JP3991598B2 - Wafer polishing method - Google Patents

Description

[0001]
BACKGROUND OF THE INVENTION
The present invention relates to a wafer polishing method for simultaneously polishing both surfaces of a semiconductor wafer using a carrier between upper and lower rotating surface plates.
[0002]
[Prior art]
High integration of semiconductor devices is rapidly progressing, and flatness required for a semiconductor wafer as a material thereof is becoming increasingly severe. Further, from the viewpoint of reducing the manufacturing cost of the semiconductor device, since the diameter of the semiconductor wafer is being increased, it is more difficult to improve the flatness. For this reason, in the semiconductor wafer processing process, a process was proposed in which a grinding process was introduced after etching, and in the mirror polishing process following grinding, a double-side polishing system with processing accuracy superior to conventional single-side polishing. Is attracting attention.
[0003]
As a double-side polishing apparatus for semiconductor wafers, a planetary gear type is usually used. This double-side polishing apparatus includes a pair of upper and lower rotating surface plates, a sun gear provided at the center of rotation between the rotating surface plates, and an annular internal gear provided at the outer peripheral portion between the rotating surface plates. Yes. The upper rotary platen can be raised and lowered, and is driven up and down by a cylinder that also serves as a pressurizing cylinder, and the rotation direction is opposite to the rotation direction of the lower rotary platen. A polishing cloth is affixed to the opposing surfaces of the upper and lower rotating surface plates.
[0004]
In the polishing operation, a plurality of carriers are set around the rotation center on the lower rotating surface plate with the upper rotating surface plate raised, and a semiconductor wafer is set on each carrier. The plurality of carriers, which are planetary gears, mesh with the inner sun gear and the outer internal gear, respectively, and have a wafer accommodation hole called a hole at a position eccentric from the rotation center.
[0005]
When the setting of the carrier and the semiconductor wafer is completed, the upper rotating surface plate is slowly lowered, and a plurality of semiconductor wafers are sandwiched between the upper and lower rotating surface plates with a predetermined pressure. Then, the rotating surface plate, the sun gear, and the internal gear are rotated while supplying the polishing liquid between the rotating surface plates. As a result, the plurality of carriers revolve around the sun gear while rotating between the rotating surface plates rotating in the opposite directions. Thus, both surfaces of the plurality of semiconductor wafers are polished simultaneously.
[0006]
The thickness of the carrier used here is set sufficiently thinner than the target value of the final finished thickness of the wafer so as to ensure that a predetermined pressure is applied to the wafer throughout the polishing process. Specifically, in the case of a 12-inch wafer whose final finished thickness target value is 750 μm , the thickness of the carrier combined therewith is about 650 μm . Further, as the material of the carrier, stainless steel is mainly used in addition to a resin reinforced with glass fiber, for example, epoxy resin, phenol resin, nylon resin.
[0007]
[Problems to be solved by the invention]
In such a double-sided mirror polishing process for semiconductor wafers, management of the polishing amount is regarded as an important technical issue not only for improving flatness but also for removing processing distortion and correcting surface roughness up to the previous process. In the past, the management was performed by managing the polishing time.
[0008]
In other words, the polishing amount in the mirror polishing process varies greatly depending on the polishing conditions, and is greatly influenced by changes in the condition due to the life of the polishing pad. For this reason, before and after mirror polishing for each batch or process, the wafer thickness is measured with an automatic or manual thickness gauge provided outside the apparatus, and the polishing rate is calculated from the polishing amount and polishing time that are the difference between them. In order to ensure a predetermined polishing amount, feedback control for setting the polishing time in the next batch or the next process is performed.
[0009]
And by proposing a process that introduces a grinding process before the mirror polishing process, the thickness variation between the wafers after grinding has been dramatically improved. As a result, the wafer thickness data can be used as the known information for the mirror polishing process. Thus, the thickness measurement before polishing can be omitted.
[0010]
However, unlike the lapping process in which both sides are processed directly between the upper and lower rotating surface plates made of a rigid body and the wafer thickness can be detected by the interval between the surface plates, both surfaces of the wafer are pressed through a polishing cloth made of a viscoelastic material. In the case of the mirror polishing process to be processed, it is impossible to detect the wafer thickness in-process by the surface plate interval, and it is impossible to perform fixed-size polishing. For this reason, offline measurement is still necessary for the thickness measurement after polishing.
[0011]
As a result, enlargement of the apparatus and a decrease in the operation rate become problems. That is, in the offline thickness measurement, a wafer drying process is required before the thickness measurement depending on the measurement model, and the apparatus becomes enlarged. Moreover, the complexity of a process is promoted and the fall of an operation rate is caused.
