JP7587255B2 - Double-sided polishing machine - Google Patents

Description

The present invention relates to a double-sided polishing device for workpieces such as semiconductor wafers.

There is a double-sided polishing device in which a workpiece held by a carrier is sandwiched between upper and lower platens and polished while revolving and rotating together with the carrier due to the rotation of a sun gear and an internal gear.
As shown in Patent Documents 1 to 3, these double-sided polishing machines are equipped with a workpiece thickness measuring unit that can measure the thickness of the workpiece in real time while the workpiece is being polished.

The workpiece thickness measuring section in the double-sided polishing machine in Patent Document 1 (JP Patent Publication 2015-47656 A) is structured with one or more workpiece measurement holes that penetrate either the upper or lower base plate, and further, when measuring the workpiece thickness, the rotation of the sun gear and the internal gear is synchronously controlled to stop the revolution of the carrier and allow only rotation, so that the thickness of the workpiece is measured while being polished.

The workpiece measuring unit in the double-sided polishing machine in Patent Document 2 (JP 2017-204609 A) and Patent Document 3 (JP 2017-207455 A) is capable of determining the measurement position (coordinates on the workpiece) of the workpiece thickness based on the rotational positions of the sun gear and internal gear during polishing, and by importing the thickness measurement position and the measured value in association with each other, it is possible to obtain the continuous thickness distribution (radial cross-sectional shape) of the workpiece through calculation processing.

JP 2015-47656 A JP 2017-204609 A JP 2017-207455 A

In the double-sided polishing device described in Patent Document 1, when the thickness of the workpiece is measured, the revolution of the carrier is stopped during polishing, so control is required to keep the carrier rotating on its axis only. In addition, stopping the revolution of the carrier, which is necessary for polishing, has the problem of adversely affecting the machining accuracy and efficiency of the workpiece.

In the devices of Patent Documents 2 and 3, as shown in FIG. 15, there may be a discrepancy between the actual thickness measurement position and the thickness measurement position calculated from the rotational position information of the sun gear and internal gear due to a time lag caused by delays in obtaining and calculating the rotational position information of the sun gear and internal gear, as well as errors in the rotational position information (backlash of various gears). FIG. 15 is a graph showing the relationship between the polar coordinate position of the carrier measurement position versus time (horizontal axis) (the dashed line is the calculated value, and the solid line is the actual measured value). There is a discrepancy between the calculated value and the actual measured value.

Therefore, it is not possible to grasp the exact thickness measurement position (the position on the coordinate system of the workpiece), and an accurate thickness distribution cannot be obtained. In addition, due to a deviation between the thickness measurement position obtained from the rotational position information and the actual thickness measurement position , there is a problem that the workpiece 21 does not exist, and the thickness of the slurry layer 19 on the carrier 20 may be mistaken for the thickness of the workpiece 21, as shown in Fig. 16. In other words, there is a problem that the thickness of the slurry layer 19 is measured due to interference of reflected light from the front and back surfaces of the slurry layer 19.

The present invention was made to solve the above problems, and its purpose is to provide a double-sided polishing machine that does not require the rotation of the carrier to be stopped during polishing, and does not mistake the thickness of the slurry layer for the thickness of the workpiece.

In order to achieve the above object, the present invention has the following configuration.
That is, the double-sided polishing apparatus according to the present invention is a double-sided polishing apparatus comprising: a ring-shaped lower surface plate having a polishing pad fixed to its upper surface and rotatable about a rotation axis; a ring-shaped upper surface plate having a polishing pad fixed to its lower surface and movable up and down above the lower surface plate and rotatable about the rotation axis; a sun gear disposed at the center of the lower surface plate; an internal gear disposed surrounding the lower surface plate; and a carrier disposed between the lower surface plate and the upper surface plate, having a through hole for holding a workpiece, meshing with the sun gear and the internal gear, revolving around the sun gear and rotating on its axis, the carrier including an optical system disposed on either the upper surface plate or the lower surface plate and rotating together with the surface plate, which irradiates the workpiece with laser light through a measurement hole provided in the surface plate, and a thickness measurement unit capable of measuring the thickness of the workpiece by receiving interference light of reflected light or transmitted light from the workpiece. the measurement hole; a position calculation unit that calculates the path of the laser light passing through the workpiece held on the carrier as position information from the center of the workpiece in relation to the passing time based on the rotational speed of the base plate, the rotational speed of the sun gear, and the rotational speed of the internal gear; a position correction unit that corrects the position information of the workpiece calculated by the position calculation unit based on the deviation between the actual passing time of the laser light passing through the actual boundary between the carrier and the workpiece, which is measured from the detection signal from the carrier detection sensor, and the calculated time of the laser light passing through the boundary between the carrier and the workpiece, which is calculated by the position calculation unit; and a control unit that terminates polishing of the workpiece when the thickness of the workpiece at the required position of the workpiece corrected by the position correction unit reaches a set thickness.

It is preferable that the control unit controls so as to end polishing of the workpiece when the thickness of the inner side immediately adjacent to the edge portion of the workpiece reaches a set thickness.
The sensor portion of the carrier detection sensor may be composed of a high frequency coil that induces an eddy current in the carrier.
The detection sensitivity of the carrier can be improved by disposing a sensor portion of the high frequency coil in the measurement hole provided in the upper surface plate and adjacent to the lower surface of the upper surface plate.

The high frequency coil and the optical system can be set so that the high frequency coil and the laser light irradiated from the optical system to the workpiece are coaxial.
It is preferable that the sensor portion of the carrier detection sensor is covered with a fluororesin or electromagnetically shielded.
It is preferable to provide a sensor amplifier for amplifying a detection signal from a sensor portion of the carrier detection sensor, and adjust the sensitivity of carrier detection on the sensor amplifier side, or set a threshold value to turn carrier detection on and off.

