JP2001021317A - Measuring method, measuring apparatus, polishing method and polishing apparatus - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a measuring apparatus which can precisely, simply, and quickly specify measuring pattern species, measure a film thickness and process end points in a film eliminating process and a film forming process, for a substrate having a pattern structure constituted of various pattern species, a substrate in an STI(shallow trench isolation) process and a substrate in a Cu process in which a barrier metal layer exists. SOLUTION: This measuring apparatus is equipped with an optical measurement part which casts a probe light 3 having a plurality of wavelength components on a substrate surface, and obtains a signal wave form of a signal light extracted from a light reflected from the substrate 1 surface or a light which has permeated the surface, and a signal processing part having at least a function which specifies a measurement position 30 on the substrate surface by comparing the signal wave form with a wave form which is previously calculated or measured.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a measuring method and a measuring apparatus, and more particularly, to a method and an apparatus for measuring the thickness of a semiconductor wafer or measuring a process end point in a film removing step or a film forming step.
And a method and a polishing apparatus for polishing a wafer in a semiconductor device manufacturing process.

[0002]

2. Description of the Related Art Densification of semiconductor devices (semiconductor devices) has been progressing without showing a limit, and as the densification has increased, multilayer wiring and the accompanying formation of interlayer insulating films and metal such as plugs and damascenes have been developed. The importance of electrode formation techniques has greatly increased. Naturally, the thickness and shape of such interlayer insulating film and metal electrode film (whether or not they are properly embedded)
Is a big challenge. Of course, monitoring of film thickness is also required in processes such as thin film formation and etching,
Of particular importance recently is the detection (measurement) of the process end point in the planarization process.

In consideration of the reduction in the depth of focus during exposure, which accompanies the shortening of the exposure wavelength in recent lithography processes, in order to secure the resolution of exposure, at least the area of the interlayer layer in the range of the exposure area is required. The demand for flattening accuracy is great. In the so-called inlay (plug, damascene) in which a metal electrode layer is buried, it is required to remove and flatten an extra metal layer after forming the metal layer.

Many methods have been proposed and implemented for locally smoothing an interlayer layer on a wafer surface by improving a film forming method and the like. However, as an efficient flattening technique for a wider wafer area, CMP is used. There is a polishing process called CMP (Ch
emical Mechanical Polishing or Planarization)
Is a process of removing irregularities on the wafer surface of a semiconductor element by using a physical action in combination with a chemical action (dissolution with an abrasive solution). It has become. Specifically, using an abrasive called a slurry in which abrasive grains (typically silica, alumina, cerium oxide, etc.) are dispersed in a soluble solvent of an object to be polished such as an acid or an alkali, and using an appropriate polishing cloth. Then, the polishing is advanced by pressing the wafer surface and rubbing by relative motion. By making the pressing force and the speed of the relative movement uniform over the entire surface of the wafer, uniform polishing can be performed over the entire surface of the wafer.

[0005] In this polishing step, the stability and reproducibility of the process are generally not as good as those of the film forming step and the etching step. Therefore, in order to solve the problem and to improve the efficiency of the polishing step, information necessary for the polishing step is provided. It is necessary to provide quick feedback, and for that purpose, it is required to constantly measure the film thickness of the interlayer and metal layers. Conventionally, a general film thickness measuring device is often used for measuring the film thickness. A minute blank portion (two-dimensional film thickness distribution, that is, a position where there is no pattern structure) of the wafer that has been cleaned after polishing is selected as a measurement position and measured by various methods. The film thickness of a portion that is actually desired to be known (it is often impossible to measure because of a pattern structure portion having a fine structure) is calculated by calibrating from this value. This method has a problem that feedback is slow.

[0006] In the polishing process for planarization, a method of measuring the film thickness with faster feedback is to measure the fluctuation in friction when polishing proceeds from the polishing layer to be removed to a different layer, by rotating the wafer or the pad. There is a method of measuring by the change of the motor torque. Also, an optical path is provided on the polishing pad, or light is radiated to the film surface using light transmitted through the wafer (infrared light) from the back surface of the wafer, and the thickness of the thin film being polished is polished by optical interference. A method for measuring is also proposed.

[0007]

A technique for quickly and simply measuring the thickness of an interlayer film, a metal layer, or the like in the above-described CMP process or the like is not definitive, despite increasing demands. . In the measurement with the above-mentioned film thickness measuring machine, the wafer which has been cleaned after polishing is transferred and held in a sufficiently stable state such as holding a stage, and then measured. At present, sufficient accuracy and reliable data can be obtained, but the apparatus itself becomes large-scale, and it takes a long time to obtain a measured value, so that feedback to the process is delayed.

Further, as a major problem, there is a problem of setting a measurement position of a film thickness on a wafer. In a semiconductor device wafer having a pattern structure, the film thickness must be measured by searching for a portion having no pattern in order to set a portion having no pattern as a measurement position. Generally, a portion having no pattern has a large area. And the position is not constant depending on the type of the semiconductor device wafer.

First, it is not easy for a device to reduce the measurement range. Further, it is not easy to quickly search and measure a small portion. In order to search for a small portion at high speed, it is necessary to have a complicated mechanism for capturing an image of a pattern, recognizing and processing the small portion, which is hard (imaging device, precision alignment mechanism, etc.) and software (image Both processing software) have a large load and are expensive. Even if it can be realized, the time for image processing, position search and positioning will greatly increase the measurement time. Further, as described above, the film thickness thus obtained is a calibration value, and the evaluation of the pattern dependence described below requires a complicated calibration (calibration) for this value. There is a problem that it does not.

Although the method based on the change in motor torque is simple and fast, it can detect only the start of polishing of a different layer, and thus is effective for detecting the end point of the process, but is not effective for measuring the film thickness. Moreover, the detection accuracy is insufficient. Further, in a method of measuring a film thickness using optical interference (a method of irradiating a laser beam and tracking a time variation of a reflected light amount) or the like, a measurement position cannot be specified. The uncertainty depending on the measurement and the error due to the measurement position are pointed out. As a memory element such as a D-RAM, the pattern may be regarded as substantially uniform (one pattern type) as a continuation of the periodic structure. In some cases, this problem can be solved. In the device, the pattern is not uniform, and the problem due to the measurement position becomes significant. Particularly, in the case of the CMP process, polishing depends on the pattern type, that is, the polishing rate and the manner of eliminating the step differ greatly depending on the pattern density and definition, so it is important to evaluate the film thickness for each different pattern type. become.