[0012]
An object of the present invention is to provide a wafer polishing method capable of performing a predetermined amount of polishing simply and accurately without measuring the thickness of the wafer and without being affected by the condition of the polishing cloth. .
[0013]
[Means for Solving the Problems]
In order to achieve the above object, the present inventors paid attention to the thickness of the carrier with respect to the finished thickness of the wafer.
[0014]
That is, conventionally, as described above, the thickness of the carrier combined with the wafer is set sufficiently thinner than the final finished thickness of the wafer so as to ensure that a predetermined pressure is applied to the wafer throughout the polishing process. . Thereby, even when the thickness of the wafer reaches the final finished thickness, the upper and lower polishing cloths are only pressed against the wafer and do not substantially contact the upper and lower surfaces of the carrier. For this reason, a predetermined pressing force from the upper rotating surface plate is applied only to the wafer until the polishing is completed.
[0015]
If polishing continues even after the wafer thickness reaches the final thickness, the wafer becomes thinner. When polishing progresses to approximately the same thickness as the carrier, the applied pressure from the upper rotating surface plate is applied to both the wafer and the carrier and dispersed. In addition, there is always no escape space for the polishing liquid supplied between the rotating surface plates via the upper rotating surface plate, and back pressure is generated in the direction of lifting the upper rotating surface plate. The pressure applied to the wafer is drastically reduced. As a result, polishing is substantially stopped. In addition, the torque of the motor that drives the upper and lower rotating surface plates is abruptly reduced. Further, in this state, although the polishing cloth is in contact with the upper and lower surfaces of the carrier, the damage that the carrier receives from the polishing cloth is surprisingly small due to hydroplane polishing.
[0016]
When the polishing is continued until the thickness of the wafer reaches substantially the same thickness as the carrier, the polishing is substantially stopped, and at this point, the torque of the motors that drive the upper and lower rotating surface plates suddenly increases. Focusing on the point of decrease and further on the fact that the damage to the carrier is minor, we planned to manage the polishing amount and the final finished thickness of the wafer based on the thickness of the carrier, and conducted an experimental study. As a result, if the carrier thickness is selected to be the same as or slightly smaller than the final finished thickness of the wafer, the rotating platen will be driven when the wafer thickness reaches a value slightly smaller than the carrier thickness. It has become clear that the amount of polishing and the thickness of the wafer after polishing can be managed easily and accurately.
[0017]
The wafer polishing method of the present invention has been invented based on such knowledge. By holding a semiconductor wafer to be polished on a carrier and rotating it between upper and lower rotating surface plates, both surfaces of the semiconductor wafer can be simultaneously applied. In the wafer polishing method for polishing,
The thickness before polishing of the carrier is the same as or slightly smaller than the final finished thickness of the semiconductor wafer, and the difference between the thickness before polishing of the carrier and the final finished thickness of the semiconductor wafer is 6 μm or less,
When the polishing of the semiconductor wafer proceeds to substantially the same thickness as the carrier, back pressure is generated in the direction of lifting the upper rotating surface plate, so that the pressure applied to the semiconductor wafer is greatly reduced. The polishing end point is detected from a decrease in driving torque.
Further, according to the present invention, the reduction in the driving torque of the rotary platen determined to be the completion of the polishing may be a reduction rate of 5 to 50% from the maximum value of the driving torque.
Further, the present invention is preferably characterized in that a target value of a final finished thickness of the semiconductor wafer is 750 μm.
In the present invention, the carrier may be made of stainless steel, or FRP in which a reinforcing fiber is combined with any of epoxy resin, phenol resin, and polyimide.
The wafer polishing method of the present invention is a wafer polishing method in which a semiconductor wafer to be polished is held on a carrier and rotated between upper and lower rotating surface plates to simultaneously polish both surfaces of the semiconductor wafer. It is the same as or slightly smaller than the final finished thickness of the semiconductor wafer.
[0018]
By doing this, when the thickness of the wafer becomes a value slightly smaller than the carrier thickness, a rapid decrease in the driving torque of the rotating surface plate starts, and by detecting the decrease, the wafer thickness can be measured, Even if the polishing time is not managed, the polishing end point can be obtained accurately.
[0019]
The difference between the thickness of the carrier and the finished thickness of the semiconductor wafer is preferably 6 μm or less. In a region where this difference exceeds 6 μm, the driving torque does not significantly decrease, and it is difficult to accurately detect the polishing end point. Note that it takes a long time to perform polishing until the finished thickness of the semiconductor wafer becomes less than the thickness of the carrier, and the wear of the carrier becomes significant, which is not preferable.