By disposing an upper cap fixed to the top of the measurement hole, a lower cap fixed to the bottom of the measurement hole, and a window body fixed coaxially to the upper cap and lower cap and having a transparent plate that allows the laser light to pass through, it is possible to prevent liquids such as slurry and water from entering the measurement hole.
It is preferable to provide a protective plate made of resin so as to cover the lower surface of the lower cap other than the transparent plate.

The thickness measurement unit has a light source monitoring circuit using a reference work, and the light source monitoring circuit can be composed of a second optical system that guides and irradiates laser light from a wavelength swept laser light source to the reference work, a second detector that detects an interference light signal of reflected light or transmitted light obtained from the reference work, and a DAQ that converts the interference light signal detected by the second detector into a digital signal.

The thickness measurement unit has a Mach-Zehnder interferometer and a detector, and the Mach-Zehnder interferometer also serves as a light source monitoring circuit, which can monitor the sweep wavelength accuracy, which can be grasped by the average value, PP value, or deviation value of a set number of measurements, for the frequency of the FFT peak value of the interference waveform acquired by the detector.

According to the present invention, the following advantageous effects are obtained.
That is, according to the present invention, it is possible to provide a double-sided polishing apparatus which does not require stopping the revolution of the carrier during polishing and which does not mistake the thickness of the slurry layer for the thickness of the workpiece.

FIG. 2 is a schematic cross-sectional view of a double-sided polishing apparatus. 11 is a plan view showing the arrangement of carriers on a lower platen. FIG. FIG. 2 is a block diagram showing an outline of a thickness measuring unit. 4A and 4B are explanatory diagrams showing the detection state of the carrier and the work when neither the work nor the carrier is present on the lower surface plate. 5A and 5B are explanatory diagrams showing the detection state of the carrier and the workpiece when the carrier enters below the high frequency coil. 6A and 6B are explanatory diagrams showing the detection state of the carrier and the workpiece when the workpiece enters below the probe. 7A and 7B are explanatory diagrams showing the detection state of the carrier and the workpiece when the laser light passes over the lines L1, L2, and L3 on the carrier. This is a simulation diagram showing the path of laser light passing momentarily over 15 workpieces, calculated from the rotation speed of the upper platen, the rotation speed of the sun gear, and the rotation speed of the internal gear. This is a simulation diagram showing the path of laser light passing momentarily over 15 workpieces, calculated from the rotation speed of the upper platen, the rotation speed of the sun gear, and the rotation speed of the internal gear. This is a simulation diagram showing the path of laser light passing momentarily over 15 workpieces, calculated from the rotation speed of the upper platen, the rotation speed of the sun gear, and the rotation speed of the internal gear. This is a simulation diagram showing the path of laser light passing momentarily over 15 workpieces, calculated from the rotation speed of the upper platen, the rotation speed of the sun gear, and the rotation speed of the internal gear. FIG. 2 is a block diagram of a control unit that controls the entire system. 13A and 13B are explanatory diagrams of the configuration of a window body incorporating a carrier detection sensor. FIG. 13 is an explanatory diagram in which the carrier detection sensor and the probe are set at different positions. 11 is a graph showing the difference between the calculated value and the actually measured value of the rotational position of the carrier with respect to time. FIG. 13 is an explanatory diagram showing a state in which the thickness of the slurry on the carrier is detected. FIG. 13 is a block diagram showing an outline of another thickness measuring unit for measuring the thickness of a workpiece. FIG. 18 is a flow diagram of signal processing showing a thickness measurement process including an optical system in the thickness measurement unit shown in FIG. 17. 13 is a graph showing a direct interference waveform from the workpiece, which is input to the computer simultaneously with the trigger signal from the laser light source and the internal clock of the A/D board, and an interference waveform input to the computer from the MZI interferometer 72 . FIG. 1 is a block diagram of a DAQ. 7A is a graph showing the state of the power spectrum when FFT processed data is not averaged (FIG. 7A) and when FFT processed data is averaged (FIG. 7B). FIG. 13 is a block diagram of the entire system showing still another embodiment of the thickness measuring unit .

Hereinafter, preferred embodiments of the present invention will be described in detail with reference to the accompanying drawings.
First, an example of a double-sided polishing machine will be described. Since a known double-sided polishing machine may be used, a brief description will be given below.
Fig. 1 is a schematic cross-sectional view of a double-sided polishing machine 10 for polishing workpieces such as silicon wafers, and Fig. 2 is a plan view showing the arrangement of carriers on a lower platen.
In the apparatus shown in FIG. 1, a carrier 20 driven by an internal gear 15 and a sun gear 16 is disposed between a lower platen 12 and an upper platen 14 which rotate in opposite directions.
Polishing pads 17 and 18 are attached to the upper surface of the lower surface plate 12 and the lower surface of the upper surface plate 14, respectively.

The carrier 20 has holes 22 for holding the workpiece 21 to be polished therein. In this embodiment, as shown in FIG. 2, five carriers 20 that mesh with the sun gear 16 and the internal gear 15 are arranged at regular intervals in the circumferential direction. Each carrier 20 has three holes 22. Therefore, a total of 15 workpieces 21 can be polished at the same time.

The carrier 20 has a planetary mechanism structure that meshes with the sun gear 16 and the internal gear 15, and revolves around the sun gear 16 and rotates about its own axis.
The carrier 20 is made of a metal such as titanium.
The lower surface plate 12 is rotatably disposed on a surface plate receiver (not shown). The upper surface plate 14 is supported via a rod 25 on a disk 24 which is supported via a suspension column 23 on a gate-shaped column (not shown) so as to be movable up and down and rotatable therewith.

The lower surface plate 12, the upper surface plate 14, the sun gear 16, and the internal gear 15 are each adapted to be rotated by a known driving mechanism (not shown).
As described above, during polishing, the lower platen 12 and the upper platen 14 are rotated in opposite directions, and the carrier 20 revolves and rotates around the sun gear 16, thereby polishing the workpiece 21 held in the through hole 22 of the carrier 20.

