixels where IT' < IR', i.e., pixels where the reflected light is greater than the transmitted light, x' = λ1 (a wavelength shorter than λ0). For pixels where IT' > IR', i.e., pixels where the transmitted light is greater than the reflected light, x' = λ2 (a wavelength longer than λ0). Thus, the value of x' (wavelength centroid) shifts (wavelength shift) based on the amount of transmitted and reflected light.

Furthermore, since the wavelength centroid correlates with film thickness, it can be used to derive film thickness. Figure 5 shows the relationship between wavelength and film thickness. In Figure 5, the horizontal axis represents wavelength, and the vertical axis represents reflectance. In the examples shown in Figure 5, the relationship between wavelength and reflectance is shown for film thicknesses of 820 nm, 830 nm, and 840 nm. As shown in Figure 5, the wavelength centroid differs depending on the film thickness. Thus, because the wavelength centroid and film thickness are correlated, it is possible to estimate the film thickness by identifying the wavelength centroid.

Returning to Figure 1, the memory unit 33 stores information regarding the relationship between film thickness and measurement parameters (in this case, the wavelength centroid). As mentioned above, there is a correlation between film thickness and the wavelength centroid. Therefore, by having pre-prepared information regarding the relationship between film thickness and the wavelength centroid, the film thickness can be derived from this relationship information and the actually measured wavelength centroid. The memory unit 33 stores this relationship information for each type of film.

The relevant information may be derived based on the theoretical reflectance corresponding to the film type and the spectral characteristics (spectral sensitivity) of the entire film thickness measuring device 1. The theoretical reflectance can be determined for each wavelength once the film type (refractive index and extinction coefficient) and film thickness are determined. The spectral characteristics of the entire film thickness measuring device 1 can be predetermined (estimated) by various methods. The following provides an example of a method for estimating the spectral characteristics of the entire film thickness measuring device 1.

The spectral characteristics of the film thickness measuring device 1 may be estimated, for example, by accumulating the spectral characteristics (spectral sensitivity) of each optical component constituting the film thickness measuring device 1. Specifically, the spectral characteristics of the film thickness measuring device 1 may be estimated by accumulating the following: the luminance spectrum of the light source 10, the spectral transmittance (transmittance spectrum) of the half mirror 11, the spectral transmittance (transmittance spectrum) of the field lens 12, the spectral transmittance (transmittance spectrum) of the lens 21, the spectral transmittance of the tilted dichroic mirror 22, the quantum efficiency (QE) or spectral sensitivity of the area sensor 23, the quantum efficiency (QE) or spectral sensitivity of the area sensor 24, and the reflectance of a bare wafer placed in place of the wafer 100.

In this case, the spectral characteristics SCT_xm,yn(λ) of the transmitting side that passes through the tilted dichroic mirror 22 are given by equation (3) below. Also, the spectral characteristics SCR_xm,yn(λ) of the reflecting side that reflects the tilted dichroic mirror 22 are given by equation (4) below. In equations (3) and (4), λ is the wavelength, xm and yn are the coordinates on the surface of the wafer 100, SC1 is the brightness spectrum of the light source 10, SC2 is the spectral transmittance of the half mirror 11, SC3 is the spectral transmittance of the field lens 12, SC4 is the spectral transmittance of the lens 21, SC5 is the spectral transmittance of the tilted dichroic mirror 22, SC6 is the quantum efficiency or spectral sensitivity of the area sensor 23, SC7 is the quantum efficiency or spectral sensitivity of the area sensor 24, and R is the reflectance on the bare wafer.
SCT_xm,yn(λ)=SC1(λ)xm,yn×SC2(λ)xm,yn×SC3(λ)xm,yn×SC4(λ)xm,yn×SC5(λ)xm,yn×SC6(λ)xm,yn×R(λ) (3)
SCR_xm,yn(λ)=SC1(λ)xm,yn×SC2(λ)xm,yn×SC3(λ)xm,yn×SC4(λ)xm,yn×SC5(λ)xm,yn×SC7(λ)xm,yn×R(λ) (4)

Furthermore, if, for example, the spatial uniformity of the spectral characteristics is high on the surface of wafer 100 (the variation in spectral characteristics for each coordinate is small), then, without considering the positional dependence for each coordinate, one may use, for example, SC1(λ)xm,yn = SC1(λ) (the same applies to SC2 to SC7).

The spectral characteristics of the film thickness measuring device 1 may be estimated, for example, by using multiple types of bandpass filters. Figure 6 is a schematic diagram showing an example of a configuration for estimating the spectral characteristics of the film thickness measuring device 1. In addition to the components included in the film thickness measuring device 1 described above, Figure 6 shows multiple types of bandpass filters 51 to 54. For example, if the rate of change of the spectrum is a continuous monotonic characteristic, the spectral characteristics SCT_xm,yn(λ) on the transmission side and SCR_xm,yn(λ) on the reflection side of the film thickness measuring device 1 can be estimated all at once for each coordinate of the wafer 100 by sequentially switching between multiple types of bandpass filters 51 to 54 using a filter wheel (not shown) or a filter slider (not shown). In this case, without individually considering the spectral characteristics of each optical component constituting the film thickness measuring device 1, the spectral characteristics of the transmitted side, SCT_xm,yn(λ), are estimated from the transmitted light amount measured by the area sensor 23, and the spectral characteristics of the reflected side, SCR_xm,yn(λ), are estimated from the reflected light amount measured by the area sensor 24. Note that when estimating spectral characteristics using this configuration, a bare wafer 500 is used instead of wafer 100.

In the configuration shown in Figure 6, bandpass filters 51 to 54 are filters that remove light in wavelength ranges other than a predetermined wavelength range. Bandpass filter 51, for example, removes light in wavelength ranges other than the first wavelength range (lowest wavelength). Bandpass filter 52 removes light in wavelength ranges other than the second wavelength range (higher wavelength than the first wavelength range). Bandpass filter 53 removes light in wavelength ranges other than the third wavelength range (higher wavelength than the second wavelength range). Bandpass filter 54 removes light in wavelength ranges other than the fourth wavelength range (higher wavelength than the third wavelength range).

