Double-sided polishing process

Abstract

Method for double-sided polishing, comprising a polishing cycle that includes the following: Performing a first polishing step in which a double-sided polishing process is carried out at a high polishing rate, wherein the double-sided polishing process simultaneously polishes both surfaces of a wafer which is held by a carrier and arranged between polishing pads, each arranged on an upper and a lower rotary disk, while the carrier is rotated and revolved, polishing medium is supplied and a thickness of the wafer is measured. Performing a second polishing step in which the double-sided polishing process is carried out at a low polishing rate, Subdividing a straight line extending from one outermost circumference of the polished wafer through a center of the polished wafer to the other outermost circumference into predetermined sections, and optically measuring the cross-sectional shape of the subdivided sections; Applying a predetermined weight for each subdivided section to the measured cross-sectional shape to quantify the flatness of each section; and Determining the polishing conditions of the first and second polishing steps in a subsequent polishing cycle based on quantified flatness, wherein A beam diameter of a measuring device used to measure the cross-sectional shape of outermost sections is smaller than the one used to measure the cross-sectional shape of the other section.

Description

TECHNICAL AREA

[0001] The present invention relates to a double-sided polishing method for simultaneously polishing both surfaces of a wafer while a polishing agent is supplied. STATE OF THE ART

[0002] The miniaturization of semiconductor devices has led to a greater need for highly flat semiconductor wafers with improved productivity, which serve as their substrate. Given these circumstances, double-sided polishing has been adopted for wafer polishing due to its higher accuracy compared to the conventional single-sided polishing.

[0003] Fig.Figure 8 shows a schematic diagram of a typical double-sided polishing device of the planetary gear type. The double-sided polishing device 101 comprises an upper and a lower turntable. The upper turntable is movable up and down and can apply a load to a wafer located between the upper and lower turntables by pressing on the lower turntable. As shown in Fig. As shown in Figure 8, the device 101 for double-sided polishing comprises a sun wheel 107 arranged inside the lower rotary disk and an inner wheel 108 arranged outside the lower rotary disk.

[0004] A carrier 105 for holding the wafer is arranged between the upper and lower turntables and can rotate, its outer circumference engaging the sun gear 107 and the inner gear 108. The carrier 105 is rotated and revolved between the upper and lower turntables according to the rotational speed of the sun gear 107 and the inner gear 108. The wafer to be polished is inserted into a holding opening 106 formed in the carrier and held there, so that the wafer can be polished without being detached from the double-sided polishing device.

[0005] Regarding the device for double-sided polishing of the planetary gear type, it is known that, since the ratio between the thickness of a support and the thickness of a polished wafer, i.e., the thickness of a finished wafer, affects the flatness of the polished wafer, the flatness is controlled by adjusting the ratio of the final thickness to the thickness of the support (see, for example, patent document 1).

[0006] Generally, improving flatness precision requires a reduced polishing rate, whereas improving productivity requires a higher polishing rate. The polishing process is therefore divided into two steps: a first polishing step and a second polishing step. The first polishing step performs coarse polishing at a high polishing rate; the second polishing step performs precision polishing at a low polishing rate. In other words, the first polishing step contributes to efficient polishing, while the second polishing step contributes to achieving precise flatness, thus improving flatness without reducing productivity.

[0007] The final thickness is set by changing the polishing time in the first polishing step. The conventional method described above must finish a wafer with the optimal thickness depending on the support thickness in order to polish the wafer to a flat surface. CN 1 02 089 121 A discloses a method for polishing wafers, wherein the two surfaces of a wafer are polished simultaneously by arranging the wafer in a sandwich-like manner and sliding it under pressure by means of at least one lower plate, which is rotaryally driven and has a flat polishing surface, an upper plate, which is arranged opposite the lower plate, is rotaryally driven and has a flat polishing surface, and a support with a wafer-holding hole for holding the wafer.CN 1 01 909 817 A discloses a method wherein the nanotopography of a ground wafer is predicted using data indicating the profile of a wafer ground with a double-sided grinding machine, wherein grinding parameters to improve the nanotopography of the subsequently ground wafers are determined based on the predicted nanotopography, and wherein the operation of the double-sided grinding machine is adjusted according to the determined grinding parameters. STATE OF THE TECHNOLOGY PATENT LITERATURE Patent document 1: JP H05 - 177 539 A Patent document 2: JP 2002 - 100 594 A BRIEF DESCRIPTION OF THE INVENTIONAL PROBLEM

[0008] However, a polished wafer with the specified final thickness does not necessarily have both high overall wafer flatness, such as Global Backside Ideal Range (GBIR), and high outer circumference flatness, such as Site Front Least Squares Range (SFQR) or Edge SFQR (ESFQR). Even if the wafer is polished to a final thickness that results in a good GBIR, a raised area or depression may occur on the outer circumference, which degrades the flatness of the SFQR and ESFQR.