During polishing, a slurry containing an abrasive is supplied onto the lower platen 12 from a slurry supply unit (not shown) provided on the upper platen 14. The slurry supply unit supplies the slurry onto the lower platen 12 through a through hole provided in the upper platen 14 via a hose (not shown).

FIG. 3 is a block diagram showing an outline of a thickness measuring unit 30 for measuring the thickness of the workpiece 21. As shown in FIG.
Reference numeral 31 denotes a known wavelength sweep type laser light source. The laser light source 31 repeatedly sweeps the wavelength over all or part of the same wavelength range, for example, at a wavelength sweep speed of 1 to 50 kHz and a wavelength variable range of 1200 to 1400 nm. The laser light emitted from the laser light source 31 is irradiated onto the measurement site of the workpiece 21 from a circulator 32, a rotary joint 33, and a probe (optical system) 34. The rotary joint 33 is disposed on a suspension support 23 which is the center of rotation of the upper surface plate 14.

The probe 34 is fixed on the upper surface plate 14 and rotates together with the upper surface plate 14. Laser light is irradiated from the probe 34 onto the workpiece 21 through a measurement hole 35 provided in the upper surface plate 14. The measurement hole 35 is preferably located at a position on the upper surface plate 14 that corresponds to approximately the center of the width of the carrier 20 (a position equivalent to 4/5 of the surface plate radius from the center of the upper surface plate 14) so that it passes evenly over all 15 workpieces 21 held by the carrier 20. The probe 34 may be fixed to the lower surface plate 12 (not shown) and configured to irradiate laser light toward the workpiece 21 from a measurement hole (not shown) provided in the lower surface plate 12.

The laser light (interference light) reflected by the front and back surfaces of the workpiece 21 is detected by the photodiode 36 via the measurement hole 35, the probe 34, the rotary joint 33 and the circulator 32, and is converted into an electrical signal (interference light signal) by the photodiode 36, which is then amplified by the amplifier.

The laser light reflected from the front surface of the workpiece 21 and the laser light reflected from the rear surface interfere with each other and are observed as interference light having a required phase.
It is also possible to observe interference light between the laser light passing through the workpiece 21 and the laser light reflected from the rear surface of the workpiece 21, then reflected from the front surface and passing through the rear surface side.
The interference light signal is converted into a digital signal by a DAQ (data acquisition device: digitizer) 37 including an A/D converter and an FFT processor, and then FFT processing is performed to obtain data on the peak value of the interference waveform. The data is then input to a thickness calculation unit 38 in the control unit 50 ( FIG. 12 ).
An external trigger signal generated by the laser light source 31 is input to the DAQ 37 .

In the thickness calculation unit 38, the center frequency (peak value) is calculated from the FFT processed data from the DAQ 37, and the thickness of the workpiece 21 is calculated by a known formula including the refractive index of the workpiece 21.
The FFT process may be performed by the thickness calculation unit 38 .

Reference numeral 40 denotes a carrier detection sensor (proximity sensor) disposed in the measurement hole 35, and has a high-frequency coil (sensor portion) 41 provided at a portion facing the lower surface plate 12 (carrier 20) (FIGS. 4 to 6). The high-frequency coil 41 is fixed to the lower surface of a hollow support 43. In order to increase the detection sensitivity of the carrier 20, the high-frequency coil 41 is preferably located as close to the carrier 20 as possible, preferably about 3 mm (about 4.5 mm at the limit). It is preferable that the high-frequency coil 41 is at the same height as the lower surface of the upper surface plate 14. In this case, the distance between the high-frequency coil 41 and the carrier 20 is about the thickness of the polishing pad 17.
A high-frequency current is applied to the high-frequency coil 41 from an AC power source (not shown).
Reference numeral 42 denotes a lead wire of the high frequency coil 41 .

In this embodiment, the high-frequency coil 41 is provided so as to be coaxial with the laser light irradiated from the probe (optical system) 34 to the workpiece 21. The laser light from the probe 34 passes through the hollow support 43 and the high-frequency coil 41 and is irradiated to the workpiece 21.
The high frequency coil 41 which is the sensor portion of the carrier detection sensor 40 is preferably covered with a fluororesin (not shown) to improve corrosion resistance.
It is also preferable to electromagnetically shield the high frequency coil 41 to prevent magnetic interaction with the steel upper platen 14 .

As described above, the laser light emitted from the probe 34 passes through the measurement hole 35 and is irradiated onto the workpiece 21. The reflected light also passes through the measurement hole 35 and is guided from the probe 34 and rotary joint 33 to the circulator 32.

When the metal carrier 20 passes below the high-frequency coil 41, i.e., below the measurement hole 35, eddy currents are generated in the carrier 20 due to electromagnetic induction, and the impedance of the high-frequency coil 41 changes due to the magnetic field caused by these eddy currents. This change in impedance is then detected to detect the presence of the carrier 20.

The presence of the carrier 20 can also be detected as follows. For example, an induced current caused by the magnetic field of the eddy current in a secondary coil (receiving coil: not shown) arranged coaxially with the high frequency coil 41 is captured, and this induced current is amplified by the sensor amplifier 39, passes through the rotary joint 33, and is processed by a processing unit including the DAQ 37 to generate a carrier detection signal. In this case, the sensitivity of the carrier 20 detection can be adjusted by the sensor amplifier 39, or a threshold value can be set in the DAQ 37 to turn carrier detection on and off.

As will be described later, a detection signal from this carrier 20 is input to a position correction section 52 (FIG. 12) in the control section 50.
The rotary joint 33 has a structure such as a slip ring, and is configured so that the transmission of optical and electrical signals to the subsequent A/D converter 37 is not impeded when the probe 34 and carrier detection sensor 40 rotate together with the upper base plate 14.