Figure 7 illustrates an example of a method for estimating spectral characteristics using the configuration shown in Figure 6. In Figure 7, the left side shows an image of one coordinate within the bare wafer 500; the upper right shows the transmitted light quantity measured by the area sensor 23 for the same coordinate while switching between bandpass filters 51-54; and the lower right shows the reflected light quantity measured by the area sensor 24 while switching between bandpass filters 51-54 for the same coordinate. In the upper right and lower right diagrams of Figure 7, the horizontal axis represents wavelength, and the vertical axis represents spectral intensity.

In the upper right diagram of Figure 7, wavelength region 151 corresponds to the first wavelength region of bandpass filter 51, wavelength region 152 corresponds to the second wavelength region of bandpass filter 52, wavelength region 153 corresponds to the third wavelength region of bandpass filter 53, and wavelength region 154 corresponds to the fourth wavelength region of bandpass filter 54. As shown in the upper right diagram of Figure 7, by determining the amount of transmitted light when using each bandpass filter 51-54, a curve 160 showing the relationship between wavelength and spectral intensity can be derived by, for example, performing curve fitting on this transmitted light data. The relational expression (polynomial) representing such a curve 160 is a relational expression showing the spectral characteristics SCT_xm,yn(λ) on the transmission side. Note that curve fitting methods include, for example, polynomial approximation and other curve fitting methods. Furthermore, a curve showing the relationship between wavelength and spectral intensity may also be derived using interpolation methods or other methods besides curve fitting.

Similarly, in the lower right diagram of Figure 7, the wavelength region corresponding to the first wavelength region of bandpass filter 51 is wavelength region 251, the wavelength region corresponding to the second wavelength region of bandpass filter 52 is wavelength region 252, the wavelength region corresponding to the third wavelength region of bandpass filter 53 is wavelength region 253, and the wavelength region corresponding to the fourth wavelength region of bandpass filter 54 is wavelength region 254. As shown in the lower right diagram of Figure 7, by specifying the amount of transmitted light when using each bandpass filter 51 to 54, a curve 260 showing the relationship between wavelength and spectral intensity is derived by performing curve fitting on this transmitted light data. The relational expression (polynomial) that shows such a curve 260 is a relational expression that shows the spectral characteristics SCR_xm,yn(λ) on the reflecting side. Note that curve fitting methods include, for example, polynomial approximation and other curve fitting methods. In addition to curve fitting, a curve showing the relationship between wavelength and spectral intensity may also be derived using interpolation methods or other methods. As described above, the spectral characteristics of the film thickness measuring device 1 can be estimated using the bandpass filters 51 to 54.

Furthermore, the spectral characteristics of the film thickness measuring device 1 may be estimated using, for example, a spectrometer. Figure 8 schematically shows an example of another configuration for estimating the spectral characteristics of the film thickness measuring device 1. Figure 8 shows a configuration in which spectrometers 60 and 70 are provided instead of the area sensors 23 and 24 of the film thickness measuring device 1 described above.

The spectrometer 60 can derive the spectral characteristic SC8(λ)xm,yn, which is the cumulative spectral characteristic of each optical component, excluding SC6(λ)xm,yn, which is the quantum efficiency or spectral sensitivity of the area sensor 23, from the transmitted spectral characteristic SCT_xm,yn(λ). The spectrometer 60 has a measuring unit 61 and a probe head 62. The spectrometer 60 spectrally analyzes the light input from the probe head 62 (light transmitted through the tilted dichroic mirror 22) for each wavelength, and derives the intensity for each wavelength in the measuring unit 61. In this way, the spectral characteristic SC8(λ)xm,yn described above is derived. The transmitted spectral characteristic SCT_xm,yn(λ) is expressed as the product of SC6(λ)xm,yn, which is the quantum efficiency or spectral sensitivity of the area sensor 23, and the spectral characteristic SC8(λ)xm,yn, which is the cumulative spectral characteristic of each optical component, as shown in equation (5) below.
SCT_xm,yn(λ)=SC6(λ)xm,yn×SC8(λ)xm,yn (5)

Similarly, the spectrometer 70 can derive the spectral characteristic SC9(λ)xm,yn, which is the cumulative spectral characteristic of each optical component, excluding SC7(λ)xm,yn, which is the quantum efficiency or spectral sensitivity of the area sensor 24, from the spectral characteristic SCR_xm,yn(λ) on the reflective side. The spectrometer 70 has a measuring unit 71 and a probe head 72. The spectrometer 70 spectrally analyzes the light input from the probe head 72 (light reflected from the tilted dichroic mirror 22) for each wavelength, and derives the intensity for each wavelength in the measuring unit 71. In this way, the spectral characteristic SC9(λ)xm,yn described above is derived. The spectral characteristic SCR_xm,yn(λ) on the reflective side is expressed as the product of SC7(λ)xm,yn, which is the quantum efficiency or spectral sensitivity of the area sensor 24, and the spectral characteristic SC9(λ)xm,yn, which is the cumulative spectral characteristic of each optical component, as shown in equation (6) below.
SCR_xm,yn(λ)=SC7(λ)xm,yn×SC9(λ)xm,yn (6)

Furthermore, while the above describes several examples of estimating the spectral characteristics of the film thickness measuring device 1 using the device itself, the estimation of the spectral characteristics of the film thickness measuring device 1 does not necessarily have to be performed using the device itself. That is, it is sufficient that the aforementioned relationship information is stored in the storage unit 33, and the spectral characteristics of the film thickness measuring device 1 used to derive this relationship information can be determined in any way.