[0009] In recent years, the yield of the latest devices has been significantly affected, particularly by the depressions and protrusions that occur on the outer circumference. Consequently, the flatness of the outer circumference has become more important than the flatness of the entire wafer. However, it is difficult to improve the flatness of the outer circumference solely by controlling the flatness of the entire wafer. Furthermore, the aforementioned method, which uses substrate thickness as a criterion, struggles to account for variations in substrate thickness caused by substrate wear over time.

[0010] Patent document 2, for example, discloses that the thickness of a wheel of a support is configured to be thinner than a wafer-holding area of ​​the support, thus improving the flatness of the outer circumference of a wafer. However, this method limits the structure of a support and cannot sufficiently improve the outer circumference.

[0011] The present invention was developed in consideration of the problems described above. It is an object of the present invention to provide a double-sided polishing method that can measure the shape of a polished wafer to its outermost perimeter with high precision, without reducing productivity, and can improve the flatness of the entire wafer, including its outermost perimeter. SOLUTION TO THE PROBLEM

[0012] To solve this problem, the present invention provides a method for double-sided polishing, which includes a polishing cycle comprising the following steps: performing a first polishing step in which a double-sided polishing process is carried out at a high polishing rate, wherein the double-sided polishing process polishes both surfaces of a wafer simultaneously, the wafer being held by a carrier and arranged between polishing pads mounted on the upper and lower rotary tables while the carrier is rotated and revolved, polishing medium is supplied, and the thickness of the wafer is measured; performing a second polishing step in which a double-sided polishing process is carried out at a low polishing rate;Subdividing a straight line extending from one outermost circumference of the polished wafer through a center of the polished wafer to the other outermost circumference into predetermined sections and optically measuring the cross-sectional shape of the subdivided sections; applying a predetermined weight for each subdivided section to the measured cross-sectional shape to quantify the flatness of each section; and determining the polishing conditions of the first and second polishing steps in a subsequent polishing cycle based on the quantified flatness, wherein a beam diameter of a measuring device used to measure the cross-sectional shape of the outermost sections is smaller than that used to measure the cross-sectional shape of the other section.

[0013] Such a double-sided polishing process can measure the shape of the polished wafer to its outermost perimeter with high precision without reducing productivity, and use the measurement result more precisely in the subsequent polishing cycle to improve the flatness of the entire wafer, including its outermost perimeter.

[0014] The step of determining the polishing conditions of the first and second polishing steps may include setting at least one of the following parameters: the polishing load, the rotational speed and revolution speed of the carrier, and the rotational speed of the upper and lower rotary discs, so that the thickness of the wafer being processed in the first and second polishing steps becomes a target thickness.

[0015] This specific setting of the polishing conditions for the first and second polishing steps allows for an improvement in the flatness of the entire wafer, including its outermost perimeter, in the subsequent polishing cycle.

[0016] The beam diameter of the measuring device, which is used to measure the cross-sectional shape of the outermost sections, is preferably 1 mm or less.

[0017] In this way, the shape of the outermost sections can be measured more reliably and with high precision.

[0018] The cross-sectional shape is preferably measured by optical reflection interferometry using an infrared laser.

[0019] In this way, the cross-sectional shape can be measured with higher precision.

[0020] The step of measuring the cross-sectional shape preferably includes: subdividing the line such that the specified sections are symmetrical with respect to the center of the wafer; calculating a difference ΔA in the thickness of the wafer between both ends of each section; comparing the difference ΔA of each section with that of the corresponding section symmetrical with respect to the center of the wafer and determining that the difference of the corresponding sections is a larger absolute value of the compared differences; and the step of quantifying the flatness includes: quantifying the flatness β of an outer circumference of the wafer from the difference ΔA of the outermost sections;Quantifying the flatness α of the entire wafer except for the outer perimeter by applying each weight to the difference ΔA of the corresponding section in such a way that the weight is greater when the section is closer to the center of the wafer, and adding the difference ΔA of the sections other than the outermost sections.

