Semiconductor wafer evaluation method

Abstract

To provide a method for evaluating semiconductor wafers, with which it is possible to objectively evaluate a type of an undulation shape of semiconductor wafers.SOLUTION: A semiconductor wafer evaluation method according to the present invention includes: a first step in which a shape of a semiconductor wafer in a thickness direction is measured and shape data is obtained; a second step in which the process of extracting a shape profile in a radial direction of the semiconductor wafer that represents a change of the shape in the thickness direction of the semiconductor wafer on the basis of the shape data is repeated for each of given angles, and a plurality of shape profiles is obtained; a third step in which a shape feature parameter of each of the shape profiles is established on the basis of each of the shape profiles; and a fourth step in which a shape of the semiconductor safer is evaluated on the basis of a change of the shape feature parameter due to the angle.SELECTED DRAWING: Figure 2

Description

Technical Field

[0001] The present invention relates to a method for evaluating a semiconductor wafer.

Background Art

[0002] The semiconductor wafer may have warps. Conventionally, when such warps occur, countermeasures such as adjusting the polishing conditions according to the warp shape have been taken.

[0003] Conventionally, as a method for evaluating the magnitude of warps of a semiconductor wafer, for example, using a Warp value through a band-pass filter (for example, Patent Document 1) has been proposed. Here, the shapes of the warps generated on the surface of the semiconductor wafer can be classified into several types, and when considering countermeasures when warps occur, it is necessary to consider not only the magnitude of the warps but also the types of warps.

Prior Art Documents

Patent Documents

[0004]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0005] Conventionally, regarding the types of warps, a human has observed and judged the shape map, but with such a method, whether appropriate countermeasures can be taken against the occurrence of warps depends on the proficiency of the operator.

[0006] Therefore, an object of the present invention is to provide a method for evaluating a semiconductor wafer that can objectively evaluate the types of warp shapes of the semiconductor wafer.

Means for Solving the Problems

[0007] The gist configuration of the present invention is as follows. (1) A method for evaluating a semiconductor wafer, comprising: a first step of measuring the shape in the thickness direction of the semiconductor wafer to obtain shape data; a second step of repeatedly extracting, at regular angles, a shape profile representing a change in the shape in the thickness direction of the semiconductor wafer in the radial direction based on the shape data to obtain a plurality of shape profiles; a third step of obtaining shape feature parameters of each of the shape profiles based on each of the shape profiles; and a fourth step of evaluating the shape of the semiconductor wafer based on a change in the shape feature parameters with respect to the angle, the method for evaluating a semiconductor wafer being characterized by including these steps.

[0008] (2) The method for evaluating a semiconductor wafer according to (1) above, wherein the shape feature parameter is the slope of a straight line obtained by linearly approximating the shape profile.

[0009] (3) The method for evaluating a semiconductor wafer according to (1) or (2) above, wherein in the fourth step, the shape of the semiconductor wafer is evaluated based on a relationship in which as the rotational symmetry of the shape in the thickness direction of the semiconductor wafer increases, the variation of the shape feature parameter with respect to the angle decreases. Here, the "rotational symmetry" means a property in which when the center of the semiconductor wafer is the center and the circumferential direction of the semiconductor wafer is the rotation direction, the shape in the thickness direction of the semiconductor wafer does not change depending on the rotation angle. The higher the rotational symmetry, the smaller the change in the shape in the thickness direction of the semiconductor wafer depending on the rotation angle.

[0010] (4) The method for evaluating a semiconductor wafer according to (3) above, wherein as an index of the variation of the shape feature parameter with respect to the angle, the difference between the maximum value and the minimum value in the entire angle range of the shape feature parameter is used.

Advantages of the Invention

[0011] According to the present invention, it is possible to provide a semiconductor wafer evaluation method that can objectively evaluate the type of waviness shape of a semiconductor wafer. [Brief explanation of the drawings]

[0012]

Figure 1

Figure 2

Figure 3A

Figure 3B

Figure 3C

Figure 3D

Figure 3E

Figure 3F

Figure 3G

Figure 3H

Figure 4

Figure 5

[0013] DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Hereinafter, embodiments of the present invention will be described in detail with reference to the accompanying drawings.