As described above, the relationship between film thickness and wavelength centroid is derived based on the theoretical reflectance corresponding to the film type and the spectral characteristics of the film thickness measuring device 1. The derivation of this relationship is performed, in detail, by deriving the expected value of the wavelength centroid (measurement parameter) from the theoretical reflectance and the spectral characteristics of the film thickness measuring device 1 (see Figure 9), and then deriving the relationship equation by plotting the expected value of the wavelength centroid and performing curve fitting (see Figure 10). Below, an example of deriving the relationship between film thickness and wavelength centroid for a specific film type will be explained.

Figure 9 illustrates the derivation of the expected value of the wavelength centroid, which is a measurement parameter. Since the film type is fixed, the theoretical reflectance values for each wavelength are determined according to the film thickness. Suppose a certain film thickness is specified and the theoretical reflectance R' values for each wavelength are determined. In this case, the expected value of the transmitted light quantity IT' measured by the area sensor 23 can be estimated based on the theoretical reflectance R' for each wavelength and the spectral characteristics SCT_xm,yn(λ) of the transmission side of the film thickness measuring device 1 for each wavelength. Since the area sensor 23 does not have a spectral function, the expected value of the transmitted light quantity IT' measured by the area sensor 23 is the integrated value of the light intensity for each wavelength. Similarly, the expected value of the reflected light quantity IR' measured by the area sensor 24 can be estimated based on the theoretical reflectance R' for each wavelength and the spectral characteristics SCR_xm,yn(λ) of the reflection side of the film thickness measuring device 1 for each wavelength. Then, from equation (1) or (2) described above, the expected value of the wavelength centroid x' can be derived from the expected value of the transmitted light amount IT' and the expected value of the reflected light amount IR'. In this way, given a defined film type, the expected value of the wavelength centroid x' at a given film thickness can be derived. Then, the expected value of the wavelength centroid x' at each film thickness can be derived while varying the film thickness condition while using the same film type. This results in a state where the expected values of the wavelength centroid x' for multiple film thickness conditions have been derived for a given film type.

Figure 10 illustrates the derivation of the relational equation by curve fitting. Currently, for a certain type of film, the expected values of the wavelength centroid x' for multiple film thickness conditions have been derived. For example, in the example shown in Figure 10, the expected values of the wavelength centroid x' for three different film thicknesses (film thickness d = 90 nm, 100 nm, 110 nm) are plotted on a graph with the horizontal axis being the wavelength centroid x' and the vertical axis being the film thickness d. In reality, more expected values of the wavelength centroid x' for different film thickness conditions are plotted. Then, by performing curve fitting on each plotted data, a curve 360 showing the relationship between film thickness d and the wavelength centroid x' is derived. The relational equation that shows such a curve 360 is the relational equation between film thickness and measurement parameters. The relational equation between film thickness and measurement parameters is expressed as a polynomial, for example, as in equation (7) below. By determining each parameter (a, b, c, ...) of the polynomial, it becomes possible to derive the film thickness d if the wavelength centroid x' is input.
dxm, yn (x') = axm, yn + bxm, ynx' + cxm, ynx' ² +... (7)

Returning to Figure 1, the memory unit 33 stores the relationship between film thickness and wavelength centroid, as shown in equation (7) above, as the aforementioned relationship information. In this case, the relationship information is the relationship between film thickness and wavelength centroid derived by plotting and fitting the expected values of the wavelength centroid corresponding to each film thickness, which are derived based on the theoretical reflectance and the spectral characteristics of the film thickness measuring device 1. The memory unit 33 stores such relationship equations for each type of film.

The analysis unit 32 derives the film thickness of the wafer 100 based on the relationship between film thickness and wavelength centroid stored in the storage unit 33 as related information, and the wavelength centroid, which is a measurement parameter related to the wafer 100, obtained by the calculation unit 31. The analysis unit 32 derives the film thickness of the wafer 100 based on the relationship according to the type of film on the wafer 100 and the wavelength centroid obtained by the calculation unit 31 (analysis step). That is, the analysis unit 32 reads the relationship according to the type of film on the wafer 100 from the storage unit 33 (reading step), and derives the film thickness d of the wafer 100 by inputting the wavelength centroid obtained by the calculation unit 31 into the wavelength centroid x' of the relationship as shown in equation (7).

Figure 11 illustrates the derivation of film thickness using the relational equation. As shown in Figure 11, when the curve 360 showing the relationship between film thickness d and the wavelength centroid x' is derived and equation (7) is derived, the value of film thickness d can be uniquely derived from the value of the wavelength centroid x' obtained by the calculation unit 31.

As mentioned above, the measurement parameters may be various parameters that correlate with the film thickness other than the wavelength centroid x'. Figures 12(a) to (c) illustrate the derivation of the relational equations related to measurement parameters other than the wavelength centroid.

Figure 12(a) illustrates the derivation of the relationship when the ratio of the transmitted light quantity IT' measured by the area sensor 23 to the reflected light quantity IR' measured by the area sensor 24 is used as the measurement parameter. The expected value x'' of the ratio is derived by equation (8) below.
x'' = IT'' / IR'' (8)
By deriving multiple expected values x'' of the ratio while changing the film thickness conditions for the same type of film, plotting the expected values x'' of each ratio, and performing curve fitting, a curve 460 showing the relationship between film thickness d and ratio x'' is derived, and the following relational equation (9) showing the curve 460 is derived.
dxm,yn(x'')=axm,yn+bxm,ynx''+cxm,ynx''2+... (9)

Similarly, when the transmitted light amount IT' is used as the measurement parameter, multiple expected values x'''' of the transmitted light amount IT' are derived using the same film type but with varying film thickness conditions. By plotting each expected value x'''' of the transmitted light amount IT' and performing curve fitting, a curve 560 showing the relationship between film thickness d and transmitted light amount x'''' is derived (see Figure 12(b)). The following relational equation (10) showing the curve 560 is then derived.
dxm,yn(x´´´)=axm,yn+bxm,ynx´´´+cxm,ynx´´´2+... (10)