Further, in the detection of the polishing end point in the CMP step in the STI step, the method of detecting the end point by the motor torque is not effective enough to reach a practical level, although there is a stopper layer. (In general, it is also reported that detection with the aging of the polishing pad becomes difficult.) Also, in the current optical methods such as laser interference,
Both constituent layers such as SiO2 and SixN are transparent films,
At the end point of the polishing step, a sufficient change is not seen as a change in the light amount, and the STI, which requires higher control accuracy than the process of the interlayer dielectric film, may not have sufficient measurement (end point determination) ability. It is pointed out. For this reason, in an actual process, control is often performed by controlling the polishing time.

[0012] Furthermore, Cu having a barrier metal layer is present.
For the measurement of the polishing end point of the substrate in the process, the method of detecting the polishing end point by the motor torque is not effective enough to reach a practical level. For this reason, in an actual process, it is inevitable that control is performed by controlling the polishing time. The present invention solves the above problems, and specifies a measurement position for a substrate having a pattern structure composed of various pattern types, and further for a substrate in an STI process and a substrate in a Cu process in which a barrier metal layer exists. A measuring apparatus and a measuring method and a measuring method and a measuring method capable of measuring a film thickness, and furthermore, measuring (detecting) a process end point in a film removing process and a film forming process accurately, simply and at a high speed. An object of the present invention is to provide a polishing apparatus using the apparatus and a polishing method using the measuring method.

[0013]

Various methods for optically measuring the thickness of a thin film on a substrate are known, and considerable accuracy is realized even in a system using an interference phenomenon. However, these are all for blank film measurement having no pattern structure (including a multilayer film). The present invention is directed to a wafer having a pattern structure on its surface, which is not two-dimensionally uniform. in this case,
A signal simply predicted from a uniform blank film having no pattern structure cannot be obtained.

In the present invention, in optical measurement in the case of having a pattern structure, light of multiple component wavelengths is used as probe light, and the wavelength dependence of reflected signal light, that is, spectral characteristics are examined to specify a pattern type. Then, if necessary, know the film thickness. In the present invention, light is irradiated with a spot diameter larger than the minimum unit of the pattern in order to specify the pattern type of the pattern structure and to measure the film thickness.

Further, a spatial coherence length controller is provided to control the coherence between the patterns inside the spot diameter of the light irradiated on the patterns. Furthermore, only the zero-order light component of the light that has been irradiated onto the pattern surface and returned is used as signal light, and the first-order and higher-order diffracted light is removed. The obtained signal light, the measurement position on the pattern is specified by comparing with the calculated or measured spectral characteristics,
Further, the film thickness is measured.

As described above, according to the present invention, the measurement position of the wafer is specified simply and accurately, and further, the film thickness measurement of the layer of the device semiconductor wafer and the detection (measurement) of the process end point are performed. Therefore, first, a step of irradiating a probe light having a plurality of wavelength components on a substrate surface having a pattern structure on its surface, extracting signal light from light reflected or transmitted from the substrate surface, and Acquiring a signal waveform of light, and identifying a measurement position (probe light irradiation position) in the substrate by comparing the signal waveform with a calculated or previously measured reference value. (Claim 1). "

Secondly, there is provided a "method of claim 1 (claim 2), wherein the calculation of the reference value is performed using the pattern density as a parameter.
Third, "the extracted signal light is characterized in that it contains only the specular reflection component (0th-order light) of the reflected or transmitted light of the probe light and does not contain the first-order or higher-order diffracted light component. To provide a measurement method according to any one of claims 1 to 2 (claim 3).

[0018] Fourthly, the measuring method according to any one of claims 1 to 3, further comprising a step of controlling a spatial coherence length of the probe light. 4) ”. Fifth, “the spot diameter on the irradiation surface of the probe light is larger than the minimum unit size of the pattern constituting the pattern structure,
2. The structure according to claim 1, wherein the size is smaller than the size of the block.
To (4) (Claim 5).

Sixth, "the spatial coherence length is adjusted to a condition that completely causes pattern interference or a condition that does not cause pattern interference at all."
(5) The method according to any one of (5) to (6). Seventhly, "the probe light is irradiated perpendicularly to the substrate surface.
Measurement method (Claim 7) ".

Eighth, "the comparison is performed using at least one selected from a cross-correlation coefficient, an auto-correlation coefficient, and a least squares method.
(7) The measuring method according to any one of (7) (claim 8). Ninth, the method further comprises the step of storing the calculated or pre-measured reference value as a data table in advance. Claim 9) "is provided.

In a tenth aspect, the method further comprises the step of: "moving a measuring point to a desired measuring position based on the specified result." Measurement method (Claim 10) "is provided. Eleventh, "the method further comprises the step of measuring the film thickness of the film on the substrate surface with respect to each of the measurement positions moved to the specified or measured points. The measurement method according to any one of claims (Claim 11) is provided.

In a twelfth aspect, the method further comprises the step of, when the film is placed in a film forming step or a removing step, measuring each film thickness transition at each of the measurement positions. The measuring method according to claim 11, wherein the measuring method is characterized in that:
I will provide a. In the thirteenth aspect, "the substrate is a wafer of a semiconductor element having a pattern structure on a surface, and the measurement is a step of forming or removing an insulating layer or an electrode layer on the wafer in a semiconductor device manufacturing process. The method according to any one of claims 11 to 12, further comprising the step of measuring (detecting) a process end point.

Fourteenth, "the removing step comprises a step of simultaneously polishing at least two or more layers transparent to the probe light."
3. The method according to claim 3 (claim 14). Also,
Fifteenth, "the removal step is performed by STI (Shallow Trench
Isolation) is a step for the purpose of the present invention.

