JPH1058440A - Manufacture of semiconductor wafer - Google Patents

Abstract

PROBLEM TO BE SOLVED: To suppress variation of the thicknesses of wafers after the division caused by variation of the thicknesses of the wafers under the specified limit when the semiconductor wafers forming the impurity diffused layers formed on both surfaces are divided into two parts in parallel with the surface with a multiple-wire saw. SOLUTION: Wafers 2 are aligned so that the respective surfaces are in contact. A spacer 4, whose size is adjusted so that the accumulated error of the thicknesses at every specified number is offset, is arranged between the wafers. Outer peripheral parts of the wafers 2 and the spacers 4 are stuck to slice stage 3. The slice stage 3 is fixed to a block 5 and attached to a multiple-wire saw. The position of the block 5 is adjusted so that the center of the thickness of the arbitrary wafer 2a agrees with the central line of the wire 6, and the wafers 2 are sliced. The wafers 2 are divided into the sets of every specified number. The respective sets are separated by the specified interval and attached to the multiple-wire saw. The interval can be adjusted so that the center of the thickness of the arbitrary wafer agrees with the central line of the wire 6 for every set.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a method for manufacturing a semiconductor wafer, and more particularly to a method for dividing a semiconductor wafer having an impurity diffusion layer formed on both surfaces into two parts.

[0002]

2. Description of the Related Art Power transistors, power MOS F
In the manufacture of so-called power devices such as ET, a wafer in which a high-concentration dopant is diffused on one side of a silicon wafer is used. The diffusion wafer (hereinafter referred to as “wafer”) is prepared by chamfering a silicon wafer (hereinafter referred to as “material wafer”) sliced to a predetermined thickness, performing lapping and etching, performing pre-diffusion in an impurity atmosphere, and further performing drive diffusion. A high concentration diffusion layer having a predetermined thickness is formed on both surfaces of the material wafer. Next, the surfaces of these wafers are coated with wax and adhered to each other to form an ingot, and the wafer is sliced in parallel with each surface using an inner peripheral blade, a multi-wire saw or the like, and divided into two. In the case of using a multi-wire saw, as shown in FIG. 6, a part of the outer periphery of the wafer 2 is attached to the slice table 3, and the slice table 3 is fixed to the block 5 and attached to the multi-wire saw. Then, the center line of the thickness of the wafer 2 a positioned substantially at the center of the large number of wafers 2 adhered in an ingot shape is made coincident with the center line of the wire 6 and then sliced. Thus, the wafer 2 is divided into two wafers. When a multi-wire saw is used, up to about 270 diffusion wafers can be divided at the same time, so that the productivity is better than when an inner peripheral blade is used.

After the wafer is divided into two parts by the above method, the wax is peeled off to obtain a wafer having a high concentration diffusion layer on one side. After performing surface grinding and chamfering on these wafers, polishing for mirror finishing is performed to complete a diffusion wafer.

[0004]

However, since the thicknesses of the material wafers vary within the allowable limits even though they are within the allowable limit, when a wafer that has been diffused is sliced with a multi-wire saw, a large number of wafers stuck in an ingot shape can be obtained. The division position shifts from the center of the thickness of the wafer as it goes from the center in the longitudinal direction to both ends. The greater the number of wafers to be sliced at a time, the greater the variation in the thickness of the wafer after division into two. A wafer whose thickness has become thinner than the limit becomes insufficient in the finishing allowance in the surface grinding and mirror finishing processes, and becomes a defective product. In order to prevent such defects from occurring, it is necessary to increase the thickness of the material wafer, which increases the material cost. In addition, when the material wafer is thickened, defective products due to insufficient finishing allowance after the division into two do not occur, but wafers with an abnormally large finishing allowance are generated, so that the grinding cost in the surface grinding and mirror finishing processes increases. .

SUMMARY OF THE INVENTION The present invention has been made in view of the above-mentioned conventional problems. In a process of dividing a wafer subjected to diffusion processing into two at once by a multi-wire saw, a variation in thickness of the divided wafer is reduced to a certain limit or less. It is an object of the present invention to provide a method of manufacturing a semiconductor wafer that can reduce material costs and grinding costs.

[0006]

In order to achieve the above object, a method of manufacturing a semiconductor wafer according to the present invention is directed to a method of dividing a semiconductor wafer having an impurity diffusion layer formed on both surfaces into two parts by slicing the semiconductor wafer in parallel with the surface. In the process, a large number of wafers are arranged in contact with each other, and sliced collectively by a wire saw.