[0021] In this way, the flatness can be specifically quantified from the measured cross-sectional shape, and the polishing conditions of the first and second polishing steps can be easily determined on the basis of the quantified flatness.

[0022] The flatness α of the entire wafer is preferably quantified such that the flatness α of the section closest to the center of the wafer is converted to 0 if the difference ΔA is 0, to 1 or 2 if the difference ΔA is positive, to -1 or -2 if the difference ΔA is negative; the flatness α of the other section is converted to 0 if the difference ΔA is 0, to 0.5 if the difference ΔA is positive, to -0.5 if the difference ΔA is negative;and the difference ΔA of each section is then added up, the flatness β of the outer circumference of the wafer is preferably quantified such that the flatness β is 0 if the difference ΔA is 0, 1 if the difference ΔA is positive, -1 if the difference ΔA is negative, and the step of determining the polishing conditions of the first and second polishing steps preferably includes determining the target thickness of the wafer to which it is to be polished in the first step in the subsequent polishing cycle by adding βxT to the target thickness in a current polishing cycle, and determining the target thickness of the wafer to which it is to be polished in the second polishing step in the subsequent polishing cycle by subtracting αxT from the target thickness in the current polishing cycle, where T is a difference between the maximum thickness and the minimum thickness of the wafer.

[0023] In this way, flatness can be quantified with higher precision, and the flatness of the entire wafer, including its outermost perimeter, can be improved more effectively.

[0024] The method according to the invention preferably involves blowing air onto a surface of the wafers removed from the carrier.

[0025] Such a method can measure the cross-sectional shape with higher precision. ADVANTAGEOUS EFFECTS OF THE INVENTION

[0026] The inventive method for double-sided polishing comprises a step of measuring the cross-sectional shape of a polished wafer, in which a straight line extending from one outermost perimeter of the wafer through the center of the wafer to the other outermost perimeter is divided into predetermined sections to measure the cross-sectional shape, and a beam diameter of a measuring device used to measure the cross-sectional shape of the outermost sections is smaller than that used to measure the cross-sectional shape of the other sections, enabling a highly precise measurement of the wafer shape to its outermost perimeter without reducing productivity.The process also includes a step of determining the polishing conditions of the first and second polishing steps in the subsequent polishing cycle based on the flatness quantified by the shape measured with high precision, which enables the improvement of the flatness of the entire wafer, including the outermost perimeter. BRIEF DESCRIPTION OF THE DRAWINGS Fig. Figure 1 is a flowchart of an example of the inventive method for double-sided polishing; Fig. Figure 2 is a schematic diagram of an example of a device for double-sided polishing that can be used for the double-sided polishing method according to the invention; Fig. Figure 3 is an illustrative view of an example of the division into sections and the quantification of flatness; Fig.Figure 4 is a schematic diagram of an example of a method for changing the beam diameter of a measuring device; Fig. Figure 5 is a diagram showing the relationship between a polishing load and the shape of the outer circumference of a wafer; Fig. Figure 6 is a diagram that shows, by way of example, the evaluation result of the flatness α and β in the first polishing cycle and the second polishing cycle; Fig. Figure 7 is a diagram showing the evaluation result of the flatness of the outer circumference of polished wafers in an example and a comparison example; and Fig. Figure 8 is a schematic diagram of a typical double-sided polishing device of the planetary gear type. DESCRIPTION OF EXECUTION FORMS

[0027] Embodiments of the present invention are described below, although the present invention is not limited to these embodiments.

[0028] As previously described, the flatness of the outer circumference, which has become particularly important in recent years, needs to be improved. The inventor of the present invention has conducted a study on this problem and discovered the following.

[0029] The inventor first addressed a conventional method for evaluating the flatness of the outer circumference of a polished wafer: the shape of the outer circumference is optically measured within a measuring range that excludes an area extending up to 2 mm from the outer circumference. The reason for excluding this area is to prevent the measurement from being influenced by a measuring laser beam reflected from the outer circumference, particularly a chamfered portion. The inventor found that excluding this area from the measuring range reduces the precision of the flatness measurement.