Similarly, when the reflected light intensity IR' is used as the measurement parameter, multiple expected values x'''' of the reflected light intensity IR' are derived using the same film type but with varying film thickness conditions. By plotting each expected value x'''' of the reflected light intensity IR' and performing curve fitting, a curve 660 showing the relationship between film thickness d and reflected light intensity x'''' is derived (see Figure 12(c)). The following relational equation (11) showing the curve 660 is then derived.
dxm,yn(x´´´´)=axm,yn+bxm,ynx´´´´+cxm,ynx´´´´2+... (11)

In the above embodiment of the film thickness measuring device 1, it was explained that measurement parameters and film thickness are derived by imaging the light transmitted and reflected by the inclined dichroic mirror 22, respectively, with the area sensors 23 and 24. This disclosure is not limited to this embodiment; for example, the film thickness of an object may be measured using a film thickness measuring device that does not include the inclined dichroic mirror 22.

Figure 13 schematically shows a modified film thickness measuring device 1B. Note that the control device 30 (calculation unit, storage unit, analysis unit) is omitted from the illustration in Figure 13. As shown in Figure 13, the film thickness measuring device 1B has a camera system 20B instead of the camera system 20 in the film thickness measuring device 1. The camera system 20B does not have a tilted dichroic mirror and has only one area sensor. Specifically, the camera system 20B has a lens 21 and one area sensor 23B. In this configuration, the area sensor 23B can measure the amount of light (the intensity of light from the wafer 100). Therefore, for example, a relational expression showing the relationship between such light amount and film thickness is pre-stored as the relational information described above, and the film thickness can be derived based on this relational expression and the measured light amount. In this case as well, the relational expression (relational information) is derived based on the theoretical reflectance according to the type of film and the spectral characteristics of the entire film thickness measuring device 1B.

Furthermore, although the above embodiment described an example of measuring the film thickness of a wafer 100 in which only one film 100b is formed on the surface of the substrate 100a, the invention is not limited to this, and the film thickness of a wafer with a multilayer film structure in which two or more films are formed on the surface of the substrate may also be measured.

Figure 14 is a schematic cross-sectional view of a wafer with a multilayer film structure according to a modified example. As shown in Figure 14, the wafer 200 has a substrate 200a, a film 200b, and a film 200c. Film 200b is the first layer laminated on the substrate 200a. Film 200c is the second layer laminated on film 200b. In other words, the second layer is the surface layer of the wafer. Here, as an example, we will explain assuming that film 200b is a silicon dioxide (SiO2) film and film 200c is a silicon nitride (SiN) film, but the type of film is not limited to these.

The following describes an example of a procedure for measuring the film thickness of a multilayer wafer 200 as shown in Figure 14. This film thickness measurement includes a first layer derivation step, a second layer deposition step, a wavelength centroid derivation step after second layer deposition, and a second layer derivation step. These steps are performed in sequence.

In the first layer derivation step, a wafer with only the first layer, film 200b, deposited is prepared, and film thickness measurement (i.e., film thickness measurement of film 200b) and evaluation are performed on this wafer. The derivation of the film thickness of only the first layer can be performed using the method described in the above embodiment. Figure 15(a) shows the distribution of the wavelength centroid measured for the first layer, and Figure 15(b) shows the distribution of the film thickness for the first layer derived based on the wavelength centroid shown in Figure 15(a). Thus, by deriving the distribution of the wavelength centroid of the first layer using the method described in the embodiment, the distribution of the film thickness of the first layer can be derived.

The second layer deposition process is performed following the first layer extraction process. In the second layer deposition process, the second layer, film 200c, is deposited so as to be stacked on film 200b. This completes the wafer 200 as shown in Figure 14.

The wavelength centroid derivation process after the second layer deposition is performed immediately following the second layer deposition process. In the wavelength centroid derivation process after the second layer deposition, the wavelength centroid after the second layer deposition is derived based on the imaging results of light from the wafer 200, similar to the wavelength centroid derivation in the above embodiment. Figure 16 shows the distribution of the wavelength centroid measured after the second layer deposition.

The second layer derivation process is performed following the wavelength centroid derivation process after the second layer deposition. Figure 17 illustrates the derivation of the second layer thickness based on the wavelength centroid measured after the second layer deposition and the thickness of the first layer. As shown in Figure 17, in the second layer derivation process, the thickness of the second layer is derived based on the wavelength centroid after the second layer deposition, which was measured (derived) in the wavelength centroid derivation process after the second layer deposition, and the thickness of the first layer, which was measured (derived) in the first layer derivation process.

As a prerequisite for performing the second layer derivation process, the memory unit 33 stores information on the relationship between the second layer thickness and the wavelength centroid (measurement parameter) for each combination of the first layer film type and thickness, and the second layer film type. Figure 18 illustrates the relationship between the second layer thickness and the wavelength centroid for each first layer thickness. In Figure 18, "SiO2 film thickness" refers to the first layer thickness. As shown in Figure 18, the relationship between the second layer thickness and the wavelength centroid changes depending on the first layer thickness. Therefore, by pre-storing information on the relationship between the second layer thickness and the wavelength centroid (see Figure 18) for each combination of the first layer film type and thickness, and the second layer film type, the first layer thickness can be measured with high accuracy.

The analysis unit 32 derives the film thickness of film 200c on wafer 200 based on relationship information corresponding to the combination of film type and film thickness of film 200b on wafer 200, and film type of film 200c, as well as the wavelength centroid of wafer 200 obtained by the calculation unit 31.