In the sixteenth aspect, the measurement method according to any one of claims 14 and 15, wherein the calculation of the reference value is performed for a film thickness whose dishing amount has been corrected in advance. Item 16) "is provided. In the seventeenth aspect, the measuring method according to claim 13, wherein the removing step comprises simultaneously polishing at least two or more types of layers that are opaque or translucent to the probe light. (Claim 17).

Eighteenthly, there is provided a "measuring method according to claim 17, wherein the two or more types of layers include a Cu layer and a barrier metal layer."
Also, in the nineteenth, `` irradiating the substrate surface with a probe light having a plurality of wavelength components, an optical measurement unit that acquires a signal waveform from signal light extracted from light reflected or transmitted from the substrate surface, 19. A signal processing unit having at least a function of specifying a measurement position (probe light irradiation position) in a substrate by comparing a signal waveform with a previously calculated or measured waveform, and any one of claims 1 to 18. A measurement device (claim 19) for performing measurement by the measurement method according to claim 1 ".

Twenty-second, "the substrate is a semiconductor element wafer having a pattern structure on its surface, and the measurement is performed in a semiconductor device manufacturing process in which an insulating layer or an electrode layer is formed on the wafer. Alternatively, it is a measurement in the removing step, and further measures (detects) the end point of the step.

Also, a twenty-first method for polishing a semiconductor device wafer having a pattern structure on the surface and an insulating layer or an electrode layer on the outermost surface layer, comprising the steps of: 19. A step of polishing the insulating layer or the electrode layer of the wafer by relatively moving the polishing pad and the wafer with an abrasive interposed therebetween, and wherein the polishing pad and the wafer are relatively moved. Performing a measurement (detection) of the film thickness or the process end point by the measurement method described above.

In a twenty-second aspect, the present invention provides a measuring apparatus, a polishing pad, and a polishing head for holding a semiconductor element wafer having an insulating layer or an electrode layer on the outermost surface layer. In a state where a polishing agent is interposed between a polishing pad and the wafer, the polishing pad and the wafer are relatively moved to polish the insulating layer or the electrode layer of the wafer, and the measurement device Performs a measurement of the insulating layer or the electrode layer or a measurement (detection) of an end point of the polishing process (claim 22). "

[0029]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Embodiments according to the present invention will be described below with reference to the drawings for easy understanding, but the present invention is not limited to these drawings. FIG. 2 is a conceptual diagram showing the measuring apparatus of the present embodiment, and FIG. 3 is a detailed view of one example of the optical measuring section of FIG. FIG. 1 is a flowchart showing the operation of the measuring apparatus according to the present embodiment, and mainly shows the operation of the signal processing unit 6 in FIG.

In FIG. 2, reference numeral 1 denotes a substrate having a pattern structure on the surface to be measured, with the pattern surface facing upward. Reference numeral 2 denotes an optical measurement unit, 6 denotes a signal processing unit, and 8 denotes a display unit. In FIG. 3, reference numeral 1 denotes a substrate having a pattern structure on the surface to be measured, with the pattern surface facing downward. 9 is a light source that emits a plurality of wavelength component lights, 10 is a spatial coherence controller (variable diameter stop), 1
1 is an optical lens, 34 is a spot diameter adjusting unit, 12 is a beam splitter, 13 and 130 are optical lenses, 14 is a mirror, 15 is an optical lens, 35 is a zero-order light sorting unit, 16 is a spectral unit (diffraction grating), Reference numeral 17 denotes a light detection unit.

In the measuring apparatus shown in FIG. 3, light having a plurality of wavelength components emitted from a light source 9 passes through a spatial coherence control unit 10, passes through an optical lens 11, and has a field of view as a spot diameter adjusting unit. Part of the light passes through the aperture 34 and is transmitted by the beam splitter 12,
The light is collimated by the laser beam 3 and is irradiated as a probe light 3 onto the pattern surface 29 of the substrate 1. The reflected light 4 from the pattern surface 29 again passes through the optical lens 13, partially reflects the beam splitter 12, passes through the optical lens 130, reflects off the mirror 14, passes through the optical lens 15,
The first-order or higher order diffracted light is removed by the stop 35 as the next light selecting unit, and only the signal light is selected and incident on the spectroscopic unit 16. The signal light is split by the splitting unit 16, and light of different wavelengths is reflected in different angle directions. The light reflected in different angle directions enters the light detection unit 17 that can independently detect light incident on different positions corresponding to different angle directions. These lights are output from the light detection unit 17 as a signal 5 to the signal processing unit 6.

The signal processing section 6 operates as shown in FIG. When the process starts, signal data (R (t)) is obtained from the optical measurement unit 2 (S1). Before going to the next stage, whether the simulation calculation of the pattern surface of the wafer to be measured is performed in advance or the pattern surface is measured in advance (S7), the spectral waveform (R _k ) of the pattern surface is determined. It has been determined as a reference value (S8). This spectral waveform is obtained for each pattern type corresponding to each pattern density and each film thickness configuration,
K sets are determined for each film thickness of the uppermost layer.

Next, the similarity between the signal data (R (t)) and the spectral waveform (R _k ) is calculated (S2). Next, the value of k is reduced by 1 (S3). Next, it is determined whether or not k = 1 (S4). If k = 1 is not satisfied, the process returns to S2 to calculate the similarity between the signal data (R (t)) and the spectral waveform ( _Rk ), and repeats this loop through S3 until YES in S4. .

If YES in S4, that is, if k = 1, S5
go to. In S5, the spectral waveform (R _k ) having the best similarity is found, and the pattern type and the film thickness corresponding to the R _k are determined as the film thickness for the measurement position of the pattern surface to be measured.
Here, in FIGS. 2 and 3, the relationship between the spot diameter of the probe light 3, the minimum pattern size, and the spatial coherence length of the signal light is important.