Figures 4A and 4B are explanatory diagrams showing the state when neither the workpiece 21 nor the carrier 20 is present on the lower surface plate 12, and neither the carrier 20 nor the workpiece 21 is detected. Figures 5A and 5B are explanatory diagrams showing the state when the carrier 20 has entered below the high frequency coil 41, and the carrier 20 is detected by the carrier detection sensor 40. Figures 6A and 6B are explanatory diagrams showing the state when the workpiece 21 has entered below the probe 34, and the carrier 20 is not detected, but the workpiece 21 is detected.

7A and 7B are explanatory diagrams showing the detection state of the carrier 20 and the workpiece 21 when the laser light passes over the lines L1, L2, and L3 on the carrier 20.
7A and 7B, the carrier 20 and the workpiece 21 are clearly identified and detected. In particular, the boundary between the carrier 20 and the workpiece 21 is clearly identified.

In the conventional system, when there is only a workpiece thickness measurement section, as shown in Figure 15, if there is a slurry film on the carrier, the workpiece thickness measurement section measures the thickness of the slurry film, making it difficult to clearly distinguish the boundary between the workpiece and the carrier, and making it impossible to accurately grasp the workpiece thickness measurement position, which is one of the reasons why an accurate workpiece thickness distribution (cross-sectional shape) cannot be obtained.

In this regard, in the present embodiment, by providing a carrier detection sensor 40, even if a film of slurry is formed on the carrier 20, it is possible to determine that this is the thickness of the slurry on the carrier 20 by detecting the carrier 20.
In this sense, it is important to grasp the boundary between the carrier 20 and the workpiece 21 .
Also, it is important to grasp the boundary between the carrier 20 and the workpiece 21 for the following reason.

In this embodiment, the cross-sectional shape of the workpiece 21 can be known by measuring the laser light passing across the workpiece 21. This cross-sectional shape can be known by measuring the thickness of the workpiece 21 from when the laser light starts to pass across the workpiece 21 (at the boundary between the carrier 20 and the workpiece 21) to when the laser light finishes passing over the workpiece 21 (at the boundary between the carrier 20 and the workpiece 21). In order to know this cross-sectional shape of the workpiece 21, it is important to grasp the boundary between the carrier 20 and the workpiece 21.
The cross-sectional shape of the workpiece 21 is grasped in relation to position information (XY coordinates or polar coordinates) from the center of the workpiece 21 .

Figures 8 to 11 are simulation diagrams showing the path of laser light passing through 15 workpieces Wf1 to Wf15 at each moment, calculated from the rotational speed of the upper base plate 14 on which the probe 34 is provided, the rotational speed of the sun gear 16, the rotational speed of the internal gear 15, etc.
As shown in FIGS. 8 to 11, the laser light passes over each workpiece 21 evenly.

FIG. 12 is a block diagram of the control unit 50 that controls the entire system.
The control unit 50 includes a thickness calculation unit 38 including a thickness distribution calculation unit 38 a, as well as a position calculation unit 51 and a position correction unit 52 .
The position calculation unit 51 is connected to a memory unit 53 into which data such as the rotational speed of the upper base plate 14, the rotational speed of the sun gear 16, the rotational speed of the internal gear 15, etc. (the settings can be changed by operating an input unit not shown) has been input in advance.
The carrier detection sensor 40 is connected to the position correction unit 52 .

In this embodiment, as shown in Figures 8 to 11, the path of the laser light passing over the workpiece 21 (path of the thickness measurement position) is calculated in the position calculation unit 51 as position information (XY coordinates or polar coordinates) from the center of the workpiece in relation to the time of passage. That is, based on data such as the rotation speed of the upper surface plate 14, the rotation speed of the sun gear 16, and the rotation speed of the internal gear 15 input to the memory unit 53, the position calculation unit 51 calculates the path of the laser light passing over the workpiece 21 as position information (XY coordinates or polar coordinates) from the center of the workpiece in relation to the time of passage using a formula for the relationship between these data.

As shown in FIG. 15, there is a discrepancy between the position information (calculated time) calculated by the above and the actual measured position (actual measured time) due to a time lag when extracting the rotational position information of the sun gear 16 and the internal gear 15, and errors in the rotational position information. For this reason, it is not possible to grasp the exact thickness measurement position (coordinates on the workpiece 21), and as mentioned above, an accurate thickness distribution cannot be obtained.

In this embodiment, the carrier 20 is detected by the carrier detection sensor 40, and the time when the laser light passes through the actual boundary between the carrier 20 and the workpiece 21 is detected (actual measurement time). The position correction unit 52 corrects the deviation between this actual measurement time and the time when the laser light passes through the boundary between the carrier 20 and the workpiece 21 (calculated time) calculated by the position calculation unit 51 so that it becomes zero.

That is, the position correction unit 52 replaces the calculated time when the laser light passes through the boundary between the carrier 20 and the workpiece 21 calculated by the position calculation unit 51 with the above-mentioned actual measurement time. That is, the position information (XY coordinates or polar coordinates) of the workpiece 21 calculated by the position calculation unit 51 is corrected by assuming that the laser light passes through the boundary between the carrier 20 and the workpiece 21 at the above-mentioned actual measurement time.
As a result, the thickness (cross-sectional shape) of the workpiece 21 on the path that the laser light passes over the workpiece 21 is calculated in association with position information (coordinates) relative to the center of the workpiece 21 from the actually measured boundary portion.

Therefore, the cross-sectional shape of the workpiece 21 at the edge portion can be obtained from the edge portion on the path through which the laser light has passed.
The thickness calculation unit 38 , the position calculation unit 51 , and the position correction unit 52 are controlled by an overall control unit 50 .
The control unit 50 controls each unit so that polishing of the workpiece 21 ends when the thickness of the workpiece 21 at the required position (actual position) reaches the set thickness.