Thus, by defining the relationship information between the film thickness of the second layer and the measurement parameters for each combination of the type and film thickness of the first layer and the type of film of the second layer, it is possible to perform highly accurate and simple film thickness measurement of film 200c when the type and film thickness of the first layer (film 200b), the type of the second layer (film 200c), and the wavelength centroid are known.

The following section provides a detailed explanation of how to derive the film thickness (thickness distribution) of the second layer in the multilayer wafer 200 described above.

First, we will explain the manufacturing process of a semiconductor device, from the various processes performed on the wafer 200 to the production of the semiconductor device (semiconductor chip) 600. Figures 19(a) to (i) and 20(a) to (j) illustrate the manufacturing process of the semiconductor device 600. Here, all states during the manufacturing process of the semiconductor device 600 are described as the wafer 200. Note that the following manufacturing process is merely an example.

As shown in Figure 19(a), first, a mirror-finish silicon substrate 210 is manufactured by, for example, slicing and polishing a single-crystal ingot. Subsequently, as shown in Figure 19(b), an oxide film 211 and a nitride film 212 are deposited on the silicon substrate 210. These films may function as insulating films in the semiconductor device 600, or they may function as masks during the etching process.

Next, as shown in Figure 19(c), a resist 213 (photoresist) is applied to the nitride film 212 by spin coating or the like. The photoresist functions as an etching mask in the photolithography process. Then, as shown in Figure 19(d), UV light is irradiated onto the wafer 200 through a mask 400 on which the circuit pattern is drawn. This changes the solubility of the resist 213, allowing for partial removal of the resist 213 in the development process described later.

Next, as shown in Figure 19(e), the excess resist 213 in the development process is dissolved, forming a mask for etching the circuit pattern. That is, the resist 213 remaining in the development process becomes a mask during the etching process. Then, as shown in Figure 19(f), the excess is removed by etching using the resist 213 as a mask, forming the circuit.

Next, as shown in Figure 19(g), unwanted resist is stripped and removed by a wet method or ashing method. Then, as shown in Figure 19(h), an oxide film 214 is deposited in the grooves formed by etching. This oxide film 214 serves as an insulating film for electrically insulating the semiconductor elements.

Next, as shown in Figure 19(i), planarization is performed by polishing away unwanted portions of the wafer 200 using methods such as mechanical chemical polishing (CMP). Then, as shown in Figure 20(a), the silicon is oxidized by thermal oxidation, forming a gate oxide film 211A from the oxide film 211. After nitriding the surface of the gate oxide film 211A, the gate electrode layer 215 is formed by chemical vapor deposition (CVD).

Next, as shown in Figure 20(b), a pattern is imprinted on the gate electrode layer 215 by photography, forming the source and drain regions. Then, as shown in Figure 20(c), dopants 216 are implanted into the source and drain regions by ion implantation. Ions are not implanted in the areas where the gate oxide film 211A remains. Afterward, the dopants 216 are activated by recovery heat treatment.

Next, as shown in Figure 20(d), an oxide film 217 is deposited by chemical vapor deposition (CVD). The oxide film 217 functions as an insulating layer. Subsequently, as shown in Figure 20(e), contact holes 218 are formed by photolithography. These contact holes 218 form source and drain electrodes and serve as holes for electrical connection.

Next, as shown in Figure 20(f), electrode metal 219 is deposited in the contact hole to form the contact 220. At this point, any unwanted film is removed by mechanical chemical polishing (CMP). Subsequently, as shown in Figure 20(g), a trench 221 is formed by photolithography. The trench becomes a resist opening for electrical wiring.

Next, as shown in Figure 20(h), a metal film 222 is deposited to fill the trench 221. Unnecessary film is removed by mechanical chemical polishing (CMP). Subsequently, as shown in Figure 20(i), the formation of trenches, deposition of metal films, and planarization are repeated to form a three-dimensional wiring layer 223.

Finally, as shown in Figure 20(j), the semiconductor device (semiconductor chip) 600 is cut by dicing with a diamond blade or stealth dicing. This concludes the manufacturing process for the semiconductor device manufacturing method.

Figure 21 illustrates the procedure for deriving the film thickness (thickness distribution) of the second layer for a multilayer wafer 200. The derivation of the second layer's thickness distribution is performed, for example, after various processing treatments have been carried out on the wafer 200. These processing treatments are performed on a wafer 200 that has at least two or more layers stacked after the processing, and may be carried out after, for example, a film deposition treatment (Figure 19(b)), a photoresist coating treatment (Figure 19(c)), an etching treatment (Figure 19(f)), a planarization treatment (Figure 19(i)), an electrode formation treatment (Figure 20(a)), a pattern formation treatment (Figure 20(b)), an insulating film formation treatment (Figure 20(d)), a contact hole formation treatment (Figure 20(e)), a contact formation treatment (Figure 20(f)), a trench formation treatment (Figure 20(g)), or a wiring formation treatment (see Figure 20(i)). The following explanation will describe an example of deriving the second layer's film thickness for a wafer 200 after a film deposition treatment.

Now, as shown in Figure 21(a), a wafer 200 is prepared with only the first layer, film 200b, deposited, before the second layer is formed. With the first layer, film 200b, formed on the surface side of the wafer 200, the thickness distribution of film 200b is obtained (first step (corresponding to the first layer extraction step described above)).

Next, as shown in Figure 21(b), a predetermined processing treatment (in this case, film deposition) is performed on film 200b to form the second layer, film 200c (second step (corresponding to the second layer deposition step described above)).

Then, as shown in Figure 21(c), after the second step, light is shone planarly onto the wafer 200 on which the film 200c has been formed, and the light from the wafer 200 is imaged. Based on the signal from the image, measurement parameters (e.g., wavelength centroid) within the plane of the wafer 200 are derived (third step (corresponding to the wavelength centroid derivation step described above)).

Finally, based on the thickness distribution of film 200b obtained in the first step and the measurement parameters (e.g., wavelength centroid) derived in the third step, the thickness distribution of film 200c is derived (fourth step (corresponding to the second layer derivation step described above)).

