JPS6159771A - Manufacture of semiconductor device - Google Patents

Abstract

PURPOSE:To attach two sheets of substrates accurately without damage by forming a spacer in such a structure that a negative resist layer is provided between the lowest and highest positive resists to the periphery of the one substrate of them. CONSTITUTION:A positive photo resist film 29A and a negative photo resist film 29B are formed to the entire part of a p type silicon substrate 21 allowing formation of n<+> type region 23 to which charges are injected. Thereafter, patterning is carried out by exposing and developing the peripheral region of film 29B having the predetermined width. Next, after forming a positive photo resist film 29C to the entire part, a spacer is formed by patterning the films 29C and 29A. A substrate 21 having the photoelectric conversion part is arranged on the fixing base, a pusher of flat plate is pressurizingly placed in contact with the rear surface of substrate 31, coupling the electrode 27 and bump 34.

Description

DETAILED DESCRIPTION OF THE INVENTION [Field of Industrial Application] The present invention relates to a method suitable for manufacturing a semiconductor device called a hybrid optical sensor.

(Prior Art) FIG. 7 is an exploded side view of the main parts of a conventional hybrid optical sensor.

In the figure, 1 and 2 are photoelectric conversion parts, 3 is a p-type indium antimony (InSb) substrate, 4 is an insulating film made of silicon dioxide (SiOz), 5 and 6 are n+ type regions 7 and 8 are indium bumps consisting of 11 and 12
1 is a charge injection treated portion, 13 is a silicon (Si) substrate, and 1
4 is an insulating film made of silicon dioxide, 15 and 16 are n
+ type regions 17 and 18 respectively represent bumps made of indium.

In FIG. 7, the photoelectric conversion semiconductor device having the photoelectric conversion parts 1 and 2, etc. and the charge injection semiconductor device having the charge injection processing parts 11 and 12, etc. are shown separated, but in reality, the indium bumps 7 and 8.17 consisting of
By applying mechanical pressure 81 to and 18 and bonding the corresponding parts, both the photoelectric conversion semiconductor device and the charge injection semiconductor device are used.

The operation of this semiconductor device is well known, but to briefly explain it, when light enters from the back surface of the p-type indium antimony substrate 3, a pn junction formed by the n+ type regions 5 and 6 is formed. A charge is generated, and the charge is injected into the n+ type regions 15 and 16 through the indium bumps 7 and 8, 17 and 18, and the injected charge is transferred to a charge coupled device, for example.
vice:ccD) or the like for detection.

[Problem that the invention seeks to solve]

As described above, when a photoelectric conversion semiconductor device and a charge injection semiconductor device are bonded together, various problems occur if the surfaces of the semiconductor devices are inclined.

FIG. 8 is a side view of essential parts of a photoelectric conversion device and a charge injection semiconductor device for explaining such a case.

In the figure, 50 indicates a photoelectric conversion semiconductor device, 50' indicates a charge injection semiconductor device, and 51 indicates a hump made of indium.

As shown, the surfaces of the photoelectric conversion semiconductor device 5° and the charge injection semiconductor device 50' are inclined.

Generally, in this type of semiconductor device, the back surface of the substrate, especially on the photoelectric conversion device side, is polished to make it thinner overall. The yield, which is maintained constant over time, is extremely low.

As described above, if it is not possible to polish the back side of the Fi-base door to make the thickness the same, the result will be that the front surface is slanted.

When these two parts are bonded together in such a state, the bumps 51 located on the left side in the figure come into contact with each other, but the bumps 51 located on the right side cannot come into contact with each other.

Furthermore, if the bumps 51 on the right side are forcibly pasted together so that they are in contact with each other, the bumps 51 on the left side will be crushed, resulting in the disadvantage that adjacent bumps 51 will come into contact with each other.

According to the present invention, regardless of whether or not the surfaces of the substrate including the photoelectric conversion portion and the substrate including the charge injection processing portion are inclined, the substrates can be bonded together reliably and without causing damage. Do it like this.

[Means for solving problems]

In the method for manufacturing a semiconductor device according to the present invention, when a substrate having a photoelectric conversion portion and a substrate having a charge injection processing portion having a function of injecting and processing charges generated in the photoelectric conversion portion are combined, at least one of the substrates is bonded to the substrate having a photoelectric conversion portion. A spacer is formed on the periphery of the substrate, the bottom layer and the top layer are positive resist films, and at least one negative resist film is interposed between them, and then a spacer is formed on the surface of the substrate. This method includes the steps of applying pressure to bumps made of soft metal, thereby bonding the substrates together, and then removing the spacer.