Incidentally, it is known that the cross-sectional shape of the workpiece (silicon wafer) 21 being polished changes from moment to moment depending on the stage of polishing.
That is, in the initial stage of polishing, the overall shape of the workpiece 21 is an upwardly convex shape, and a large sagging shape is observed on the outer periphery of the workpiece 21. As polishing progresses, the overall shape of the workpiece 21 approaches flatness, but the sagging shape remains on the outer periphery of the workpiece 21. At this time, the thickness of the workpiece 21 becomes approximately the same as that of the carrier 20. As polishing progresses further, the overall shape of the workpiece 21 becomes approximately flat, and the amount of sagging on the outer periphery of the workpiece 21 decreases. Thereafter, as polishing progresses, the shape of the workpiece 21 gradually becomes concave in the center, and the outer periphery of the workpiece 21 becomes cut up.

For the above reasons, in order to obtain a workpiece 21 with high flatness on the entire surface and outer periphery, it is common to polish the workpiece 21 until the thickness of the workpiece 21 becomes approximately equal to the thickness of the carrier 20, and polishing of the workpiece 21 is completed at this stage.
It is known from experience that when polishing of the workpiece 21 is completed, the portion of the workpiece 21 other than the edge portion has an extremely high degree of flatness (for example, a surface roughness of 10 nm or less).

Even in this case, the outer periphery (edge) of the workpiece 21 is rounded off as described above, and therefore, the outer periphery of the workpiece 21 is usually rendered unusable.
Therefore, it is better to determine when polishing of the workpiece 21 is complete when the thickness of the area (recognized as coordinates) just inside the edge of the workpiece 21 (for example, about 2 mm from the edge) reaches the set thickness.

In this embodiment, as described above, the cross-sectional shape of the workpiece 21 on any path that the laser light has passed through can be obtained. To obtain an accurate cross-sectional shape of the entire workpiece 21, it is ideal to obtain the cross-sectional shape on or near the diameter of the workpiece 21, but the cross-sectional shape on or near the diameter of the workpiece 21 does not necessarily have to be used to determine the end time of polishing.

In other words, as described above, the end of polishing of the workpiece 21 can be determined by ending the polishing of the workpiece when the thickness of the part immediately inside the edge of the workpiece 21 reaches the set thickness. Therefore, at any position where the laser light, which is the measurement light, crosses the workpiece 21 as shown in Figures 8 to 11, it can be determined whether the thickness of the part immediately inside the edge has reached the set thickness. Therefore, the thickness of, for example, 15 workpieces 21 can be determined almost simultaneously.

As described above, in this embodiment, the thickness of the workpiece 21 can be measured in real time while the workpiece 21 is being polished without stopping the revolution of the platen, and the provision of the carrier detection sensor 40 prevents the thickness of the slurry layer 19 from being mistaken for the thickness of the workpiece 21, and by correcting the deviation from the calculated measurement position, the thickness can be measured at the exact position (coordinates) of the workpiece 21.

13A and 13B are explanatory diagrams showing an embodiment in which the carrier detection sensor 40 is provided in a window body 45 provided in the measurement hole 35.
The window body 45 has an upper cap 60 and a lower cap 61 .
The upper cap 60 and the lower cap 61 are formed of an elastic body such as rubber, and are connected by two connecting shafts 64 as is clear in FIG. 13B.
The upper cap 60 is inserted into and fixed to the upper portion of the measurement hole 35 provided in the upper surface plate 14. Similarly, the lower cap 61 is inserted into and fixed to the lower portion of the measurement hole 35 provided in the upper surface plate 14.
A thin transparent plate 65 is fitted into through holes provided at the same vertical position in the upper cap 60 and the lower cap 61. The laser light from the probe 34 is irradiated onto the workpiece 21 through this transparent plate 65, and the reflected light from the workpiece 21 is also guided from this transparent plate 65 to the probe 34.

The carrier detection sensor 40 is fitted in a through hole provided in parallel to the through hole in which the transparent plate 65 of the lower cap 61 is fitted. The lead wires 42 are led out from a through hole provided in the upper cap 60 to the outside of the upper cap 60.
The lower cap 61 is inserted into and fixed at the bottom of the measurement hole 35 so that its lower surface is positioned slightly above the lower surface of the upper surface plate 14. A resin protective plate 66 is adhered and fixed to cover the lower surface of the lower cap 61 to prevent slurry or water from entering the measurement hole 35.

An O-ring 68 is fitted between the outer wall surfaces of the upper cap 60 and the lower cap 61 and the inner wall surface of the measurement hole 35 to provide a liquid-tight seal.
Furthermore, the gaps between the outer wall surface of the upper cap 60 and the inner wall surface of the measurement hole 35 , between the outer wall surface of the lower cap 61 and the inner wall surface of the measurement hole 35 , and between the lead wire 42 and the through hole are filled with sealant to ensure liquid-tightness.
Since the carrier detection sensor 40 in this embodiment is disposed within the window 45 provided in the measurement hole 35 , it is possible to prevent liquids such as slurry or water from entering the measurement hole 35 .

In each of the above embodiments, the carrier detection sensor 40 is disposed in the measurement hole 35 through which the laser light passes, but the probe 34 (measurement hole 35) through which the laser light passes and the carrier detection sensor 40 may be provided at different positions on the upper surface plate 14.
FIG. 14 is an explanatory diagram in which the carrier detection sensor 40 and the probe 34 are set at different positions.

In this embodiment, the five carriers 20 are the same shape and are arranged at equal intervals on the lower surface plate 12, and the positions of the three through holes 22 are also set to be arranged at the same angle relative to the center of the lower surface plate 12. The positions of the probes 34 (i.e., the measurement holes 35) and the carrier detection sensors 40 are both set to be at the same distance from the center position of the upper surface plate 14, which is 4/5 of the radius of the upper surface plate 14, and to be in the same position relative to the corresponding carriers 20.

Under the above conditions, the five carriers 20 revolve and rotate in the same state.
By setting the above conditions, the detection of the carrier 20 by the carrier detection sensor 40 is time-similar to the detection of the carrier 20 at the probe 34 position (thickness detection position), and the thickness of the workpiece 21 can be measured and the presence of the carrier 20 can be detected in the same manner as described above.