Normally, the reflected light from the pattern structure portion irradiated with the probe light is applied to each layer of the device (laminated thin film),
This can be regarded as superposition of light waves from each part, and the waveform of the wavelength dependence (spectral characteristics) is significantly different from that of the blank film due to the complicated interference effect. It is not easy to analytically calculate the value of the film thickness to be measured from such a curve. As a result of various experiments, we found that the complexity of this interference effect depends on the pattern type of the pattern surface irradiated with the probe light, the spatial coherence length and spot diameter of the probe light, the definition of the pattern, the diffraction component, etc. It is discovered that the interrelationship is due to the diversity.

If these are not in a limited relationship, the signal light obtained from the reflected light is not determined even for the same pattern shape and the same film thickness, and as a result, the film thickness obtained by analyzing the signal light is also It is not determined. Therefore, the present invention provides a method for measuring a film thickness on a substrate having a pattern structure.
First, the pattern type is specified. For this purpose, first, white light (or a component obtained by splitting the light) is emitted as one of light sources having a plurality of wavelength components as probe light on the pattern.

Irradiation may be performed from the pattern surface side of the substrate having the pattern structure or from the back surface side to detect transmitted light. In the latter case, the multi-component wavelength in the infrared region is used. A light source is required. Next, in the case where the irradiation spot diameter is larger than the minimum unit forming the pattern structure and the pattern is uniformly distributed over the entire surface of the wafer (such as a pattern structure of a D-RAM). As described in [Prior Art], a stable and reproducible signal independent of a place is obtained as average information. In this case, it is not necessary to pay much attention to the measurement position (irradiation position) in both the so-called in-line measurement in which the wafer is measured after the process and the simultaneous (in-situ) measurement in the process. It is possible to calculate the film thickness and to know the end point of the process. However, a CP having an uneven pattern distribution
In the pattern structure of a device such as U or ASIC, the signal waveform greatly differs depending on the irradiation position, and the reproducibility depending on the irradiation position cannot be obtained.

As a result of a detailed examination of the pattern structure of a device such as a CPU and an ASIC, it has been found that, in many cases, a pattern is divided into a plurality of separated blocks having different optical characteristics even though the pattern is not uniform. I noticed that I can do it. In fact, we have found that in various devices, the signal reproducibility within each block is good under conditions where the probe light is incident vertically. Although the film configuration and the pattern density are different for each of these blocks, the waveforms are the same if the pattern density and the pattern definition are the same under the common conditions of the film configuration. Here, the pattern density is shown in FIG.
Corresponds to the ratio of the area of the electrode layer portion of the substrate having the pattern structure (sum of the area of the A portion) to the total area (sum of the area of the A portion + sum of the area of the B portion). Regarding the significance of pattern density, refer to Japanese Patent Application No. 11-17 / 1999.
1094. The definition of the pattern means the magnitude relationship between the minimum unit length of the pattern and the spatial coherence length, that is, the definition of the minimum coherence of the probe light having the dimension of the minimum unit of the pattern. If the minimum unit size is 2 μm,
When the spatial coherence length of the probe light is 5 μm, the definition is high, and when the spatial coherence length of the probe light is 1 μm for the same pattern size, the definition is low. Here, the dimension of the minimum unit of the pattern corresponds to the length of the portion A + the length of the portion B, taking the pattern of FIG. 5 as an example.

In the measuring apparatus shown in FIGS. 2 and 3 according to the embodiment of the present invention, the probe light is vertically incident and the spatial coherence length is adjusted. In this measuring apparatus, the spatial coherence length is made sufficiently larger than the minimum dimension of the pattern to be measured or, conversely, made sufficiently small. The significance of the adjustment of the spatial coherence length is disclosed in detail in Japanese Patent Application Laid-Open No. H11-47485. The spatial coherence length can be increased by reducing the aperture of the variable diameter stop 10 in FIG. 3, and can be reduced by increasing the aperture. This adjustment is made by comparing the measured values with the simulation calculated values.

The reflection from the substrate surface when there is a pattern structure will be described in more detail below. In the reflected light from the thin film when the substrate surface has the pattern structure, the interference phenomenon due to the film thickness (interference due to amplitude division) and the interference due to the pattern (interference due to wavefront division) are added. Since the interference by the pattern is a phenomenon between the patterns only within the range of the (spatial) coherence length on the irradiation surface of the probe light, it does not occur when the minimum unit width of the pattern is larger than the spatial coherence length. Therefore, in a coarse pattern larger than the spatial coherence length, the signal waveform is determined by simply adding the light intensities from each part of the pattern. We have devised and implemented using this to perform signal separation from low-definition patterns. (Japanese Patent Application No. 11-171094) The spot diameter of the probe light must be larger than the minimum unit of the pattern forming the pattern structure therein even if it is the minimum, and at the maximum it is possible to use one pattern type. Must be smaller than one corresponding block.
Here, that the irradiation spot diameter is larger than the minimum unit means that the spot diameter has a dimension capable of completely irradiating at least the entire minimum unit of the pattern, and that the spot diameter is smaller than the block size. Means that the spot can fit inside the block.

It is preferable that only the 0th-order reflected light be selected from the signal light and that the first-order and higher-order diffracted light components be removed. This is to eliminate the influence of the fluctuation of the contribution of the diffraction component due to the difference in the definition of the pattern.
289175. Even when the probe light is taken into consideration as described above, it is not easy to directly and analytically determine the film thickness from the signal light obtained from the reflected light from the pattern surface. Therefore, as shown in the flowchart of FIG. 1, the signal processing unit of FIG. 2 performs a comparison operation of the similarity between the signal and a reference value such as a spectral waveform calculated by simulation or a spectral waveform (learning wafer or the like) measured in advance. The fitting is done by.

The method of calculating the similarity in this fitting is not particularly limited, but preferably, any one method selected from a cross-correlation coefficient, an auto-correlation coefficient, or a least squares method can be used. . Since pattern information such as pattern density and film thickness, etc. are used as parameters in correspondence with each reference value used for this fitting operation, not only the measurement position but also the film thickness information is obtained corresponding to this measurement position. It becomes possible.