In slicing the diffusion wafer having the above structure, a spacer for adjusting an error in the thickness of the wafer is arranged between a plurality of diffusion wafers.

Further, in slicing a diffusion wafer,
A predetermined number of diffusion wafers may be brought into contact with each other to form a set, and the sets may be arranged at predetermined intervals.

[0009]

According to the above construction, when a semiconductor wafer having impurity diffusion layers on both surfaces is sliced at once by a wire saw, the surfaces of the wafers are arranged in contact with each other. There is no need to bond the surfaces of the wafers as in the conventional case.

When a large number of wafers to be sliced are arranged in contact with each other, errors in the thicknesses of the wafers are accumulated, and the deviation of the division position gradually increases. Therefore, by arranging spacers for each arbitrary number, and adjusting the thickness of the spacers so as not to affect the influence of the accumulated error of the thickness of the wafer on the number of wafers to other wafers, the thickness of the wafer after the division into two is obtained. The variation of the overall becomes small. However, in the case of this method, it is necessary to prepare spacers of various thicknesses or to manufacture them each time.

As another means for reducing the variation in the thickness of the wafer after the division into two, there is a method in which wafers to be sliced are grouped by a predetermined number, and the groups are arranged at predetermined intervals. If the position of each set is adjusted so that the reference position of each set, for example, the center of the thickness of the wafer positioned at the center of each set matches the center line of the wire 6 wound on the work roller of the wire saw, the position of each set is adjusted to 2 The variation in the thickness of the wafer after the division is further reduced.

Next, an embodiment of a method of manufacturing a semiconductor wafer according to the present invention will be described with reference to the drawings. FIG. 1 is a side view of a wafer sticking jig, FIG. 2 is a front view of a wafer stuck on a slice table, and FIG. 3 is a schematic view showing a slicing method using a multi-wire saw, each showing a first embodiment.

The procedure for slicing a wafer using a multi-wire saw is as follows. (1) Divide the wafers into a plurality of sets, for example, 25 wafers as one set. If one batch consists of 250 sheets, 10
Divided into pairs. (2) The total thickness Tt of the wafer is measured for each set, and the difference δ from the reference value T0 is determined. Set the reference thickness of the wafer to t0
, T1, t2, t3,...
.., T25, Tt = t1 + t2 + t3 +... + T25 T0 = t0.times.25.delta. = Tt-T0 (3) When slicing a wafer with a multi-wire saw, the thickness of the spacer sandwiched between each pair is set as follows. decide. The spacer is used to prevent the difference δ between the actual measured value Tt of the total thickness of one set of wafers and the reference value T0, that is, the influence of a thickness error in one set of wafers, from the next set onward. This is a disk having the same diameter as the wafer. The thickness of the spacer is determined based on the above-mentioned δ and the wire pitch of the multi-wire saw. If δ is a positive value, the thickness of the spacer is reduced by that amount, and if δ is a negative value, the thickness of the spacer is increased by that amount. In this way, spacers of various thicknesses corresponding to the positive and negative of δ are manufactured in advance, and a spacer having a suitable thickness is selected from these and used. (4) On the substrate 1a of the wafer bonding jig 1 shown in FIG. 1, a slice table 3 having an arc-shaped concave portion closely contacting the outer periphery of the wafer 2 is placed. Then, the outer peripheral surfaces of the respective wafers 2 and the spacers 4 are attached to the concave portions of the slice table 3 with wax. At this time, in order to bring the respective surfaces of the wafer 2 and the spacer 4 into contact with each other, the clamp 1b of the wafer bonding jig 1 is tightened to hold the wafer 2 and the spacer 4 between the vertical plate 1c and the holding plate 1d. (5) As shown in FIG. 2, the slice table 3 to which the wafer 2 and the spacer are adhered is fixed to the block 5,
To the multi-wire saw. (6) The position of the block is adjusted so that one wire 6, which is wound on the work roller of the wire saw, is located at a position where the wafer incorporated in the center position of an arbitrary set is exactly divided into two. For example, as shown in FIG.
The block 5 is moved so that the position of the thickness of the third wafer 2 a is １ ／ and the center line of the thirteenth wire 6 is coincident.
Adjust the position of. The position adjustment between the wafer 2 and the wire 6 is performed using a microscope. (7) After the above procedure, all wafers are sliced simultaneously. (8) After slicing is completed, the slice table 3 is removed from the block 5, and the wax is melted to separate the wafer 2 from the slice table 3. Thereby, 5 having a diffusion layer on one side is obtained.
00 wafers are obtained. These wafers are subjected to surface grinding, chamfering, and mirror finishing as in the conventional method.