The measurement hole 35 in which the probe 34 is located may have a structure in which the carrier detection sensor 40 is removed from Figures 4 to 6 and 13, and the carrier detection sensor 40 may simply have a structure in which a sensor portion or the like is disposed in a hole provided in the base plate.
The reference numeral 35a denotes a slurry supply hole.

Next, Fig. 17 is a block diagram showing an outline of another embodiment of the thickness measuring unit 30 for measuring the thickness of the workpiece 21. Fig. 18 is a signal processing flow diagram showing the thickness measuring process including an optical system. The same reference numerals are used for the same members as those in the above embodiment.
17 and 18 will be described together below.
Reference numeral 70 denotes a known wavelength sweep type laser light source.

A laser beam having a wavelength swept at a constant wavelength sweep speed of 20 to 50 kHz in a certain range including a wavelength of 1260 to 1350 nm is emitted from the laser light source 70. When a silicon wafer (workpiece) is to be measured, the wavelength sweep range is preferably a range of infrared light of 1200 to 1400 nm, which is centered around 1300 nm and easily transmits silicon, that is, the wavelength of the silicon is easily transmitted through the silicon.
The laser light emitted from the laser light source 70 is irradiated onto the portion to be measured of the workpiece 21 via the circulator 32, the optical rotary joint 33, and the probe (lens system) 34.

The laser light reflected from the front surface of the workpiece 21 and the laser light reflected from the rear surface interfere with each other and are observed as interference light having a required phase.
The interference light signal is input to a DAQ (data acquisition device: digitizer) 37 incorporating an A/D converter and an FFT analyzer, and converted into a digital signal (FIG. 4, step 1: S1).
An external trigger signal generated by a laser light source 70 is input to the DAQ 37 .
As the DAQ 37, for example, the ADQ14OCT manufactured by TELEDYNE SP DEVICES can be suitably used.

A known MZI (Mach-Zehnder) interferometer 72 is preferably used as the MZI interferometer 72, for example, the "Thorlabs INT-MZI-1300" manufactured by Thorlabs.
The MZI interferometer 72 takes in a portion (about 5%) of the laser light emitted from the laser light source 70 into the MZI optical system 73, detects the phase difference of the interference light by a differential photodetector 74, and generates a sampling clock signal (step 2: S2).
The MZI optical system 73 and the differential photodetector 74 constitute an MZI interferometer 72 .

FIG. 19 shows the direct interference light waveform from the workpiece, which is simultaneously input into the computer using the internal clock of the A/D board from the trigger signal of the laser light source 70, and the sampling clock signal (interference light waveform: one sweep) input into the DAQ 37 from the MZI interferometer 72.
19, the sampling clock signals (interference optical signals) from the MZI interferometer 72 are spaced at equal intervals. These interference optical signals are also spaced at equal intervals in the optical frequency space (k-space), and their zero crossings are sampled in the DAQ 37 and used as the sampling clock signals (k-clock signals) (Step 3: S3).

In the DAQ 37, the interference light signal from the work is resampled using the k-clock signal as a reference (step 4: S4).
Then, in the FFT processor 76 (FIG. 20) in the DAQ 37, the resampled signal from the work is multiplied by a window function using the k-clock signal as a reference to perform FFT processing (specifically, DFT: Discrete Fourier Transform), and data on the peak value of the resampled interference waveform is obtained (Step 5: S5).

More specifically, within the FFT processor 76, the work interference waveform resampled in k-space is multiplied by a window function (Hanning or Hamming) at 1024, 2048, 4096, or 8192 points to perform FFT (DFT) processing to obtain peak value data of the interference waveform, and this peak value data is output to the calculation unit 38.
For example, when FFT processing is performed at 8192 points, data can be acquired at a high speed of 122 kHz.

The calculation unit 38 calculates the center frequency (peak value) from the FFT-processed data from the DAQ 37, and can measure the thickness of the workpiece 21 by calculating it using a known formula that includes the refractive index of the workpiece 21.

FIG. 20 is a block diagram of the DAQ 37.
The interference light signal from the workpiece 21 is sampled by an A/D converter (step 1: S1).
The sampling clock signal (interference optical signal) from the MZI interferometer 72 is sampled by an A/D converter and used as a k-clock signal (step 3: S3).

Then, the interference light signal from the workpiece is resampled based on the k-clock signal (step 4: S4).
Then, the resampled signal is multiplied by a window function and subjected to FFT processing (specifically, DFT: Discrete Fourier Transform) (step 5: S5).
The output of the FFT (DFT) processing is converted to a power spectrum (PSD) in step 6 (S6), and further subjected to signal averaging in step 7. The averaging is preferably a moving average.

When the light transmittance of the workpiece 21 is low due to the influence of the measurement conditions, the light interference intensity is weak, and the acquired power spectrum has a low peak at the center frequency indicating the thickness of the workpiece, making the peak difficult to distinguish as it blends into the noise. Therefore, in this embodiment, an averaging process (step 7: S7) is performed on the power spectrum to cancel the effects of randomly occurring noise and make the center frequency peak apparent.

Fig. 21 is a graph showing the state of the power spectrum when averaging is not performed (Fig. 21A) and when averaging is performed (Fig. 21B). As can be seen from Fig. 21, when averaging is performed, noise is reduced and the center frequency peak becomes clearer.
This allows the peak of the central frequency to be detected with high accuracy, improving the accuracy of measuring the thickness of the workpiece 21.

The averaged data is output to the calculation unit 38 via the interface.
As described above, the calculation unit 38 calculates the center frequency (peak value) from the FFT-processed data from the DAQ 37, and can measure the thickness of the workpiece 21 by performing calculations using a known equation that includes the refractive index of the workpiece 21 (step 8: S8).