As a specific calculation example, when the film configuration is constant, the pattern type is determined only by the pattern density, and the pattern type (distribution state of the metal layer including silicide) is changed as a parameter. The calculation is performed by changing the thickness of the upper layer as a parameter. For the calculation, a refractive index, an absorption coefficient, and a film thickness are used as film constants. In the calculation, as described above, for at least one type of substrate, the definition of the pattern is increased at all the measurement positions in correspondence with the spatial coherence length of the probe light that is actually measured, that is, the pattern It is performed under the condition that the interfering completely occurs or the condition that the definition of the pattern becomes low at all the measurement positions, that is, the condition that the interpattern interference does not occur at all.

When the pattern density is changed as a parameter, the pattern density may be calculated by changing the pattern density as an unknown value from 0 to 100% evenly. However, a more efficient method is to use a block of the semiconductor element to be measured. The value of the pattern density for each pattern may be obtained in advance, the calculation may be performed only with the value of the pattern density, and the fitting may be performed only with the calculated value. In this way, the calculation amount and the calculation time can be reduced, and the calculation amount and the required time of the fitting calculation can be reduced.

The pattern density can be determined from design data such as CAD data. However, in actual patterns, there may be a difference from design values. It can also be calculated. In order to perform the fitting calculation faster and more efficiently, in each block portion of the pattern structure of the semiconductor device to be measured, a plurality of sets of simulations calculated in the assumed (top layer) film thickness range. It is a preferable method to store the spectral waveform in the memory in advance as a reference value and sequentially retrieve the data and compare it with the signal waveform. In this way,
The measurement position can be specified and the film thickness can be determined faster than performing simulation calculation every time fitting is performed. Although it is a so-called lookup table, it is preferable that the number of spectral waveforms be appropriately selected and prepared according to the required accuracy of film thickness measurement.

Also, in the case where the process end point is detected by comparing with a waveform at a process target thickness obtained in advance by a learning wafer (dummy wafer) or the like instead of a calculated waveform, actual measurement or the like is performed in advance. The signal waveform of the semiconductor element wafer having the pattern structure is acquired for each block, stored in the memory, and compared with the actually measured signal waveform, whereby each measurement position is specified quickly and appropriately, and the film It becomes possible to measure the thickness and to measure (detect) the process end point.

That is, the block at the irradiation position (measurement position) is determined and specified based on the signal waveform, and the film thickness is obtained for each measurement position. Further, when the present measuring apparatus is used in a film forming process of a semiconductor device manufacturing process or a removing process such as CMP, there is a possibility that probe light is irradiated with a target characteristic value at a process end point as a reference value in advance. It will be acquired for all irradiation positions. This makes it possible to specify the irradiation position, calculate the film thickness, and control the process, regardless of which irradiation position is used for measurement.

Since the measurement position can be specified in this manner, in the case of a measurement device having a mechanism for moving the irradiation position, a desired measurement can be performed based on the acquired signal waveform and pattern information obtained in advance from design values and the like. Since the probe light can be moved to the position, there is no need for a person to adjust the measurement position while observing the pattern (block). As described above, in the description of the measurement of the present invention, the film whose thickness is to be measured has been described as a light-transmitting dielectric film (interlayer insulating film).
The present invention is applicable even when the film to be measured is a metal layer. In general, when forming a buried electrode layer (inlay), if the metal layer formed on the entire surface of the substrate (wafer) is removed by etching or polishing, a part with and without a metal layer appears at the end of the process I do. Signal waveform of reflected light
(Waveform of spectral characteristic) is usually smooth with respect to the wavelength in the metal layer. When the metal layer disappears and the underlying pattern layer appears, the signal waveform is largely fluctuated under the influence of the pattern layer below the underlying dielectric layer, and generally the waveform is undulated. By observing this change, the end point of polishing the metal layer can be detected. In this case, by comparing the similarity between the signal waveform and the reference value in the case where the base appears, the measurement position can be specified, and further, the end point of polishing can be detected.

Further, when the polishing step includes a step of polishing two or more kinds of layers at the same time, the measurement can be preferably performed according to the present invention. Examples of this, Cu layer and SiO ₂ in FIG. 8
The cross section of the pattern structure having the barrier metal layer 37 between the layers is shown. As the barrier metal, tantalum nitride (TaN), tantalum (Ta), titanium nitride (TiN), or the like is used. Since the optical properties (optical constants) of these materials are different from Cu, the signal contribution of the Cu layer to the signal waveform is different from the signal contribution of the barrier metal layer to the signal waveform. Therefore, the signal in the stage before polishing shown in FIG. 8A, the signal in the stage where the barrier metal layer is exposed (pattern structure appears) in FIG.
Since the signals at the stage where both the barrier metal layer and the underlayer are exposed in (c) are different, the reflection characteristics at each stage are calculated by simulation or pre-measured and used as reference values to obtain the film thickness at each stage. Alternatively, the measurement (detection) of the process end point can be appropriately performed. Further, since the Cu layer and the barrier metal layer can transmit the probe light when the layer thickness is thin, FIGS.
Not only in the stage (c), but also in the intermediate stage, C
The thickness of the u layer and the barrier metal layer can be measured.
Thus, also in this case, the film thickness and the process end point can be easily measured (detected) with high accuracy.