[0030] Next, the inventor sought a method to solve this problem and came up with the following: The beam diameter of a measuring device is changed according to the position at which the device measures the cross-sectional shape of a wafer. Specifically, the cross-sectional shape of the wafer's outermost perimeter is measured using a measuring device with a smaller beam diameter. This method can add the outermost perimeter, which is normally an excluded area, to the measuring range to improve measurement accuracy while preventing a significant increase in measurement time. This allows the shape to be measured with high precision right up to the outermost perimeter.

[0031] Finally, the inventor came up with the following: In the form measurement process, a straight line extending from one outermost perimeter of the wafer through the wafer's center to the other outermost perimeter is divided into predetermined segments. Each of these segments is assigned a predetermined weight, in particular such that the weight is greater the segment is closer to the wafer's center. The flatness is evaluated in each segment, to which the corresponding weight is added. This method allows for a higher degree of precision in evaluating flatness. The result of the evaluation is used to determine the polishing conditions in the subsequent polishing cycle accordingly, thereby enabling a more effective improvement in the flatness of the entire wafer, including that of the outermost perimeter. The inventor thereby completed the invention.

[0032] A device for double-sided polishing, which can be used for the double-sided polishing method according to the invention, is now described. As in Fig. As shown in Figure 2, the double-sided polishing device 1 has a cylindrical upper rotary disk 2 and a cylindrical lower rotary disk 3. Polishing pads 4 are attached to the upper and lower rotary disks 2 and 3, with all polishing surfaces facing each other. The polishing pads 4 are made of a non-woven fabric impregnated with a urethane resin or urethane foam. A sun gear 7 is located in the lower rotary disk 3. An inner gear 8 is located outside the lower rotary disk 3. The upper and lower rotary disks 2 and 3, the sun gear 7, and the inner gear 8 have the same axis of rotation and can rotate independently around this axis.

[0033] The carrier 5 is provided with holding openings 6 to hold a wafer W. A plurality of carriers 5 can be arranged between the upper and lower turntables 2, 3. The holding openings 6 provided in each of the carriers 5 allow a plurality of wafers W to be polished in each polishing cycle. Each carrier 5 engages the sun gear 7 and the inner gear 8 and can rotate about its axis and, according to the rotational speed of the sun gear 7 and the inner gear 8, revolve around the aforementioned axis of rotation between the upper and lower turntables. The wafers W are inserted and held in the holding openings 6 of the carriers 5. The upper turntable 2 is lowered so that the wafers W and the carriers 5 are inserted between the turntables, thereby applying a polishing load to the wafers W.The rotary disk 2 and the lower rotary disk 3 are rotated in opposite directions, while a polishing agent supplied by a nozzle 10 is poured through through openings 9 formed in the upper rotary disk 2 into a space between the rotary disks, so that both surfaces of the wafer W are polished simultaneously.

[0034] The device for double-sided polishing also includes a measuring device for measuring the shape of the polished wafers removed from the carriers, as well as a transfer robot for transferring the wafer to the measuring device.

[0035] An example of a measuring device is a device for measuring the thickness of a wafer using an infrared laser with a variable wavelength that can be transmitted through the wafer. The measuring device can adjust the diameter of the laser beam emitted onto the wafer.

[0036] The double-sided polishing method according to the invention uses the double-sided polishing apparatus of this type to polish wafers by a first polishing step with a high polishing rate and a subsequent second polishing step with a low polishing rate. The first polishing step removes deformations and depressions that have arisen on the wafer surface mainly from a previous process and is performed with a high polishing load and a high polishing rate to improve productivity. The second polishing step mainly restores flatness and is performed with a low polishing rate. The polishing conditions of these steps are suitably determined based on the flatness evaluation result in the previous polishing cycle. The aforementioned double-sided polishing apparatus, which can be used for the invention, is now described by way of example and does not limit the invention.

[0037] The inventive method for double-sided polishing is now described in more detail with reference to Fig. 1 described.

[0038] As in Fig. As shown in Figure 1, a current polishing cycle begins while the thickness of the wafer to be polished is measured (S1 in Fig. 1) The wafer thickness is preferably measured by optical reflection interferometry using an infrared laser, which improves the measurement accuracy. In the polishing cycle, the first polishing step is performed first, as described above (S2 in Fig. 1) The polishing conditions of the first polishing step are determined in the previous polishing cycle, as described below.