The DAQ 37 or the calculation unit 38 calculates a peak value by taking a simple average of two or more FFT values or a moving average of two or more FFT values. The peak value (measurement value) calculation is repeated at the speed of the sweep frequency of the laser light source 70. This makes it possible to average out the measurement variations and to measure the thickness of the workpiece 21 with high accuracy.
A series of processes in the DAQ 37 is controlled by a control unit 78 separate from the calculation unit 38. The control unit 78 has a built-in FPGA that allows high-speed processing conditions and the like to be rewritten.

In particular, the k-clock signal generated by the DAQ 37 is a clock signal derived from the MZI interferometer 72 and is extremely accurate. The interference light signal from the workpiece 21 is resampled and calibrated based on this k-clock signal, so that even if there is some deviation in the sweep wavelength from the wavelength-swept laser light source 70 , the thickness of the workpiece 21 can be measured with high accuracy.

In the double-sided polishing machine 10, the relative speed between the workpiece 21 and the probe 34 is high, and in one embodiment, the workpiece passes through at a speed of approximately 1400 mm/sec, taking a maximum of 200 msec. In order to measure the shape (cross-sectional shape) of the workpiece, 5 or more measurement points per mm are required.
Therefore, the number of measurement points required per second is 7055 or more.
The above is calculated each time the wafer passes.

When the wavelength sweep of the laser light source 31 is 30 kHz (33.3 μs/sweep), data can be acquired approximately 30,000 times per second, but by suitably using the DAQ37, for example, an ADQ14OCT from TELEDYNE SP DEVICES, the above FFT processed data can be acquired, and if the DAQ37 or the calculation unit 38 averages every four times, 7,500 points can be acquired per second, and the above 7,055 points or more can be acquired, which is sufficient to handle thickness measurement (shape measurement) of the workpiece 21 in a double-sided polisher in which the workpiece 21 rotates at high speed. In this way, in this embodiment, data can be acquired at high speed and at multiple points within the DAQ37 while the workpiece 21 is being polished, so that the workpiece 21 can be polished with high precision.

When the wavelength sweep of the laser light source 70 is 20 kHz (50 μs/sweep), data can be acquired approximately 20,000 times per second. If the DAQ 37 or calculation unit 38 averages every two times, this amounts to 10,000 points per second, which is sufficient for measuring the thickness (shape) of the workpiece 21 in a double-sided polishing machine, in which the workpiece 21 rotates at high speed.
It is also possible to perform window function + FFT processing in the calculation unit 38. However, since the calculation unit 38 is involved in controlling the entire polishing apparatus, the data processing speed is slow and only about 45 points can be obtained per second, making it impossible to obtain a satisfactory number of data points.
In this respect, in the present embodiment, as described above, programming is performed by FPGA within the DAQ 37, and high speed processing can be performed by a separate control unit 78.

In addition, since there are various types of workpieces to be machined, the light transmittance of the workpiece 21 to be measured may also differ depending on the type of workpiece 21 to be machined. Accordingly, the intensity of the interference light signal from the workpiece 21 varies depending on the type of workpiece 21 to be measured. Specifically, when the light transmittance is low, the intensity of the interference light signal tends to be low.
Therefore, it is preferable to provide a function for switching or adjusting the amplification factor of the detector (photodiode) 36 that detects the interference light signal to an appropriate amplification factor according to the type and transmittance of the workpiece 21 to be measured. The adjustment and switching of the amplification factor is preferably performed by a controller (calculation unit) 38.
This enables accurate thickness measurement for various types of workpieces with different light transmittances, and even if the workpiece being measured has a low intensity of the interference light signal, it is possible to accurately measure the thickness without being affected by noise, etc.

FIG. 22 is a block diagram of the entire system showing still another embodiment of the thickness measuring unit 30.
The same members as those in FIG. 3 are designated by the same reference numerals, and the description thereof will be omitted.
In this embodiment, a light source monitoring circuit 80 using a reference wafer (work) is added.
In this light source monitoring circuit 80, the laser light emitted from the laser light source 70 and separated is irradiated onto the reference wafer 86 through the circulator 82 and the probe 84. The circulator 82 and the probe 84 constitute a second optical system. The laser light (interference light) reflected on the front and back surfaces of the reference wafer 86 is detected by the photodiode 88 (second detector) via the probe 84 and the circulator 82, and is further converted into a digital signal by the DAQ 37 and input to the calculation unit 38.

The calculation unit 38 calculates the average value of the voltage output from the photodiode 88 (average value within a set time) from the input digital signal value. The voltage value output from the photodiode 88 fluctuates in proportion to the increase or decrease in the amount of laser light emitted from the laser light source 70. The fluctuation in the voltage value is appropriately monitored, and if the average value of the voltage falls below a predetermined threshold value, it is determined to be abnormal and an alarm is issued.

Alternatively, the light source monitoring circuit 80 can also be used to monitor the sweep wavelength accuracy of the laser light source 70. Since the thickness (and its distribution) of the reference wafer 86 is constant, if there is a deviation in the measured thickness, it can be determined that the sweep wavelength accuracy of the laser light source 70 has fluctuated, and this can serve as a guide for understanding the cause of the fluctuation. The sweep wavelength accuracy can be understood by the average value, PP value, deviation value, etc. of the frequency of the FFT peak value of the acquired interference waveform measured a set number of times (for example, 1000 times).
The timing of monitoring can be set by the program. For example, it can be set to immediately before measuring the thickness of the workpiece.

The MZI (Mach-Zehnder) interferometer 72 can also be used as a light source monitoring circuit.
In this case, the MZI (Mach-Zehnder) interferometer 72 serves as the second detector. The MZI (Mach-Zehnder) interferometer 72 monitors the swept wavelength accuracy, which can be grasped by the average value, PP value, or deviation value of a set number of measurements, for the frequency of the FFT peak value of the interference waveform acquired by the MZI (Mach-Zehnder) interferometer 72. If the change in the swept wavelength accuracy exceeds the allowable value, an alarm is issued.

In each of the above embodiments, a workpiece thickness measuring unit in a double-sided polishing apparatus for a workpiece has been described as an example, but it goes without saying that the present invention can also be applied to a workpiece thickness measuring unit in a single-sided polishing apparatus (not shown) in which a workpiece is held on the underside of a polishing head and the underside of the workpiece is polished between the polishing head and a polishing pad on a base plate.