[0014] 1 is a flowchart showing a semiconductor wafer evaluation method according to one embodiment of the present invention. As shown in FIG. 1, in this embodiment, first, the shape of the semiconductor wafer in the thickness direction is measured to obtain shape data (first process: step S101).

[0015] Here, the shape data in the thickness direction of a semiconductor wafer is data representing the shape of the semiconductor wafer, obtained, for example, by measuring the thickness of the semiconductor wafer at measurement points uniformly distributed over the entire surface of the semiconductor wafer. More specifically, the shape data can be obtained by measuring the thickness at each point set at a predetermined interval in the radial direction and a predetermined interval in the circumferential direction of the semiconductor wafer. As an example, the thickness can be measured at each point set at equal intervals of 1 mm in the radial direction of the semiconductor wafer and equal intervals of 1° in the circumferential direction of the semiconductor wafer. For example, in the case of a wafer with a diameter of 300 mm, 149 points at 1 mm intervals from 1 mm to 149 mm (excluding the wafer center and outer periphery) are multiplied by 360 points at 1° intervals from 0° to 359°, and then one point for the thickness at the center of the wafer is added to obtain shape data consisting of 149 × 360 + 1 = 53,641 points of semiconductor wafer thickness data. The order of measurement at each point is not particularly limited.

[0016] The thickness of the semiconductor wafer can be measured using any known measuring device, such as an infrared measuring device or a spectroscopic interference measuring device.

[0017] Next, in this embodiment, based on the shape data, a shape profile representing the change in the shape of the semiconductor wafer in the thickness direction of the semiconductor wafer in the radial direction of the semiconductor wafer is extracted repeatedly at a certain angle to obtain multiple shape profiles (second process: step S102).

[0018] Fig. 2 is a diagram showing a shape profile of a semiconductor wafer at one angle. Fig. 2 shows an example of a semiconductor wafer with a diameter of 300 mm, where the shape in the thickness direction was measured at equal intervals of 1 mm in the radial direction of the semiconductor wafer. Such shape profiles are repeatedly obtained at regular angles (e.g., at equal intervals of 1°) to obtain multiple shape profiles (e.g., 360).

[0019] Next, in the present embodiment, shape feature parameters of each shape profile are obtained based on each shape profile (third step: step S103). The "shape feature parameters" referred to here are parameters that enable evaluation of the shape of the semiconductor wafer based on changes in the above angles of the shape feature parameters, as will be described in detail in the fourth step described later. In the example shown in FIG. 2, the shape feature parameter is the slope of a straight line obtained by linearly approximating the shape profile. More specifically, such a slope can be, for example, the slope when the shape profile (entirety) is linearly approximated by the least squares method (FIG. 3A), or the slope of a straight line connecting the thickness value at the center of the semiconductor wafer and the thickness value at the outer peripheral edge of the semiconductor wafer (FIG. 3B). Alternatively, it can be the slope intermediate between the maximum slope and the minimum slope (FIG. 3C).

[0020] The shape feature parameters are not limited to the slopes as described above. For example, when the shape profile is approximated by an arc (for example, when approximated within a range where the coefficient of determination R 2 is less than or equal to a predetermined value), the radius of curvature (FIG. 3D), the height of the arc (the length of the foot of the perpendicular dropped from the shape profile to the straight line connecting the thickness value at the center of the semiconductor wafer and the thickness value at the outer peripheral edge of the semiconductor wafer, at the location where the length of the foot is maximum) (FIG. 3E), the area enclosed by the straight line connecting the thickness value at the center of the semiconductor wafer and the thickness value at the outer peripheral edge of the semiconductor wafer and the shape profile (FIG. 3F), the perimeter length of the shape profile (FIG. 3G), the radial distance (from the center) to the position of the height of the arc (the height at the location where the height is maximum) (FIG. 3H), etc. can all be candidates.

[0021] Next, in the present embodiment, the shape of the semiconductor wafer is evaluated based on changes in the shape feature parameters with respect to the angle (fourth step: step S104). More specifically, in this fourth step, the shape of the semiconductor wafer is evaluated based on the relationship that as the rotational symmetry of the shape in the thickness direction of the semiconductor wafer increases, the variation in the shape feature parameters with respect to the above angle decreases.