[Effect]

When the above-mentioned method is adopted, even if the surfaces of the substrates to be bonded are inclined, when pressure is applied between the substrates, the pressure applied to each bump will be approximately equal due to the effect of the spacer.
Accidents such as strong pressure being applied only to a specific bump and causing it to collapse, or pressure not being applied to a bump, etc., are eliminated.

〔Example〕

1 to 6 are cross-sectional side views of essential parts of a semiconductor device at key points in the process for explaining one embodiment of the present invention, and the following description will be made with reference to these figures.

Refer to FIG. 1(a) First, a normal charge injection semiconductor device as shown in the figure is prepared. Since this is achieved by a technique that has been widely used in the past when manufacturing this type of semiconductor device, a description of the manufacturing process will be omitted.

In the figure, 20A is a pad portion, 20B is a charge injection portion, 21 is a p-type silicon substrate, 22 is an insulating film made of silicon dioxide, and 23 is an n-type region 2 into which charges are injected.
4 is an electrode made of aluminum (A1), 25 is a wiring made of aluminum, 26 is a passivation film made of silicon dioxide, 27 is an electrode made of aluminum,
Reference numeral 28 indicates electrodes made of aluminum. Although not shown in the drawings, in addition to the above-mentioned structure, the substrate 21 is provided with a charge processing part, such as an incline type COD, which transfers and detects the charge injected into the n+ type region 23. These are defined as substrates having a charge injection treated portion.

Refer to FIG. 2 (bl) A positive photoresist film 29A is formed on the entire surface to a thickness of about 2 [μm], and a negative photoresist film 29B is formed to a thickness of about 8 [μm].

To carry out the present invention, a spacer made of a photoresist film with a thickness of about 10 [μm] is required, but it is almost impossible to form such a thick photoresist film at once. Have difficulty.

Therefore, it is necessary to stack a photoresist film, but in general, it is not possible to use a positive photoresist because it can be easily dissolved and removed using, for example, acetone. Preferably, photo
When increasing the thickness by laminating resist films, negative photo
It is effective for patterning when a resist film is used in combination.

(C) In order to pattern the negative photoresist film 29B, the portion indicated by the symbol DRI, that is, the negative photoresist It! A peripheral area of a predetermined width at 29B is exposed.

Refer to Figure 3 (d+ After development, the negative photoresist film 29
B is patterned.

(Ill) A positive photoresist film 29C is formed on the entire surface to a thickness of about 1 [μm].

(fl In order to pattern the positive photoresist film 29G and the positive photoresist film 29A, a portion indicated by the symbol DR27 is exposed to light.

See Figure 4 ((a) After development, a positive photoresist film 29C
and 29A are patterned to complete the spacer.

As described above, by laminating the Bon type photoresist films 29A and 29C and the negative type photoresist film 29B in a sandwich structure, a sufficiently thick spacer can be obtained.
Furthermore, when patterning, relatively thin photoresist films can be patterned one after another, so there is no possibility of the pattern being disrupted, which is the case when patterning a thick photoresist film.

See Figure 5 (hl) A normal photoelectric conversion semiconductor device as shown in the figure is prepared.This is realized by a technique that is conventionally widely used when manufacturing this type of semiconductor device, similar to the charge injection semiconductor device described above. Therefore, a description of the manufacturing process will be omitted.

In the figure, 31 is a p-type InSb substrate, 32 is an insulating film made of silicon dioxide, and 33 is an n+ type region 34 for creating a pn junction for photoelectric conversion between the p-type InSb substrate 31 and the p-type InSb substrate 31. Each bump is shown in the figure.

Note that a substrate having this configuration is defined as a substrate having a photoelectric conversion portion.

(11) The substrate having the charge injection treated portion and the substrate having the photoelectric conversion portion are faced to each other as shown in the figure, and a pressing force is applied as shown by the arrow to bond the electrode 27 and the bump 34.

In this case, for example, the substrate 21 is placed on a fixed table, and the substrate 3
A flat presser is brought into contact with the back surface of the substrate 21 or 31 to apply a pressing force. For example, a spring is disposed on the back surface of the presser, and even if there is an inclination on the surface of the substrate 21 or 31, The inclination is absorbed by the spring, and the back surface of the substrate 31 can be pressed almost evenly.