[0020]
In the present invention, the distance between the polishing cloth and the carrier is smaller than in the conventional case, and the contact pressure between the polishing cloth and the carrier is also increased, so that the wear of the carrier easily proceeds, and if the thickness is reduced, the final finished thickness of the wafer is also worn. The amount is reduced, which affects the sizing accuracy. Carrier wear occurs due to friction with the polishing cloth affixed to the surface plate, so the carrier material is a material that has high wear resistance and a low friction coefficient with the polishing cloth, and an alkaline pH 8-12. Those having high chemical resistance in the polishing liquid are preferred. Examples of the carrier material satisfying such conditions include stainless steel, FRP in which a reinforcing fiber such as glass fiber, carbon fiber, or aramid fiber is combined with a resin such as epoxy resin, phenol resin, or polyimide.
[0021]
DETAILED DESCRIPTION OF THE INVENTION
Embodiments of the present invention will be described below with reference to the drawings. FIG. 1 is a configuration diagram of a polishing apparatus for explaining an embodiment of the present invention, and FIG. 2 is a view taken along line AA in FIG.
[0022]
The polishing apparatus includes an annular lower surface plate 1 that is horizontally supported, an annular upper surface plate 2 that faces the lower surface plate 1 from above, a sun gear 3 that is disposed inside the annular lower surface plate 1, and a lower surface plate. 1 and a ring-shaped internal gear 4 disposed on the outer side of 1.
[0023]
The lower surface plate 1 is rotationally driven by a motor 11. The upper surface plate 2 is suspended from the cylinder 5 via a joint 6 and is rotationally driven in the reverse direction by a motor different from the motor 11 that drives the lower surface plate 1. A polishing liquid supply system including a tank 7 for supplying the polishing liquid to the lower surface plate 1 is also provided. The sun gear 3 and the internal gear 4 are also rotationally driven independently by a motor 12 different from the motor that drives the surface plate.
[0024]
On the opposing surfaces of the lower surface plate 1 and the upper surface plate 2, a polishing cloth made by impregnating a urethane resin into a nonwoven fabric or a polishing cloth made of foamed urethane or the like is affixed.
[0025]
A plurality of carriers 8 are set on the lower surface plate 1 so as to surround the sun gear 3. Each set carrier 8 meshes with the inner sun gear 3 and the outer internal gear 4, respectively. Each carrier 8 is provided with an eccentric hole 9 for accommodating the semiconductor wafer 10. The thickness of each carrier 8 is set to be the same as or slightly smaller than the target value of the final finished thickness of the wafer 10.
[0026]
In order to polish the wafer 10, a plurality of carriers 8 are set on the lower surface plate 1 with the upper surface plate 2 raised, and the wafers 10 are set in the holes 9 of each carrier 8. The upper surface plate 2 is lowered and a predetermined pressure is applied to each wafer 10. In this state, the lower surface plate 1, the upper surface plate 2, the sun gear 3, and the internal gear 4 are rotated at a predetermined speed in a predetermined direction while supplying polishing liquid between the lower surface plate 1 and the upper surface plate 2. .
[0027]
As a result, a so-called planetary motion in which the plurality of carriers 8 revolve around the sun gear 3 while rotating between the lower surface plate 1 and the upper surface plate 2 is performed. The wafer 10 held by each carrier 8 is in sliding contact with the upper and lower polishing cloths in the polishing liquid, and both upper and lower surfaces are polished simultaneously. The polishing conditions are set so that both surfaces of the wafer 10 are uniformly polished.
[0028]
During polishing, the torque of the motor 11 that drives the lower surface plate 1 or the torque of the motor that drives the upper surface plate 2 is monitored. Then, when the torque decreases from a stable value by a preset ratio, for example, 10%, the upper surface plate 2 is raised to finish the polishing. As a result, the final finished thickness of the wafer 10 is managed with high accuracy and stability to be slightly less than or equal to the thickness of the carrier before polishing.
[0029]
More specifically, here, the thickness of each carrier 8 is set to be the same as or slightly smaller than the final finished thickness of the wafer 10. For this reason, the thickness of the wafer 10 before polishing is sufficiently larger than the thickness of the carrier 8. Therefore, at this stage, the upper and lower polishing cloths substantially contact only the upper and lower surfaces of the wafer 10 and do not press the upper and lower surfaces of the carrier 8.
[0030]
In the initial stage of polishing, the wafer 10 that is sufficiently thicker than the carrier 8 is intensively pressed, and the variation in the thickness of the plurality of wafers 10 that have been charged is eliminated. During this time, the torque of the motor 11 that drives the lower surface plate 1 or the torque of the motor that drives the upper surface plate 2 suddenly increases, and when the variation in the thickness of the plurality of wafers 10 is resolved, the peak is reached and the stable region is reached. enter.