[0022] As an example, as an index of the variation due to the angle of the shape feature parameter, the difference between the maximum value and the minimum value in the entire angular range of the shape feature parameter (the inclination in this embodiment) (hereinafter referred to as the range) can be used. As illustrated at four angles in FIG. 4, the range can be obtained from the inclination at each angle (in the case of FIG. 4, since the maximum value is 0.42 at an angle of 0° and the minimum value is -0.33 at an angle of 270°, the range = 0.42 - (-0.33) = 0.75). At this time, it is preferable to perform moving averaging at each angle in order to remove noise. The variation index can be various, such as the difference between 90% of the maximum value and 110% of the minimum value, the standard deviation, or the number of changes between the upwardly convex shape and the downwardly convex shape in the entire angular range.

[0023] Accordingly, the larger the above range, the lower the rotational symmetry of the shape in the thickness direction of the semiconductor wafer. Thus, the types of warps of the semiconductor wafer can be classified and the shape of the semiconductor wafer can be evaluated. Such an evaluation can be performed by converting the value of the above range into a continuous index and evaluating it as a continuous value, or by providing a plurality of threshold values and performing multi-stage classification, or by providing one threshold value and performing a determination such as pass / fail. The second to fourth steps can be executed, for example, by a processor of a computer.

[0024] According to the method for evaluating a semiconductor wafer of the present embodiment, what is performed in the first step is measurement, the plurality of shape profiles obtained in the second step can be uniquely obtained from the shape data obtained in the first step, the shape feature parameters (the inclination in this embodiment) at each angle obtained in the third step can also be uniquely obtained from the plurality of shape profiles obtained in the second step, and the evaluation performed in the fourth step can also be uniquely calculated using a predetermined index (the range in this embodiment, and the range can be uniquely obtained). Therefore, according to the semiconductor wafer evaluation method of this embodiment, the type of waviness shape of the semiconductor wafer can be objectively evaluated without being influenced by subjective factors such as the skill level of the operator.

[0025] A semiconductor wafer polishing apparatus includes a carrier plate with a holding hole formed therein and a surface plate with a polishing pad attached thereto. The semiconductor wafer is polished by pressing the polishing pad against the semiconductor wafer while housed in the holding hole and rotating the carrier plate and surface plate relative to each other. As the carrier plate and surface plate rotate relative to each other, the semiconductor wafer housed in the holding hole also rotates within the holding hole. This is called the rotation of the semiconductor wafer. When the semiconductor wafer rotates smoothly within the holding hole, i.e., when the semiconductor wafer has good rotational properties, the entire semiconductor wafer is polished more uniformly, making it less likely to produce waviness. On the other hand, when the semiconductor wafer does not rotate smoothly within the holding hole, i.e., when the semiconductor wafer has poor rotational properties, uneven polishing of the semiconductor wafer occurs, resulting in waviness. Regarding the rotational ability of semiconductor wafers during polishing, for example, if the polishing pad lacks surface roughness and the polishing pad's ability to hold the semiconductor wafer is weak, it is believed that sufficient rotation is not possible. Furthermore, if the shape of the polishing pad itself hinders the rotation of the semiconductor wafer, the wafer becomes even more difficult to rotate. For this reason, if the above evaluation results show that the range is equal to or greater than a first predetermined value and equal to or less than a second predetermined value (the second predetermined value is greater than the first predetermined value), the polishing pad is deemed to lack surface roughness, and measures such as initializing the surface of the polishing pad by long-term dressing can be taken. Furthermore, if the range exceeds the second predetermined value, measures such as initializing the surface and shape of the polishing pad by dressing it to form a shape that does not hinder rotation can be taken.

[0026] Here, the shape characteristic parameter is preferably the slope of a straight line obtained by first-order approximation of the shape profile, because the slope can be easily calculated and is suitable for evaluating the type of waviness.

[0027] In the fourth step, it is preferable to evaluate the shape of the semiconductor wafer based on the relationship that the higher the rotational symmetry of the shape in the thickness direction of the semiconductor wafer, the smaller the variation in the shape characteristic parameters due to the angle, because by using such a relationship, the type of waviness of the semiconductor wafer can be appropriately evaluated.