Further, when the polishing step includes a step of simultaneously polishing two or more kinds of layers transparent to the probe light, the measurement can be preferably performed according to the present invention. As an example of this, FIG. 9 shows STI (Shallow Trench Isola).
The steps for the purpose of (Option) are shown. In this step, as conceptually shown in FIG. 9, the element portion composed of the Six N layer 39 and the Si 40 is electrically separated by arranging the SiO ₂ insulating layer 28 two-dimensionally. . FIG.
In this case, both the SiO ₂ insulating layer 28 and the Six N layer 39 are transparent to the probe light,
It is transparent because it has _two layers. FIG. 9A shows a state before polishing, and FIG. 9B shows an ideal state after polishing is finished.
Shows a state in which is formed. SiO ₂ insulating layer 28 and Six
The polishing rate is different from that of the N layer 39, and the SiO ₂ insulating layer is generally faster than the Six N layer. Although this degree depends on the polishing conditions, in the example using a certain polishing agent, the ratio of the Six N layer to the SiO ₂ insulating layer is one. Therefore, in polishing, the film thickness always decreases at this rate.
As shown in (c), a structure state in which the trench portion is depressed, that is, dishing occurs. The reason for this is that, due to the difference in the polishing rate, the SiO ₂ insulating layer has the thickness indicated by 43 while the Six N layer is removed by the thickness indicated by 42. The ratio of each removed film thickness is constant without depending on the size of the removed film thickness. Accordingly, the remaining film thickness T ₁ of the Six N layer 39
It can be seen that the remaining film thickness T ₂ of the SiO ₂ insulating layer 42 is related to the remaining film thickness T ₂ by one fixed relationship. That T ₁ is determined is T ₂ if Kimare, T ₂ is T ₁ is determined if Kimare. This correspondence is obtained in advance by experiments and measurements, and the thicknesses T ₁ and T ₂ of the obtained correspondence are used in the simulation calculation as a reference value. Of course, the reference value may be based on prior measurement. In this way, the dishing amount is predicted and used for measuring the film thickness or setting the process end point, thereby preventing a measurement error from occurring when a model as shown in FIG. Accuracy can be improved.

In the above description of the measuring apparatus, the signal light is extracted from the reflected light of the probe light, but it goes without saying that the signal light may be extracted from the transmitted light. Also, for the sake of simplicity of description, the case where the signal to be obtained is a spectral characteristic signal has been described, but the signal to be obtained is not limited to the spectral characteristic signal, and its Fourier transform value signal is also included in the scope of the present invention. Needless to say. In this case, a spectral characteristic value calculated by simulation or a Fourier transform value of a spectral characteristic value measured in advance is used as a reference value. When used as a signal for obtaining a Fourier transform value signal, usually, the measured spectral characteristic signal is electrically Fourier transformed in the signal processing device 6. Improvement in measurement accuracy can be expected by performing fitting on a Fourier transform signal obtained by removing, for example, inconvenient frequency components from the Fourier transform signal.

The measuring apparatus according to the present invention can be applied to a polishing apparatus for simultaneous (in-situ) measurement in which measurement is performed while performing a flattening step. FIG. 4 shows a polishing apparatus according to the present invention.
Reference numeral 21 denotes a polishing head for holding the wafer 1 having a pattern structure, reference numeral 20 denotes a polishing pad, reference numeral 19 denotes a surface plate on which the polishing pad is fixed, reference numeral 22 denotes an abrasive supply mechanism, reference numeral 23 denotes an abrasive, reference numeral 2 denotes an optical measuring unit, and reference numeral 6 denotes a signal. A processing unit 8 is a display unit. Wafer 1
Of the wafer 1 is pressed against the polishing pad 20, and a relative motion is given to the surface to be polished of the wafer 1 and the polishing pad 20 by respective driving mechanisms (only the example of the driving direction is indicated by an arrow). Polishing is performed while the abrasive 23 is supplied by the abrasive supply mechanism 22 between the processing surfaces of the pad 20. At this time, it is a preferable method to make a part of the polishing pad and the surface plate light-transmitting, and exchange probe light and reflected light therefrom. Since the semiconductor element wafer is constantly moving during polishing, the measurement position to be irradiated with the probe light is not generally determined, and signals from various measurement positions are continuously obtained. In such a case, by using the measuring apparatus of the present invention, by comparing the obtained signal with a reference value, the irradiation position is specified, the film thickness is obtained for each of the specified measurement positions, and the data is processed. This enables process control in the film removal process. Further, the present polishing apparatus specifies a measurement position and moves the probe light to a desired measurement position, thereby obtaining a polishing property (polishing speed, step-elimination property, etc.) depending on the pattern (measurement position) of the semiconductor element wafer. However, even when there is a difference in the pattern dependency, it is also possible to always select and process only data from the same desired measurement position (block). Also, data from various measurement positions (blocks) can be sorted for each measurement position (block) and processed individually. In this way, it is possible to control the film thickness in various patterns.

The above-described polishing apparatus performs measurement using probe light and reflected light that pass through the light transmitting window 18 provided on the surface plate and the polishing pad. Needless to say, the measurement can also be performed by irradiating the protruding wafer portion with the probe light. As described above, the example in which the measuring apparatus and the method of the present invention are applied to the polishing apparatus and the method for use in the flattening step has been described. However, the measuring apparatus and the method of the present invention may be applied to other manufacturing processes of a semiconductor device, for example, a film forming step or etching. Needless to say, it can be preferably applied to the process.

[0054]

[Example 1] Actual 6 inch wafer (substrate)
The SiO ₂ interlayer insulating film of the above semiconductor element was polished by a CMP apparatus, and an end point of the polishing was detected. The schematic explanatory diagram is shown in FIG. The polished element is shown in FIG.
Has a block structure of A to F as shown in FIG. 1, and each block has a pattern type having a different pattern density from each other. The probe light irradiation is performed by using a CMP apparatus shown in FIG. 4 and polishing the polishing pad 20 (epoxy) on the lower surface and the platen 19 (both with a diameter of 600 mm) of about 1 cm × 4 c.
A rectangular hole of m was formed, and the observation was performed from an observation window provided with a quartz light transmitting window 18 on the same surface as the pad surface. One or both of a hard coating and an anti-reflection film for reducing reflection loss of probe light are formed on both surfaces of the light-transmitting window to prevent scratches due to wafer contact or the like.