[0039] The second polishing step is then performed (S3 in Fig.1) The polishing conditions for the second polishing step are also determined in the previous polishing cycle, as described below. The second polishing step is complete when the wafer thickness has reached the target final thickness (S4 in Fig. 1) The polished wafers are removed from the carriers (S5 in Fig. 1) At least one of the wafers is transferred to the measuring device to measure the cross-sectional shape of the wafer (S6 in Fig. 1) Before measuring the cross-sectional shape of the wafer, air is preferably blown onto the wafer surface to remove the polishing agent and water adhering to the wafer (S7 in Fig. 1) This improves measurement accuracy.

[0040] Then the cross-sectional shape of the polished wafer is measured. In this measurement, the straight line extending from one outermost perimeter of the wafer through the wafer's center to the other outermost perimeter is divided into predefined sections, and the cross-sectional shape of these sections is measured optically. The measurement is performed in two steps: measurement of the outermost sections, and measurement of the other sections. The cross-sectional shape of the entire wafer, excluding the outermost sections, is measured first (S8 in Fig. 1) In this measurement, the beam diameter of the measuring device can be approximately 3 mm, which is the same diameter as conventionally used. However, this value does not limit the invention. After this measurement, the beam diameter is changed to a smaller one in order to measure the cross-sectional shape of the outermost sections (S9 in Fig.1) After the beam diameter has been changed, the cross-sectional shape of the outermost sections is measured (S10 in Fig. 1).

[0041] The outermost section comprises a portion of the chamfered part (a portion with chamfer widths X1 and X2 of the main wafer surfaces) and the excluded outermost perimeter of 2 mm. When measuring this section, which includes the 2 mm outermost perimeter and the portion of the chamfered area, using optical reflection interferometry with a beam diameter of 1 mm or less reduces the intensity of the scattered light from the X1 and X2 portions of the chamfered part to less than half that of the incident beam. This allows the reflected light to be reliably received from a portion of the outermost perimeter up to 1 mm away. This is preferred because the cross-sectional shape of the outer perimeter can be measured stably.In particular, a beam diameter of about 0.06 mm is preferable because this diameter provides a focal depth of about 1.1 mm, which makes it easy for an optical measuring system to adjust the focus regardless of variations in the wafer thickness, so that the flatness of the outer circumference can be easily measured.

[0042] A beam diameter of 0.04 mm or less is impractical because this diameter makes it difficult for the optical system to adjust the focus.

[0043] The method according to the invention measures the shape of the outermost sections with the measuring device using a beam diameter that is smaller than that used to measure the shape of the other sections. This allows the shape measurement to be achieved up to the outermost circumference in a short time, thus preventing a reduction in productivity. The method can also prevent the influence of a laser beam reflected from the chamfered part during the measurement, thereby improving the precision of the shape measurement.

[0044] As in Fig. As shown in Figure 3, the straight line extending from an outermost circumference of the wafer through the center of the wafer can, for example, be divided into sections such that the sections are symmetrical with respect to the center of the wafer. Fig.Figure 3 shows the position of the center of a wafer with C, and the position of the outermost circumference of the wafer with n or n'. The value of n is an integer of 2 or greater. The thickness of the wafer at position n is given by A. n shown. In Fig. The outermost sections are a section from n to n-1 and a section from n' to n-1'. A complete section, excluding the outermost sections, consists of sections from n-1 to n-1'.

[0045] The cross-sectional shape is preferably measured by optical reflection interferometry with an infrared laser to improve measurement accuracy. Measuring the cross-sectional shape can be done in conjunction with measuring the wafer thickness.

[0046] As in Fig.As shown in Figure 4, the beam diameter can be reduced, for example, by focusing a laser beam 11 from t1 to t2 using a focusing lens 12. If the beam diameter is 0.06 mm, then the focal depth d is approximately 100 mm. This diameter makes it possible to reduce the size and cost of the measuring device.