The thickness variation at the center of the wafer divided into two using the above method was 62 μm. The variation in thickness at the center of the wafer when the conventional method is used is 140 μm, which means that the variation in thickness has been reduced to 44%.

FIGS. 4 and 5 show a second embodiment of the present invention.
FIG. 4 is a side view of a wafer bonding jig, and FIG. 5 is a schematic view showing a slicing method using a multi-wire saw. The procedure for slicing a diffusion wafer is as follows. (1) Divide the wafers into a plurality of sets, for example, 25 wafers as one set. If one batch consists of 250 sheets, 10
Divided into pairs. (2) On the substrate 1a of the wafer bonding jig 1 shown in FIG. 4, a slice table 7 having an arc-shaped concave portion closely contacting the outer periphery of the wafer 2 is placed. The slicing table 7 is manufactured to have a length capable of adhering only 25 wafers. Next, the outer peripheral surface of one set of 25 wafers 2 is attached to the concave portion of the slice table 7 with wax. At this time, in order to bring the respective surfaces of the set of wafers 2 into contact with each other, the clamp 1b of the wafer bonding jig 1 is tightened, and the vertical plate 1c and the holding plate 1
d, a set of wafers 2 is held. The pasting operation is performed for all the sets. In order to prevent damage to the wafer surface, dummy wafers may be attached to both ends of each set of wafers. (3) After fixing a slice table to which a set of 25 wafers are adhered to independent blocks, each block is attached to a multi-wire saw in series. At this time, a predetermined interval is provided between each set of wafers. (4) As shown in FIG. 5, the position of each block 8 is adjusted such that the center line of the wire 6 comes to a position where the wafer 2a assembled at the center position of each set is accurately divided into two. The position adjustment between the wafer 2 and the wire 6 is performed using a microscope. (5) After the above procedure, all wafers are sliced simultaneously. (6) After slicing is completed, the slice table 7 is removed from the block 8, and the wax is melted to separate the wafer 2 from the slice table 7. Thereby, 5 having a diffusion layer on one side is obtained.
00 wafers are obtained. These wafers are subjected to surface grinding, chamfering, and mirror finishing as in the conventional method.

The thickness variation at the center of the wafer divided into two using the above method was 53 μm. When the conventional method is used, the thickness variation at the central portion of the wafer is 140 μm, which means that the thickness variation has been reduced to 38%.

In the case of the second embodiment, it is not necessary to select or pinch the spacer, but it is necessary to adjust the position of the wafer with respect to the wire 6 (adjustment of the block fixing position) for each set. The variation in the thickness of the divided wafer is smaller than that in the first embodiment.

[0018]

As described above, according to the present invention, when a semiconductor wafer having a diffusion layer formed on both surfaces is divided into two at a time, the wafer is divided into a plurality of sets, and spacers are arranged or spaced in each set. In this case, the deviation of the dividing position due to the variation in the thickness of the wafer is minimized, so that the variation in the thickness after the division is significantly reduced. Therefore, even if the thickness of the material wafer is made thinner than before, defective products due to insufficient surface grinding and mirror finishing allowance do not occur, and material cost and grinding cost are reduced.

[Brief description of the drawings]

FIG. 1 is an explanatory diagram of a sticking operation according to a first embodiment.

FIG. 2 is a front view of a wafer attached to a slice table.

FIG. 3 is a schematic diagram showing a slicing method according to the first embodiment.

FIG. 4 is an explanatory diagram of a sticking operation according to a second embodiment.

FIG. 5 is a schematic diagram showing a slicing method according to a second embodiment.

FIG. 6 is a schematic view showing a slicing method according to a conventional technique.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 1 Wafer bonding jig 2, 2a Wafer 3, 7 Slice table 4 Spacer 5, 8 Block 6 Wire

Claims (3)

[Claims] 1. A diffusion wafer for slicing a semiconductor wafer having an impurity diffusion layer formed on both sides thereof in parallel with the surface.
A method for manufacturing a semiconductor wafer, comprising: in a dividing step, arranging a plurality of wafers in contact with each other and slicing them collectively by a wire saw. 2. The method of manufacturing a semiconductor wafer according to claim 1, wherein a spacer for adjusting a thickness error of the wafer is arranged between a plurality of diffusion wafers. 3. The method for manufacturing a semiconductor wafer according to claim 1, wherein a predetermined number of diffusion wafers are brought into contact with each other to form a set, and the sets are arranged at predetermined intervals.