In addition to the first to fourth steps described above, a relational information derivation step may be performed, for example, during the execution of the second step. The relational information derivation step is a process of deriving relationship information between the measurement parameters and the thickness distribution of the film 200c, based on the thickness distribution of the film 200b acquired in the first step. If relational information has been derived, the fourth step derives the thickness distribution of the film 200c based on the relational information and the measurement parameters derived in the third step. The first to fourth steps and the relational information derivation step will be described in detail below. In the following description, the configuration of Figure 1 will be referred to as appropriate, but the wafer 100 in Figure 1 will be replaced with a multilayer wafer 200.

In the first step, the thickness distribution of the first layer, film 200b, may be derived using the method described above. Specifically, in the film thickness measuring device 1, the light source 10 irradiates light onto the wafer 200 in a planar manner, area sensors 23 and 24 image the light from the wafer 200, the calculation unit 31 of the control device 30 derives measurement parameters (in this case, the wavelength centroid) based on the signals from the area sensors 23 and 24, and the analysis unit 32 derives the film thickness (thickness distribution) of film 200b based on the relational information stored in the storage unit 33 and the measurement parameters obtained by the calculation unit 31. In this case, the relational information is derived based on the theoretical reflectance according to the type of film and the spectral characteristics (system spectral sensitivity) of the entire film thickness measuring device 1. The theoretical reflectance and spectral characteristics here are as described above.

In the second step, after the completion of the first step, a second layer, film 200c, is formed so as to be laminated on film 200b. The related information derivation step is performed simultaneously during the execution of the second step. While this explanation assumes the related information derivation step is performed during the second step, the related information derivation step may also be performed before or after the completion of the second step.

In the related information derivation process, specifically, based on the thickness distribution of film 200b acquired in the first step and the thin-film information of film 200c acquired in advance, the theoretical reflectance for each predetermined region (for example, each pixel) is derived for each thickness distribution of film 200c. Thus, the theoretical reflectance described here is the "theoretical reflectance for each thickness distribution of film 200c, which is the second layer." "For each thickness distribution of film 200c" can be rephrased as "for each candidate film thickness of film 200c." As mentioned above, once the type of film (thin-film information such as the refractive index and extinction coefficient of the film) and the film thickness are determined, the theoretical reflectance value for each wavelength is determined. For example, the storage unit 33 stores the thickness distribution of film 200b acquired in the first step and the thin-film information of film 200c acquired in advance. The calculation unit 31 then refers to the storage unit 33 to identify the thin film information of the film 200c, and derives the theoretical reflectance for each candidate film thickness (thickness distribution) of the film 200c according to this thin film information.

Figure 22 illustrates the derivation of theoretical reflectance for each candidate second layer thickness. In the example shown in Figure 22, under the condition that the film type is constant, the theoretical reflectance for each wavelength is derived for the second layer thickness candidate "d1", for the second layer thickness candidate "d2", ..., for the second layer thickness candidate "dn". In this way, the theoretical reflectance is derived for each second layer thickness candidate.

In the related information derivation process, measurement parameters (e.g., wavelength centroid) are derived for each predetermined region (for example, each pixel) for each thickness distribution (candidate film thickness) of the second layer, based on the theoretical reflectance for each thickness distribution of the second layer and the previously acquired spectral characteristics. For example, the storage unit 33 stores the theoretical reflectance for each thickness distribution of the second layer and the previously acquired spectral characteristics of the film thickness measuring device 1. The calculation unit 31 then derives the expected value of the measurement parameter (e.g., wavelength centroid) for each thickness distribution of the second layer by referring to the storage unit 33.

Figure 23 illustrates the derivation of the wavelength centroid, a measurement parameter (more specifically, the derivation of the expected value of the wavelength centroid). Since the film type is now determined, the theoretical reflectance values for each wavelength are determined according to the film thickness of the second layer. For example, suppose the theoretical reflectance R values for each wavelength are determined for the candidate second layer thickness "d1". In this case, the expected value of the transmitted light quantity IT' measured by the area sensor 23 can be estimated based on the theoretical reflectance R for each wavelength and the spectral characteristics SCT_xm,yn(λ) of the transmission side of the film thickness measuring device 1 for each wavelength. Since the area sensor 23 does not have a spectral function, the expected value of the transmitted light quantity IT' measured by the area sensor 23 is the value obtained by integrating the light intensity for each wavelength. Similarly, the expected value of the reflected light quantity IR' measured by the area sensor 24 can be estimated based on the theoretical reflectance R for each wavelength and the spectral characteristics SCR_xm,yn(λ) of the reflection side of the film thickness measuring device 1 for each wavelength. Then, using equation (1) or (2) above, the expected value of the wavelength centroid can be derived from the expected value of the transmitted light amount IT' and the expected value of the reflected light amount IR'. As shown in Figure 23, the expected value of the wavelength centroid is derived for each candidate thickness of the second layer, while varying the candidate thicknesses "d2", "d3", ... "dn". Through this process, the expected value of the measurement parameter (in this case, the wavelength centroid) can be derived for each thickness distribution of the second layer.

In the relationship information derivation process, relationship information is further derived showing the relationship between the measurement parameters and the thickness distribution of the second layer for each predetermined region (for example, each pixel), based on the expected values of the measurement parameters (in this case, the wavelength centroid) for each thickness distribution of the second layer. For example, the storage unit 33 stores the derived expected values of the measurement parameters (in this case, the wavelength centroid) for each thickness distribution of the second layer. Then, the calculation unit 31 refers to the storage unit 33 to identify the expected values of the measurement parameters for each thickness distribution of the second layer and derives relationship information between the measurement parameters and the thickness distribution of the second layer.