[0047] The difference T (T max -T m i n ) between the maximum thickness T max and the minimum thickness T min The cross-sectional shape of the measured cross-section is then calculated. This cross-sectional shape corresponds to that of the one in Fig. 3 sections shown between A n and A n 'The weight predetermined for each subdivided section is applied to the measured cross-sectional shape to quantify the flatness of each section (p. 11 in Fig.1) Flatness can be quantified, for example, by a difference ΔA in wafer thickness between the two ends of each subdivided section. The quantification is described below with reference to Fig. 3 described. The difference (ΔA n-1 = A n-1 -A n The difference (ΔA) in the wafer thickness between the two ends n-1 and n of the outermost section is calculated. n-1 ' = A n -1'-A n The wafer thickness between the two ends n-1' and n' of the other outermost section, which is symmetrical to this outermost section with respect to the wafer center, is also calculated. The flatness β of the outer circumference of the wafer is determined by a larger value of absolute values ​​of ΔA. n -1 and ΔA n-1' quantified. The value of the difference ΔA is negative if the outer circumference has the form of a raised area; the value of the difference ΔA is positive if the outer circumference has the form of a depression.

[0048] The flatness α of the entire wafer, excluding the outer perimeter, is quantified similarly as follows: The difference in wafer thickness between both ends of each section (from ΔAn-2 to A0 and from ΔAn- 2' to A 0'The difference ΔA of each section, with the exception of the outermost sections, is calculated. The difference ΔA of each section is compared to that of the corresponding section symmetrical to it with respect to the wafer center. The difference ΔA of these sections is determined to be a larger absolute value of the compared differences. The weight, predetermined for each section, is applied to the difference ΔA of the corresponding section. The resulting difference ΔA of each section is summed. The flatness α of the entire wafer is the sum of the summed differences. The value of the difference ΔA is negative if the overall shape of the wafer is concave; the value of the difference ΔA is positive if the overall shape of the wafer is convex.

[0049] The weight can be predetermined, for example, so that the weight is greater when the section is closer to the center of the wafer. Tables 1 and 2 show a specific example of the flatness obtained after weighting in the [reference to be added]. Fig. The example shown in Figure 3, where n=3, that is, where there are three pairs of symmetrical sections with respect to the wafer center, is illustrated in Table 1. Specifically, Table 1 shows an example of the quantified flatness of the outermost section. As shown in Table 1, if the shape is flat, the weight and flatness are 0; if the shape is a protrusion or a depression, the weight is 1; if the shape is a protrusion, the flatness is -1; and if the shape is a depression, the flatness is 1. These values ​​can be considered the flatness β of the outer circumference.

[0050] Table 2 shows an example of the quantified flatness of section 1-2 (outer) and section C-1 (inner). In Table 2, the number of the "+" symbol for concave and convex shapes represents the degree of the shape, and the larger the number, the greater the degree. If the shape is flat, the weight and flatness are 0; if the shape is a protrusion or depression, the weight for section 1-2 is 0.5 and the flatness of this section is 0.5 if the shape is convex, or -0.5 if the shape is concave; if the shape is a protrusion or depression, the weight for section C-1 is 1 or 2, and the flatness of this section is 1 or 2 if the shape is convex, or -1 or -2 if the shape is concave. The flatness a of the entire wafer can be calculated from the flatness of the sections by comparing and adding the symmetrical sections as described above. [Table 1] CROSS-SECTIONAL SHAPE OUTER SECTION PLAINITY (SECTIONS 2-3) SURVEY -1 FLAT 0 IN-DEPTH 1 [Table 2] CROSS-SECTIONAL SHAPE PLANNING OF SECTIONS 1-2 PLANNING OF SECTION C-1 CONVEX++ 0,5 2 CONVEX + 1 FLAT 0 0 CONCAVE+ -0,5 -1 CONCAVE++ -2

[0051] The polishing conditions of the first and second polishing steps in the subsequent polishing cycle are determined on the basis of the aforementioned quantified flatness (S12 in Fig. 1) These conditions can be determined such that both the first and second polishing steps polish the wafers to their respective target thicknesses by adjusting at least one of the following parameters: polishing load, rotational speed and revolution speed of the carrier, and rotational speed of the upper and lower rotary tables. These target wafer thicknesses after the respective polishing steps can be determined, for example, as follows.

[0052] The target wafer thickness of the first polishing step in the subsequent polishing cycle is calculated by multiplying the obtained flatness β of the outer circumference by the difference T between the maximum and minimum wafer thicknesses and adding the result to the target thickness in the current polishing cycle. For example, if the shape of the outer circumference is a raised area, adding the value βxT to the target thickness in the current polishing cycle will decrease the target thickness, since the value of β is negative. The target thickness can be decreased by increasing the polishing load or by increasing the rotational speed of the rotary discs and the rotational and revolution speeds of the carrier to increase the polishing rate. Conversely, if the shape of the outer circumference is a depression, adding the value βxT to the target thickness in the current polishing cycle will increase the target thickness, since the value of β is positive.The target thickness can be increased by reducing the polishing load or by reducing the rotational speed of the rotary discs and the rotational speed and revolution speed of the carrier to reduce the polishing rate.