10 Double-sided polishing machine 12 Lower surface plate 14 Upper surface plate 15 Internal gear 16 Sun gear 17, 18 Polishing pad 19 Slurry layer 20 Carrier 21 Workpiece 22 Through hole 23 Suspension pole 24 Disk 25 Rod 30 Thickness measuring section
31 Laser light source 32 Circulator 33 Rotary joint 34 Probe 35 Measurement hole 35a Slurry supply hole 36 Photodiode 37 DAQ
38 Thickness calculation unit 39 Sensor amplifier 40 Carrier detection sensor 41 High frequency coil (sensor unit)
42 Lead wire 43 Support 50 Control unit 51 Position calculation unit 52 Position correction unit 53 Memory unit 60 Upper cap 61 Lower cap 64 Connecting shaft 65 Transparent plate 66 Protective plate 68 O-ring 70 Laser light source 72 MZI interferometer 73 MZI optical system 74 Differential photodetector 76 FFT processor 78 Control unit 80 Light source monitoring circuit 82 Circulator 84 Probe 86 Reference wafer 88 Photodiode

Claims (14)

A double-sided polishing apparatus comprising: a ring-shaped lower platen having a polishing pad fixed to its upper surface and rotatable about a rotation axis; a ring-shaped upper platen having a polishing pad fixed to its lower surface and movable up and down above the lower platen and rotatable about the rotation axis; a sun gear disposed at the center of the lower platen; an internal gear disposed surrounding the lower platen; and a carrier disposed between the lower platen and the upper platen, having a through hole for holding a workpiece, meshing with the sun gear and the internal gear, and revolving around the sun gear and rotating on its own axis,
a thickness measuring unit that includes an optical system that is provided on either the upper or lower surface plate and rotates together with the surface plate, irradiating the workpiece with a laser beam through a measurement hole provided in the surface plate, and that receives interference light of reflected light or transmitted light from the workpiece to measure the thickness of the workpiece;
a carrier detection sensor capable of detecting the carrier passing directly above or directly below the measurement hole;
a position calculation unit that calculates a path of the laser light that crosses the workpiece held by the carrier from a rotation speed of the platen, a rotation speed of the sun gear, and a rotation speed of the internal gear as position information from a center of the workpiece in association with a passing time;
a position correction unit that corrects position information of the workpiece calculated by the position calculation unit based on a difference between an actual passing time of the laser light passing through the actual boundary between the carrier and the workpiece, which is actually measured from a detection signal from the carrier detection sensor, and a calculated time of the laser light passing through the boundary between the carrier and the workpiece, which is calculated by the position calculation unit;
and a control unit that terminates polishing of the workpiece when the thickness of the workpiece at the required position of the workpiece corrected by the position correction unit reaches a set thickness. The double-sided polishing device according to claim 1, characterized in that the control unit controls the polishing of the workpiece to end when the thickness of the inner side immediately adjacent to the edge portion of the workpiece reaches a set thickness. The double-sided polishing machine according to claim 1 or 2, characterized in that the sensor part of the carrier detection sensor includes a high-frequency coil that induces eddy currents in the carrier. The double-sided polishing machine according to claim 3, characterized in that the sensor part of the high-frequency coil is disposed in the measurement hole provided in the upper platen and adjacent to the lower surface of the upper platen. The double-sided polishing machine according to claim 3 or 4, characterized in that the high-frequency coil and the optical system are set so that the high-frequency coil and the laser light irradiated from the optical system to the workpiece are coaxial. The double-sided polishing machine according to any one of claims 1 to 5, characterized in that the sensor portion of the carrier detection sensor is coated with fluororesin. The double-sided polishing machine according to any one of claims 1 to 6, characterized in that the sensor portion of the carrier detection sensor is electromagnetically shielded. A double-sided polishing machine according to any one of claims 1 to 7, characterized in that a sensor amplifier is provided to amplify the detection signal from the sensor part of the carrier detection sensor, and the sensitivity of the carrier detection is adjusted on the sensor amplifier side, or a threshold value is set to turn the carrier detection on and off. A double-sided polishing machine according to any one of claims 1 to 5, characterized in that the measurement hole is provided with an upper cap fixed to the top of the measurement hole, a lower cap fixed to the bottom of the measurement hole, and a window body fixed coaxially to the upper cap and the lower cap and having a transparent plate through which the laser light passes. The double-sided polishing device according to claim 9, characterized in that a resin protective plate is fixed to cover the lower surface of the lower cap other than the transparent plate. The thickness measuring unit includes:
A light source monitoring circuit using a reference work is provided,
The light source monitoring circuit includes a second optical system that guides and irradiates a laser beam from a wavelength swept laser source onto the reference workpiece;
A second detector that detects an interference light signal of reflected light or transmitted light obtained from the reference work;
11. The double-sided polishing apparatus according to claim 1, further comprising a DAQ for converting the interference light signal detected by the second detector into a digital signal. The double-sided polishing machine according to claim 11, characterized in that the light source monitoring circuit monitors the amount of light of the laser light from the average value of the voltage output from the second detector. The double-sided polishing machine according to claim 11, characterized in that the light source monitoring circuit monitors the sweep wavelength accuracy, which can be grasped by the average value, P-P value, or deviation value of a set number of measurements, for the frequency of the FFT peak value of the interference waveform acquired by the second detector. The thickness measuring unit includes:
A Mach-Zehnder interferometer;
a detector; and
11. The double-sided polishing apparatus according to claim 1, wherein the Mach-Zehnder interferometer also serves as a light source monitoring circuit, and the light source monitoring circuit monitors the sweep wavelength accuracy, which can be grasped by an average value, a PP value, or a deviation value of a frequency of an FFT peak value of the interference waveform acquired by the detector, measured a set number of times.

Images