As shown in FIG. 3, the optical measuring section adjusts the light using the xenon lamp 9 as a light source through the variable-diameter diaphragm 10 so that the spatial coherence length completely interferes with the pattern. After adjustment, the probe light having a spot diameter of 0.5 mm is perpendicularly incident on the surface of the wafer 1, and the reflected light passes through a pinhole (scattered light, first-order or more diffracted light is removed), and the wavelength is resolved by the diffraction grating 16. Then, the light was detected by the photodiode-type linear sensor (512 elements) 17 so that light of different wavelengths was directed in different directions. The measurement wavelength range is about 400 nm to 800 nm. Further, while comparing the measurement signal waveform with the reference value,
By adjusting 0, the probe light completely caused pattern interference. The signal output from the linear sensor is processed by a personal computer having a signal processing unit. Prior to signal processing, xenon lamp deterioration, linear sensor sensitivity reduction,
In order to normalize the fluctuation of the magnitude of the signal intensity due to the above-mentioned factors, the previously measured spectral intensity information of the light source light is used as a coefficient at the time of processing.

The polishing was performed at a polishing pressure of about 100 g / cm ² using an abrasive (slurry) in which silica particles were dispersed in an alkaline solvent. The effect on the light quantity (mainly scattering loss) due to the presence of the slurry was 1% or less. The SiO ₂ interlayer insulating film on the semiconductor element surface of the wafer is formed to a thickness of about 1000 nm by plasma CVD. This SiO
_{2 The} interlayer insulating layer is polished and removed.
Polishing to finish polishing with a remaining film thickness of nm is performed.
Here, the pattern density of each of the blocks A to F shown in FIG. 7 has been obtained in advance by calculation from an image observed with a microscope. Using this pattern density and information such as the thickness of the lower layer, the refractive index, and the absorption coefficient, the thickness of the SiO ₂ interlayer insulating layer is set to 5 nm in the range from 1020 nm to 480 nm.
The simulation was performed in advance to calculate the spectral reflectance for each pitch. The calculated values are 109 per block, 654 in total, and these are stored in the internal memory as a data table. These 654 pieces of data correspond to k sets of reference values in step S8 in FIG.

While polishing both the wafer and the polishing pad at a rotation speed of 60 rpm, a signal (spectral reflectance data) was obtained from the light transmitting window 18. In this example, the signal acquisition time is about 2 msec for signal acquisition of one point, and measurement of eight points is performed while the surface plate 19 makes one rotation. For the fitting, the data for the eight signals was compared with the data for each block stored in advance. The cross-correlation coefficient between the signal waveform and each reference value is calculated for comparison, the reference value when the value of the cross-correlation coefficient is the largest is determined, and the pattern type and the film thickness at the measurement position are determined. FIG. 7 shows the state of the determination. If the cross-correlation coefficient is less than a predetermined value, it is considered as a signal from an unsuitable portion for measurement such as a wafer edge, and is excluded from the analysis data. The above-described calculation of the cross-correlation coefficient and determination of the pattern type and the film thickness of the measurement position for each of the eight data points are performed by the following signal acquisition (generally, after one rotation of the platen, the light transmission window of the polishing pad). Is again under the wafer).

In this way, during polishing, the pattern type and the film thickness of the reference value showing the best cross-correlation coefficient with the signal waveform are successively determined, and these pattern types and the film thicknesses are allocated to the corresponding blocks. When the change in the film thickness of each block was measured simultaneously (in-situ), the target remaining film thickness reached 500 nm in terms of block A, and polishing was completed. The film thickness of the block A of this wafer is actually
When measured by M observation, it was within an error of about 3% of the target remaining film thickness, and it was confirmed that the process end point can be detected by this example. [Embodiment 2] In this embodiment, instead of the simulation calculation of the spectral reflectance performed in Embodiment 1, the final target film thickness (5
(Nm) of each of the blocks A to F shown in FIG.
Were measured and stored as a data table.
Then, polishing for terminating the polishing with the remaining film thickness of about 500 nm in the block F (blank portion) is performed. All other experimental conditions were the same as in Example 1.

As the polishing progresses and the end point of the process is approached by time management, signal acquisition is started and compared with a waveform corresponding to a final target remaining film thickness (500 nm) stored in advance. The calculation of cross-correlation coefficient was started. Polishing was terminated when the cross-correlation coefficient at a residual film thickness of 500 nm in terms of the pattern F exceeded a reference value, and the film thickness of the block F of this wafer was actually measured by ellipsometry. This is within an error range of about 3%, and it was confirmed that the process end point can be detected by this example. According to the present embodiment, the labor of the simulation calculation can be saved and the amount of the fitting calculation can be reduced, so that the pattern type can be specified more quickly.

As described above, according to the first and second embodiments, the measurement position and the pattern type at the measurement position can be specified simply, with high accuracy, at high speed, and simultaneously (in-situ), and the film thickness is measured. It was confirmed that detection (measurement) was possible.

[0061]

As described above, according to the present invention, in the measurement of a semiconductor device wafer, even if the semiconductor device wafer has a pattern structure or a plurality of pattern types, the wafer in the STI step However, even in the case of a process in which there is a barrier metal, the adjustment of probe light, the appropriate optical system arrangement, and the accurate simulation calculation technology enable simple, high-precision, high-speed, as needed and simultaneously (in-sit
u), the pattern type at the measurement position can be specified, the film thickness can be measured, the end point of the process can be detected (measured), and the process can be controlled. Further, it is possible to provide a polishing apparatus and a polishing method capable of performing the above-mentioned measurement, detecting an end point with high accuracy and easily.

[Brief description of the drawings]

FIG. 1 is a flowchart showing an operation of the measuring device of the present embodiment.

FIG. 2 is a schematic diagram showing an outline of the measuring device of the present embodiment.

FIG. 3 is a detailed view of an example of an optical measuring section 2 of the measuring apparatus of FIG.

FIG. 4 is a diagram illustrating a polishing apparatus of the present invention.

FIG. 5 is a diagram illustrating the concept of pattern density.

FIG. 6 is a diagram for explaining Examples 1 and 2;

FIG. 7A is a diagram of a chip of a semiconductor element of Examples 1 and 2, and this chip includes A, B, C, D, E, and F;
Each pattern type is divided by each block.
(B) shows the distribution of signals obtained in one signal acquisition time in the first and second embodiments.

FIG. 8 is a diagram illustrating polishing when a barrier metal layer is present.

FIG. 9 is a diagram illustrating an STI process.