[0031]
When the polishing further proceeds and the thickness of the wafer 10 becomes slightly larger than the thickness of the carrier 8, the pressure applied from the upper surface plate 2 is applied to both the wafer 10 and the carrier 8 and dispersed as described above. . In addition, since there is no escape space for the polishing liquid supplied between the surface plates 1 and 2 via the upper surface plate 2 and back pressure is generated in the direction of lifting the upper surface plate 2, hydroplane polishing (slip The pressure applied to the wafer 10 is drastically reduced. As a result, the polishing is substantially stopped, and the torque of the motor 11 that drives the lower surface plate 1 and the torque of the motor that drives the upper surface plate 2 are rapidly reduced.
[0032]
Therefore, by detecting the time when these torques are reduced by a preset ratio from the maximum value, for example, 10%, and finishing the polishing, the final finished thickness of the wafer 10 is equal to or slightly less than the thickness of the carrier 8. Matches a small value with high accuracy. Further, the work of measuring the thickness of the wafer 10 before and after polishing and the work of calculating the polishing time from the polishing speed and managing this time are not required.
[0033]
The rate of decrease from the maximum value of the torque that is judged as the end of polishing is preferably 5 to 50%, particularly preferably 5 to 30%. If it is less than 5%, detection itself is difficult, and if it exceeds 50%, the polishing rate is slow, useless polishing time is generated, and carrier wear increases, which is not preferable.
[0034]
【Example】
Next, the implementation results of the present invention will be described in comparison with a conventional example. Common polishing conditions are as follows.
[0035]
Wafer used: 12-inch silicon wafer polisher: Double-side polisher AC2000P made by Peter Wolters
Abrasive Cloth: Rodel Nita Foam Urethane Abrasive MHN15A
Polishing liquid: Nalco 2350 20 times diluted polishing pressure: 200 g / cm 2
Carrier used: Stainless steel [0036]
Carriers having thicknesses of 745 μm, 740 μm, and 720 μm were used. FIG. 3 shows the change over time of the motor torque value of the upper surface plate in each polishing. Polishing was completed when each torque value decreased by 10% from the maximum value. When the first carrier having a thickness of 745 μm was used, the torque value suddenly decreased 40 minutes after the start of polishing, and the final finished thickness when polishing was stopped at 10% reduction of the maximum value was controlled to 751 μm. .
[0037]
On the other hand, when the second carrier having a thickness of 740 μm was used, the torque value decreased rapidly about 60 minutes after the start of polishing. When the polishing was stopped at a reduction of 10% of the maximum value, the final finished thickness was 744 μm. When the third carrier having a thickness of 720 μm was used, the torque value suddenly decreased 80 minutes after the start of polishing. The final finished thickness was 726 μm when polishing was stopped with a reduction of 10% of the maximum value.
[0038]
As an example of the present invention, a carrier having a thickness of 744 μm, which is 6 μm thinner than the final finished thickness (750 μm) of the wafer, is used, and the polishing is finished when the motor torque value of the upper surface plate decreases by 10% from the maximum value. The polishing method was carried out continuously for 10 cycles. Further, as a conventional example, a carrier that is 100 μm thinner than the final finished thickness (750 μm) is used, the wafer thickness is measured for each cycle, the polishing rate is calculated, and the polishing time for the next cycle is calculated from the polishing rate. A time-controlled polishing method was performed 10 times. Further, as a reference example, continuous polishing was performed with a constant polishing time in order to reveal the instability of the polishing rate in the initial state of the polishing cloth.
[0039]
FIG. 4 shows the final finished thickness in each case. At the initial use of the polishing cloth, the polishing rate becomes unstable. In general, since the polishing agent is not sufficiently impregnated into the polishing cloth at the initial stage, the polishing is not promoted and the polishing rate is low. As the cycle is repeated, the polishing rate gradually increases. In the case of the conventional polishing time management type, the polishing time is set from the polishing rate of the previous cycle. In the first cycle of the polishing cloth and the first cycle after the apparatus is stopped, the polishing time is set with reference to the past actual values, so that the accuracy of the finished thickness tends to become unstable. However, in the case of the method of the present invention, high accuracy can be stably obtained by the torque management based on the carrier thickness.
[0040]
Together with the final finished thickness, the in-plane distribution of the thickness was also evaluated. The results are shown in FIG. In the case of the conventional method of polishing time management type, surface sagging is likely to occur on the outer peripheral portion of the wafer after polishing, and the TTV average value was slightly over 1 μm. Improved. The reason why the surface sagging of the wafer outer peripheral portion is suppressed by the method of the present invention can be considered that the load applied to the outer peripheral portion of the wafer is alleviated by the carrier when the polishing of the wafer proceeds to near the thickness of the carrier.