[0028] Furthermore, it is preferable to use the difference between the maximum and minimum values of the shape characteristic parameter over the entire angle range as an index of the variation of the shape characteristic parameter depending on the angle, because this can be easily calculated and is a highly accurate index.

[0029] Although the present embodiment has been described above, the present embodiment is not limited to the above embodiment. For example, the shape of a semiconductor wafer may be evaluated using machine learning as follows. First, an artificial intelligence model is created in advance by machine learning using a machine learning unit (first processor) of a computer, with the shape characteristic parameters of the shape profile at each angle as explanatory variables (input) and the type of waviness of the semiconductor wafer as the objective variable (output) (sufficient learning data necessary for machine learning is prepared in advance). Then, in a fourth step, when the shape characteristic parameters at each angle obtained in the third step are input into the created artificial intelligence model, the machine learning unit outputs the type of waviness of the semiconductor wafer and evaluates the shape of the semiconductor wafer. Note that any known machine learning algorithm, such as a neural network, can be used.

[0030] In the above case, the shape feature parameters may be determined in advance (for example, those shown in FIGS. 3A to 3H), or the computer may have a feature extraction unit (second processor), and for example, shape feature parameters such as those shown in FIGS. 3A to 3H may be extracted as feature amounts. That is, the feature extraction unit may extract the feature amount from the shape profile, and in the third step, the shape feature parameter, which is the extracted feature amount, may be obtained. Known techniques such as deep learning can be used for the extraction of the feature amount.

[0031] Hereinafter, embodiments of the present invention will be described, but the present invention is not limited to the following embodiments at all.

Embodiment

[0032] First, by a known method, the types of the undulations of the wafer were divided into three levels based on the rotational symmetry of the shape profile, and a total of nine semiconductor wafers with a diameter of 300 mm, three for each level, were prepared. Then, the above first step to the fourth step were performed on these semiconductor wafers. In the first step, measurement points were set at equal intervals of 1 mm in the radial direction and at equal intervals of 1° in the circumferential direction of the semiconductor wafer, and using a flatness measuring instrument, the thickness data of the semiconductor wafer was measured to obtain shape data. As the shape feature parameter, as shown in FIG. 3A, the slope when the shape profile (entire) was linearly approximated by the least squares method was used. As an index of the variation of the shape feature parameter with respect to the angle, the difference (range) between the maximum value and the minimum value in the entire angular range of the shape feature parameter was used. FIG. 5 is a diagram showing the evaluation results.

[0033] As shown in FIG. 5, all three semiconductor wafers for which the range was calculated to be less than 0.3 by this method were semiconductor wafers that were considered to have high rotational symmetry by known methods. Also, all three semiconductor wafers for which the range was calculated to be 0.3 or more and 0.6 or less by this method were semiconductor wafers that were considered to have medium rotational symmetry by known methods. Further, all three semiconductor wafers for which the range was calculated to be more than 0.6 by this method were semiconductor wafers that were considered to have low (non-uniform) rotational symmetry by known methods. Thus, the evaluation results in this example are able to obtain the same results as known methods, and as already explained, no subjective elements enter into this evaluation.

Claims

1. A method for evaluating a semiconductor wafer, comprising: a first step of measuring the shape in the thickness direction of the semiconductor wafer to obtain shape data; a second step of repeatedly extracting, at regular intervals, a shape profile representing a change in the shape in the thickness direction of the semiconductor wafer in the radial direction based on the shape data to obtain a plurality of shape profiles; a third step of obtaining shape feature parameters of each of the shape profiles based on each of the shape profiles; a fourth step of evaluating the shape of the semiconductor wafer based on a change in the shape feature parameters with respect to the angle, wherein the shape feature parameter is the slope of a straight line obtained by linearly approximating the shape profile, and in the fourth step, the shape of the semiconductor wafer is evaluated based on a relationship in which the variation of the shape feature parameter with respect to the angle decreases as the rotational symmetry of the shape in the thickness direction of the semiconductor wafer increases. A method for evaluating a semiconductor wafer, characterized by this.

2. The method for evaluating a semiconductor wafer according to claim 1, wherein a difference between a maximum value and a minimum value in the entire angular range of the shape feature parameter is used as an index of the variation of the shape feature parameter with respect to the angle.

Images