[Explanation of symbols]

DESCRIPTION OF SYMBOLS 1 Substrate (wafer) 2 Optical measurement part 3 Probe light 4 Reflected light 5 Signal 6 Signal processing part 7 Measurement value signal 8 Display part 9 Light source (xenon lamp) 10 Spatial coherence length control part (variable diameter stop) 11 Optical lens REFERENCE SIGNS LIST 12 beam splitter 13 optical lens 14 mirror 15 optical lens 16 spectral part (diffraction grating) 17 light detecting part 18 light-transmitting window 19 surface plate 20 polishing pad 21 polishing head 22 abrasive supply mechanism 23 abrasive 24 wafer 25 thermal oxide film 26 Electrode layer 27 Antireflection film 28 Dielectric layer 29 Pattern surface 30 Measurement position (probe light spot) 31 Polishing pad rotation direction 32 Polishing pad swinging direction 33 Chip 34 Probe light diameter adjustment unit (field stop) 35 0th-order light extraction unit (Aperture) 36 Cu layer 37 Barrier metal layer 38 SiO ₂ layer 39 Six N layer 4 0 Si 41 Trench 42 Removed film thickness of Six N layer layer 43 Removed film thickness of SiO ₂ insulating layer 130 Optical lens A Pattern portion B Non-pattern portion T ₁ Six N Remaining film thickness of N layer T ₂ SiO ₂ Remaining insulating layer 42 Film thickness

Claims (22)

[Claims] A step of irradiating a substrate surface having a pattern structure on its surface with probe light having a plurality of wavelength components; extracting signal light from light reflected or transmitted from the substrate surface; Obtaining a signal waveform, and identifying a measurement position (probe light irradiation position) in the substrate by comparing the signal waveform with a calculated or previously measured reference value. Measurement method. 2. The method according to claim 1, wherein the calculation of the reference value is performed using a pattern density as a parameter. 3. The method according to claim 1, wherein the extracted signal light includes only a regular reflection component (0-order light) of the reflected or transmitted light of the probe light and does not include a first-order or higher-order diffracted light component. Item 3. The measurement method according to any one of Items 1 to 2. 4. The method according to claim 1, further comprising the step of controlling the spatial coherence length of the probe light. 5. A spot diameter of the probe light on the irradiation surface is larger than a minimum unit size of a pattern constituting the pattern structure and smaller than a block size. The measurement method according to claim 1. 6. The measuring method according to claim 4, wherein the spatial coherence length is adjusted to a condition for completely pattern interference or a condition for no pattern interference at all. 7. The measuring method according to claim 1, wherein said probe light is irradiated on said substrate surface perpendicularly. 8. The method according to claim 1, wherein the comparison is performed using at least one selected from a cross-correlation coefficient, an auto-correlation coefficient, and a least squares method. Measurement method. 9. The method according to claim 1, further comprising the step of storing the calculated or pre-measured reference value as a data table in advance. 10. The measuring method according to claim 1, further comprising the step of moving a measuring point to a desired measuring position based on the specified result. 11. The method according to claim 1, further comprising the step of measuring a film thickness of the film on the substrate surface at each of the measurement positions shifted to the specified or the measurement point. Measurement method described in the section. 12. The method according to claim 11, further comprising a step of measuring a change in film thickness for each of said measurement positions when said film is placed in a film forming step or a removing step. 12. The measuring method according to 11. 13. The method according to claim 13, wherein the substrate is a wafer of a semiconductor element having a pattern structure on a surface, and the measurement is performed in a film forming step or a removing step of an insulating layer or an electrode layer on the wafer in a semiconductor device manufacturing process. 13. The method according to claim 11, further comprising the step of measuring (detecting) a process end point. 14. The measuring method according to claim 13, wherein said removing step comprises a step of simultaneously polishing at least two or more types of layers transparent to said probe light. 15. The method according to claim 15, wherein the removing step is STI (Shallow Tren).
The method according to claim 14, wherein the step is a step for the purpose of (Ch Isolation). 16. The measuring method according to claim 14, wherein the calculation of the reference value is performed for a film thickness whose dishing amount has been corrected in advance. 17. The method according to claim 13, wherein said removing step comprises a step of simultaneously polishing at least two or more layers opaque or translucent to said probe light.
The measurement method described. 18. The method according to claim 17, wherein said two or more types of layers include a Cu layer and a barrier metal layer. 19. An optical measurement section for irradiating a substrate surface with probe light having a plurality of wavelength components and acquiring a signal waveform from signal light extracted from light reflected or transmitted from the substrate surface; 19. A signal processing unit having at least a function of specifying a measurement position (probe light irradiation position) in the substrate by performing comparison with a waveform calculated or measured in advance, and a signal processing unit, and any one of claims 1 to 18. A measuring apparatus characterized in that the measurement is performed by the measuring method described in the above. 20. The method according to claim 20, wherein the substrate is a wafer of a semiconductor element having a pattern structure on a surface, and the measurement is performed in a step of forming or removing an insulating layer or an electrode layer on the wafer in a semiconductor device manufacturing process. 20. The method according to claim 19, wherein the measurement is a measurement and a measurement (detection) of a process end point is performed.
The measuring device as described. 21. A method of polishing a semiconductor device wafer having a pattern structure on the surface and having an insulating layer or an electrode layer on the outermost layer, wherein an abrasive is interposed between the wafer and the polishing pad. And polishing the insulating layer or the electrode layer of the wafer by moving the polishing pad and the wafer relative to each other in a state where the polishing pad is in an inclined state.
8. A step of measuring (detecting) a film thickness or a process end point by the measuring method according to any one of 8. 22. A polishing apparatus comprising: the measuring device according to claim 20; a polishing pad; and a polishing head for holding a wafer of a semiconductor element having an insulating layer or an electrode layer on the outermost surface layer. With an abrasive between them,
By relatively moving the polishing pad and the wafer, the insulating layer or the electrode layer of the wafer is polished, and the measuring device measures the insulating layer or the electrode layer or the end point of the polishing process. A polishing apparatus for performing measurement (detection).