[0053] Fig. Figure 5 shows the result of the evaluation of variations in the shape of the wafer's outer circumference in an example where the polishing load is changed. As in Fig. As shown in Figure 5, when the polishing load increases, the shape of the outer circumference tends to be a depression; when the polishing load decreases, the shape of the outer circumference tends to be a raised area. The shape of the outer circumference can therefore be adjusted by changing the polishing load based on this relationship.

[0054] The target wafer thickness for the second polishing step in the subsequent polishing cycle is calculated by subtracting the flatness α of the entire wafer from the target thickness in the current polishing cycle. For example, if the flatness of the entire wafer is concave, subtracting the flatness α of the entire wafer (αxT) from the target thickness in the current polishing cycle increases the target thickness, since the value of α is negative. Conversely, if the flatness of the entire wafer is convex, subtracting the flatness α of the entire wafer (xT) from the target thickness in the current polishing cycle decreases the target thickness, since the value of α is positive. In these cases, the polishing load, the rotational speed of the rotary discs, and the rotational speed and revolution speed of the carrier can be adjusted in the same way as described above.

[0055] Finally, the subsequent polishing cycle is carried out under the aforementioned specific polishing conditions (S13 in Fig. 1) The method can use a control device to automatically perform the quantification of flatness from the measured cross-sectional shape and the determination of polishing conditions based on the quantified flatness, which improves productivity.

[0056] The inventive method for double-sided polishing can measure the shape of the polished wafer up to its outer circumference with high precision and improve the flatness of the entire wafer, including its outermost circumference, by precisely using the flatness quantified by this measured shape in the subsequent polishing cycle. This method can also prevent an increase in measurement time and thus a reduction in productivity. EXAMPLE

[0057] The present invention is described in more detail with reference to an example and a comparative example, although the present invention is not limited to this example. (Example)

[0058] A silicon single-crystal wafer with a diameter of 300 mm was manufactured according to the method described in Fig. The inventive method for double-sided polishing shown in Figure 1 is double-polished. This example began with the preparation of wafers by cutting a silicon single-crystal block grown using the Czochralski (CZ) method into wafers, as well as chamfering, overlapping, and etching the wafers. The polishing was carried out with a Fig.The device shown in Figure 2 is designed for double-sided polishing. The double-sided polishing device uses urethane foam polishing pads on both sides of the upper and lower rotary discs and a polishing compound containing colloidal silicon dioxide granules, the pH of which has been adjusted to between 10.0 and 11.0.

[0059] The polishing conditions for the first polishing step in the first polishing cycle were as follows: the polishing load was 150 g / cm²; the rotational speeds of the upper and lower rotary disks, the sun wheel, and the inner wheel were determined such that the difference between the rotational speed of the carrier and the rotational speed of the upper rotary disk, and the difference between the rotational speed of the lower rotary disk and the rotational speed of the carrier, was 10 rpm; the rotational speed of the carrier was 2.5 rpm; the target thickness of the first polishing step was 785 µm; the target thickness of the second polishing step, the final thickness, was 780 µm.

[0060] After the first polishing cycle, the flatness was assessed. The result was that the outer circumference had a raised shape and the overall shape was concave. As in Fig.The values ​​shown in (A) were used to determine the polishing conditions in the second polishing cycle as follows: the polishing load of the first polishing step was increased to reduce the target thickness to 784.909 µm (785 µm + β(-1)× T (0.091)). The polishing load of the second polishing step was decreased to increase the target thickness to 780.136 µm (780 µm - α(-1.5) × T (0.091)).

[0061] After the second polishing cycle, the flatness was assessed. The result was as described in Fig. 6 at (B) showed that the flatness of the entire wafer, including the outermost perimeter, was improved.

[0062] The beam diameter of the measuring device used to measure the outermost sections was 0.06 mm.