[0041]
FIG. 6 shows a lowering behavior of the upper surface plate motor torque and an influence of the lowering of the torque on the wafer thickness and the carrier thickness.
[0042]
As can be seen from this figure, the torque reduction phenomenon of the motor that drives the upper platen is remarkable at the start of the reduction, and the time from the start of the reduction elapses and the reduction speed becomes slow. As the motor torque decreases, the finished thickness of the wafer approaches the thickness of the carrier. By stopping polishing when the motor torque reduction rate is in the range of 5 to 30%, the finished thickness of the wafer is +6 μm with respect to the carrier thickness. It is managed as follows. Further, within this range, the wear of the carrier does not substantially occur.
[0043]
More specifically, if the reduction rate of the motor torque for stopping polishing is 10%, the finished thickness of the wafer becomes +5 to 6 μm of the carrier thickness. That is, a carrier 5 to 6 μm thinner than the finished thickness of the wafer is used here. If the reduction rate of motor torque for stopping polishing is 20%, the finished thickness of the wafer is +4 to 5 μm of the carrier thickness. If the reduction rate of the motor torque for stopping polishing is 30%, the finished thickness of the wafer is +3 to 4 μm of the carrier thickness. If the reduction rate of the motor torque at which polishing is stopped is 40%, the finished thickness of the wafer is +3 μm of the carrier thickness, but the carrier is slightly worn. If the reduction rate of the motor torque for stopping the polishing is 50%, the finished thickness of the wafer matches the carrier thickness, but the wear of the carrier increases and the polishing time also increases.
[0044]
In the above embodiment, the change in the torque value for determining the polishing end time is detected by the torque value of the motor that drives the upper platen, but may be detected by the torque value of the motor that drives the lower platen. In short, the polishing end time may be determined from the change in the torque value of the motor involved in the rotational drive of at least one of the upper and lower rotary surface plates.
[0045]
In the above embodiment, the carrier is caused to perform a planetary motion combining a revolving motion and a rotational motion. However, the carrier may be rotated at a fixed position.
[0046]
【The invention's effect】
As described above, according to the wafer polishing method of the present invention, complicated operations such as measurement of the thickness of the wafer after polishing and setting of a polishing time based on the result are not required. In addition, despite the simple work, the finished thickness of the wafer is not affected by the condition of the polishing cloth. For this reason, high-precision sizing is easily and stably performed.
[Brief description of the drawings]
FIG. 1 is a configuration diagram of a polishing apparatus for explaining an embodiment of the present invention.
FIG. 2 is a view taken along the line AA in FIG.
FIG. 3 is a graph showing a relationship between torque change of a surface plate drive motor and carrier thickness.
FIG. 4 is a graph showing the finished thickness of a wafer for a conventional example, a reference example, and an example of the present invention.
FIG. 5 is a graph showing in-plane variations in wafer thickness for a conventional example and an example of the present invention.
FIG. 6 is a graph showing motor torque reduction behavior and the effect of the reduction on wafer thickness and carrier thickness.
[Explanation of symbols]
1 Lower surface plate 2 Upper surface plate 3 Sun gear 4 Internal gear 8 Carrier 10 Wafers 11 and 12 Motor

Claims (4)

In a wafer polishing method for simultaneously polishing both surfaces of the semiconductor wafer by holding the semiconductor wafer to be polished on a carrier and rotating between the upper and lower rotating surface plates,
The thickness before polishing of the carrier is the same as or slightly smaller than the final finished thickness of the semiconductor wafer, and the difference between the thickness before polishing of the carrier and the final finished thickness of the semiconductor wafer is 6 μm or less,
When the polishing of the semiconductor wafer proceeds to substantially the same thickness as the carrier, back pressure is generated in the direction of lifting the upper rotating surface plate, so that the pressure applied to the semiconductor wafer is greatly reduced. A wafer polishing method, wherein a polishing end point is detected from a decrease in driving torque . 2. The wafer polishing method according to claim 1 , wherein a decrease in the driving torque of the rotating surface plate determined to be the end of the polishing is 5 to 50% from a maximum value of the driving torque . The wafer polishing method according to claim 1, wherein a target value of a final finished thickness of the semiconductor wafer is 750 μm . 3. The wafer polishing method according to claim 1 , wherein the carrier is made of FRP in which a reinforcing fiber is combined with any of stainless steel, epoxy resin, phenol resin, or polyimide .

Images