Figure 24 illustrates the derivation of relationship information between measurement parameters and the thickness distribution of the second layer. Currently, the expected value of the wavelength centroid is derived for multiple film thickness conditions of the second layer. For example, in the example shown in Figure 24, the expected value of the wavelength centroid is derived for each of the candidate second layer film thicknesses "d1", "d2", ... "dn", and plotted on a graph with the x-axis representing the wavelength centroid and the y-axis representing the film thickness of the second layer. While only three patterns are plotted in the example shown in Figure 24, in reality, many more expected values of the wavelength centroid with different film thickness conditions would be plotted. Then, by performing curve fitting on each plotted data, a curve 560 showing the relationship between film thickness and the wavelength centroid is derived. The relational equation showing such a curve 560 is the relationship between the measurement parameters and the thickness distribution of the second layer. This relational equation is expressed as a polynomial, as in equation (7) above. By determining each parameter (a, b, c, ...) of the polynomial, it becomes possible to derive the film thickness d when the wavelength centroid x' is input.

As described above, the relational information may be relational formulas obtained through fitting or other means for each predetermined area (for example, each pixel). Alternatively, the relational information may be a table instead of relational formulas. The relational information is stored in the storage unit 33.

In the third step, in the film thickness measuring device 1, the light source 10 irradiates light onto the wafer 200 on which the film 200c is formed in a planar manner, the area sensors 23 and 24 capture images of the light from the wafer 200, and the calculation unit 31 of the control device 30 derives measurement parameters (in this case, the wavelength centroid) based on the signals from the area sensors 23 and 24.

In the fourth step, the thickness distribution of the second layer is derived based on the relational information (relationship information between the wavelength centroid and the thickness distribution of the second layer) derived in the relational information derivation step, based on the thickness distribution of the film 200b acquired in the first step, and the wavelength centroid derived in the third step. For example, the storage unit 33 stores the relational information derived in the relational information derivation step and the wavelength centroid derived in the third step. Then, the analysis unit 32, by referring to the storage unit 33, derives the thickness distribution of the second layer (film 200c) corresponding to the wavelength centroid derived in the third step.

Figure 25 illustrates the derivation of the thickness distribution of the second layer. As shown in Figure 25, when a curve 560 showing the relationship between the film thickness (thickness distribution of the second layer) d and the measurement parameter, the wavelength centroid x', is derived, and the relationship shown in equation (7) above is derived, the value of the film thickness (thickness distribution of the second layer) d can be uniquely derived from the value of the wavelength centroid x'.

Next, the effects and advantages of the semiconductor device manufacturing method according to this embodiment will be described.

In recent years, with the increasing integration of semiconductor devices, ensuring uniformity of the thickness of each layer during the manufacturing process has become crucial. Figure 26 illustrates the challenges of a semiconductor device 800 related to a comparative example. As shown in Figure 26, if the thickness of each layer is not uniform during integration, it can lead to failures such as wiring defects and void formation. With the miniaturization of semiconductor devices, the thickness of each layer has become extremely thin, for example, less than 100 nm, requiring high-precision measurement of thickness. One known method for measuring thickness involves detecting reflected interference light from the semiconductor device with a spectrometer to obtain a spectral distribution and estimate the thickness of each layer. However, this measurement method measures thickness at specific points, and the measurement time becomes extremely long when, for example, it is necessary to derive the thickness distribution across the entire wafer with high precision. Therefore, measuring the thickness distribution across the entire wafer is not practical, and instead, point measurements of the thickness at predetermined intervals along the wafer's scribe lines (lines used for division by a dicer, etc.) have been performed.

In this regard, the semiconductor device manufacturing method according to this embodiment comprises: a first step of acquiring the thickness distribution of the first layer while the first layer is formed on the surface side of the wafer 200; a second step of forming a second layer by performing a predetermined processing on the first layer; a third step of irradiating the wafer 200 with light in a planar manner after the second step, imaging the light from the wafer 200, and deriving measurement parameters within the plane of the wafer 200 based on the imaging signal; and a fourth step of deriving the thickness distribution of the second layer based on the thickness distribution of the first layer acquired in the first step and the measurement parameters derived in the third step.

In the semiconductor device manufacturing method according to this embodiment, after the formation of the second layer, light is irradiated onto the wafer 200 in a planar manner, and the light from the wafer 200 is imaged, thereby deriving measurement parameters within the plane of the wafer 200. Then, the thickness distribution of the second layer is derived based on the thickness distribution of the first layer obtained before the formation of the second layer and the measurement parameters within the plane of the wafer 200 derived after the formation of the second layer. In this way, because the thickness distribution of the first layer is obtained in advance before the formation of the second layer, the thickness distribution of the second layer, which is the surface layer of the wafer, can be appropriately (with high accuracy) derived based on the measurement parameters within the plane of the wafer 200 derived after the formation of the second layer and the thickness distribution of the first layer. Furthermore, since the thickness distribution of the second layer is derived by taking into account the image results of the light irradiated onto the wafer 200 in a planar manner, the thickness distribution of the second layer within the entire plane of the wafer 200 can be derived all at once (in a short period of time). This significantly reduces the measurement time required to derive the in-plane thickness distribution of the entire wafer 200, compared to, for example, the case where the thickness of the second layer is measured at specific points using a spectrometer or the like. As described above, the semiconductor device manufacturing method according to this embodiment allows for the rapid and highly accurate deriving of the thickness distribution of the wafer 200. Furthermore, in the case of point measurement as in the comparative example, it is necessary to save spectral data at each measurement point for manufacturing records, making data management complicated. However, with the measurement method according to this embodiment, only the reflection and transmission images need to be recorded, thus simplifying data management.