[0063] The polishing cycle described above was repeated to evaluate the flatness of the outer circumference of the polished wafer. The result is shown in Fig.7. The relative value of the assessed flatness of the outer perimeter is a value converted from a standard value between 0 and 100, where a value closer to 0 represents better flatness. As in Fig. As shown in Figure 7, it was confirmed that the flatness in the example was improved more than in the comparison example. (Comparative example)

[0064] The polishing cycle was repeated under the same conditions to perform the same evaluation as in the example, except that the polishing conditions were not changed after each polishing cycle.

[0065] The result of the evaluated flatness of the outer circumference of the polished wafer is given in Table 7. As in Fig. As shown in 7, the flatness was worse than in the example.

Claims

[1] Method for double-sided polishing comprising a polishing cycle comprising: Performing a first polishing step in which a double-sided polishing process is carried out at a high polishing rate, wherein the double-sided polishing process simultaneously polishes both surfaces of a wafer which is held by a carrier and arranged between polishing pads, each arranged on an upper and a lower rotary disk, while the carrier is rotated and revolved, polishing medium is supplied and a thickness of the wafer is measured. Performing a second polishing step in which the double-sided polishing process is carried out at a low polishing rate, Subdividing a straight line extending from one outermost circumference of the polished wafer through a center of the polished wafer to the other outermost circumference into predetermined sections, and optically measuring the cross-sectional shape of the subdivided sections; Applying a predetermined weight for each subdivided section to the measured cross-sectional shape to quantify the flatness of each section; and Determining the polishing conditions of the first and second polishing steps in a subsequent polishing cycle based on quantified flatness, wherein A beam diameter of a measuring device used to measure the cross-sectional shape of outermost sections is smaller than the one used to measure the cross-sectional shape of the other section. [2] Method for double-sided polishing according to claim 1, where the step of determining the polishing conditions of the first and second polishing steps involves setting at least one of the following parameters: a polishing load, a rotational speed and a revolution speed of the carrier as well as a rotational speed of the upper and lower rotary disk, so that the thickness of the wafer treated in the first and second polishing steps becomes a target thickness. [3] Method for double-sided polishing according to claim 1 or claim 2, wherein the beam diameter of the measuring device used to measure the cross-sectional shape of the outermost sections is 1 mm or less. [4] Method for double-sided polishing according to any one of claims 1 to 3, wherein the cross-sectional shape is measured by optical reflection interferometry with an infrared laser. [5] Method for double-sided polishing according to any one of claims 1 to 4, wherein The step of measuring the cross-sectional shape includes: Subdivide the line such that the specified sections are symmetrical with respect to the center of the wafer; calculate a difference ΔA in the thickness of the wafer between both ends of each section; Comparing the difference ΔA of each section with that of the corresponding section symmetrical about the center of the wafer and determining that the difference of the corresponding sections is a larger value of the absolute values ​​of the compared differences; and The step of quantifying flatness includes: Quantifying the flatness β of an outer circumference of the wafer from the difference ΔA of the outermost sections; quantifying the flatness α of the entire wafer except for the outer circumference by applying each weight to the difference ΔA of the corresponding section, such that if the section is closer to the center of the wafer, the weight is greater, and summing the difference ΔA of the sections except for the outermost sections. [6] Method for double-sided polishing according to claim 5, wherein The flatness α of the entire wafer is quantified such that the flatness α of the section closest to the center of the wafer is converted to 0 if the difference ΔA is 0, to 1 or 2 if the difference ΔA is positive, to -1 or -2 if the difference ΔA is negative; the flatness α of the other section is converted to 0 if the difference ΔA is 0, to 0.5 if the difference ΔA is positive, to -0.5 if the difference ΔA is negative; and the difference ΔA of each section is then summed. The flatness β of the outer circumference of the wafer is quantified such that the flatness β is 0 if the difference ΔA is 0, 1 if the difference ΔA is positive, -1 if the difference ΔA is negative, and The step of determining the polishing conditions of the first and second polishing steps includes determining the target thickness of the wafer to which polishing is to be performed in the first polishing step, which is to be carried out in the subsequent polishing cycle, by adding βxT to the target thickness in a current polishing cycle, and determining the target thickness of the wafer to which polishing is to be performed in the second polishing step, which is to be carried out in the subsequent polishing cycle, by subtracting αxT from the target thickness in the current polishing cycle, where T is a difference between the maximum thickness and the minimum thickness of the wafer. [7] Method for double-sided polishing according to any one of claims 1 to 6, further comprising blowing air onto a surface of the wafer removed from the support.

Images