Figure 27 illustrates the effects of the thickness distribution derivation according to this embodiment. Figure 27(a) shows the correct data for the thickness distribution of the second layer, Figure 27(b) shows the derivation result (film thickness analysis result) of the thickness distribution of the second layer, and Figure 27(c) shows the error between the correct data and the derivation result for the thickness distribution of the second layer. As shown in Figure 27(c), the error between the correct data and the derivation result for the thickness distribution of the second layer can be kept within ±1 nm, demonstrating that the thickness distribution of the wafer surface layer can be derived with high accuracy.

The semiconductor device manufacturing method further includes a relationship information derivation step, which derives relationship information between measurement parameters and the thickness distribution of the second layer based on the thickness distribution of the first layer obtained in the first step. In the fourth step, the thickness distribution of the second layer may be derived based on the relationship information derived in the relationship information derivation step and the measurement parameters derived in the third step. In this way, by deriving and using relationship information between the measurement parameters and the thickness distribution of the second layer under the conditions of the obtained thickness distribution of the first layer, the thickness distribution of the wafer surface layer can be derived quickly and accurately from the measurement parameters derived in the third step.

In a semiconductor device manufacturing method, the relationship information derivation step may involve deriving the theoretical reflectance for each thickness distribution of the second layer based on the thickness distribution of the first layer obtained in the first step and the thin-film information of the second layer obtained in advance, and deriving measurement parameters for each thickness distribution of the second layer based on the theoretical reflectance for each thickness distribution of the second layer and the spectral characteristics obtained in advance, and deriving relationship information between the measurement parameters and the thickness distribution of the second layer. In this way, by specifying the thickness distribution of the first layer and the thin-film information of the second layer, the theoretical reflectance for each thickness distribution of the second layer can be derived with high accuracy. Furthermore, by specifying the theoretical reflectance and spectral characteristics for each thickness distribution of the second layer, measurement parameters for each thickness distribution of the second layer can be derived, and the aforementioned relationship information can be derived with high accuracy. With this configuration, the thickness distribution of the wafer surface layer can be derived with high accuracy from the measurement parameters derived in the third step using the relationship information.

In a semiconductor device manufacturing method, the related information derivation step may be performed during the execution of the second step. This allows for the rapid derivation of the thickness distribution of the wafer surface layer after the execution of the second step.

In semiconductor device manufacturing methods, the relational information may be a relational expression obtained through fitting, or a table. By using such relational information, the thickness distribution of the wafer surface layer can be derived quickly and with high accuracy.

In semiconductor device manufacturing methods, the measurement parameter may be the wavelength centroid. With this configuration, the thickness distribution of the wafer surface layer can be accurately derived by using the wavelength centroid, which has a high correlation with film thickness, as the measurement parameter.

In a semiconductor device manufacturing method, the processing steps may include any of the following: film deposition, photoresist coating, etching, planarization, electrode formation, pattern formation, insulating film formation, contact hole formation, contact formation, trench formation, or wiring formation on the wafer 200. The thickness distribution of the second layer formed after these processes can be derived quickly and accurately using the method described above. However, after pattern formation, conventional point measurement for film thickness is difficult due to difficulties in alignment, etc. In this respect, the method according to this embodiment eliminates the need for alignment, etc., after pattern formation, allowing for easy deriving of the thickness distribution of the wafer surface layer.

In this embodiment, for convenience, an example of a two-layer film structure has been described. However, the contents of this disclosure may also be applied to the manufacture of semiconductor devices having three or more multilayer films. That is, the method of this disclosure may be used for measuring the film thickness (derivation of thickness distribution) of the outermost layer in a multilayer film of three or more layers. In this case, the "first layer" in this disclosure refers to each layer located below the outermost layer, and the "second layer" refers to the outermost layer.

200...Wafer, 200b...Film (First Layer), 200c...Film (Second Layer), 600...Semiconductor Device.

Claims (6)

A first step involves irradiating the wafer with light in a planar manner while a first layer is formed on the surface side of the wafer, imaging the light from the wafer, and obtaining the thickness distribution of the first layer within the plane of the wafer based on the imaging signal.
A second step involves forming a second layer by performing a predetermined processing treatment on the first layer,
A third step is performed after the second step, in which light is irradiated onto the wafer in a planar manner, the light from the wafer is imaged, and the distribution of measurement parameters within the wafer surface is derived based on the signal obtained from the image.
The method comprises a fourth step of deriving the thickness distribution of the second layer based on the thickness distribution of the first layer obtained in the first step and the distribution of the measurement parameters derived in the third step ,
The process further includes a relationship information derivation step, which derives relationship information showing the relationship between the measurement parameters and the thickness of the second layer for each predetermined region, based on the thickness distribution of the first layer obtained in the first step.
A semiconductor device manufacturing method comprising the fourth step of deriving the thickness distribution of the second layer based on the relational information derived in the relational information derivation step and the distribution of the measurement parameters derived in the third step . In the aforementioned relationship information derivation process,
Based on the thickness distribution of the first layer obtained in the first step and the thin film information of the second layer obtained in advance, the theoretical reflectance for each predetermined region is derived for each thickness distribution of the second layer.
A semiconductor device manufacturing method according to claim 1, comprising: deriving the measurement parameters for each predetermined region for each thickness distribution of the second layer based on the theoretical reflectance for each thickness distribution of the second layer and the spectral characteristics obtained in advance; and deriving the relational information that shows the relationship between the measurement parameters and the thickness of the second layer for each predetermined region. The semiconductor device manufacturing method according to claim 2 , wherein the related information derivation step is performed during the execution of the second step. The semiconductor device manufacturing method according to claim 1 , wherein the relational information is a relational expression or a table obtained by fitting for each predetermined region. The semiconductor device manufacturing method according to claim 1, wherein the measurement parameter is the wavelength centroid. The semiconductor device manufacturing method according to any one of claims 1 to 5, wherein the processing is one of the following processes applied to the wafer: film deposition, photoresist coating, etching, planarization, electrode formation, pattern formation, insulating film formation, contact hole formation, contact formation, trench formation, or wiring formation.