That is, when the surface of the substrate 21 or 31 has an inclination, when the back surface of the substrate 31 is pressed with a presser, the bumps 34 existing on the thicker portion of the substrate 21 or 31 will first be connected to the electrodes 27. The substrate 31 comes into contact with the spacer surface in a state where the bumps 34 are slightly crushed.

Furthermore, when a pressing force is applied, the presser can no longer advance in the portion where the substrate 31 contacts the spacer surface, and the pressing force acts to deform the spring;
At the thinner portion of the substrate 21 or 31, the force of the presser effectively presses the substrate 31, so the bumps 34 present at that portion come into contact with and bond with the electrode 27, and the In this case, the slight crushing of the bump 34 is stopped when the substrate 31 comes into contact with the surface of the spacer in the vicinity.

In this way, the electrodes 27 and the bumps 34 are bonded over the entire surface, and the bonding of the substrates 21 and 31 is completed.

(See FIG. 6 0) The semiconductor device is completed by removing the spacers made of photoresist films 29A to 29C.

In this case, since the bottom layer and the top layer are composed of positive photoresist films 29A and 29C that are easy to dissolve and remove, a negative photoresist film that requires a special solution to remove them. Even if 29B is stacked, the entire spacer can be easily removed by, for example, dipping it in acetone to dissolve the bottom and top layers.

In the above embodiment, the spacer was constructed using the positive photoresist film 29A, the negative photoresist film 29B, and the positive photoresist film 29G. As long as this is true, the gap between them may be composed of multiple layers of a negative type photoresist film and a positive type photoresist film. In addition, the thickness of the photoresist film, that is, the thickness of the spacer is the same as bump 3.
Needless to say, the decision will be made in consideration of 4.

Furthermore, although the bumps 34 in the above embodiments were formed only on the substrate side having the photoelectric conversion portion, they may be formed on the substrate side having the charge injection processing portion, or they may be formed on both substrates. It's okay.

Furthermore, in the above embodiment, an InSb substrate having a pn junction was used as the substrate having the photoelectric conversion portion, but in addition,
Water SFa (Hg)Cd made by epitaxial growth on an extremely thin InSb substrate bonded to a sapphire plate or a cadmium (Cd)/tellurium (Te) plate.
-Te substrate or the like can be used.

〔Effect of the invention〕

In the method for manufacturing a semiconductor device according to the present invention, when bonding a substrate having a photoelectric conversion portion and a substrate having a charge injection processing portion having a function of injecting and processing charges generated in the photoelectric conversion portion, A spacer is formed around at least one of the substrates, the bottom layer and the top layer are positive resist films, and at least one negative resist film is interposed between them, and then a spacer is formed on the surface of the substrate. The process includes the steps of bonding the substrates together by applying pressure to a bump made of a certain soft metal, and then removing the spacer.

In this way, even if the thickness of the substrate is not uniform, when pressure is applied between the substrates, the pressure applied to each bump will be approximately equal due to the effect of the spacer, and strong pressure will only be applied to specific bumps. Accidents such as the bumps collapsing due to the application of bumps and coming into contact with adjacent bumps, or insufficient pressing force being applied to the bumps resulting in poor bonding, etc., are eliminated, and the manufacturing yield is significantly improved. Moreover, since the spacer is constructed by laminating photoresist films, the bottom layer and the top layer are positive photoresist films that easily dissolve. Since it can be removed, the implementation of n starting from zero is extremely easy.

[Brief explanation of drawings]

1 to 6 are cross-sectional side views of essential parts of a semiconductor device at key points in the process for explaining one embodiment of the present invention, and FIG. 7 is a cross-sectional side view of essential parts of a semiconductor device for explaining a conventional example. Side view, No. 8
Each figure shows a side view of a main part of a semiconductor device to explain problems when bonding substrates together. In the figure, 20A is a pad portion, 20B is a charge injection portion, 21 is a p-type silicon substrate, 22 is an insulating film made of silicon dioxide, 23 is an n4s type region, 24 is an electrode made of aluminum, and 25 is made of aluminum. wiring,
26 is silicon dioxide, 9A and 29C are positive photoresist films, 29B is a negative photoresist film,
31 is a p-type InSb substrate, 32 is an insulating film made of silicon dioxide, and 33 is an n+ type region 34 which is a bump. Patent Applicant Fujitsu Ltd. Representative Patent Attorney Akira Aitani Representative Patent Attorney Hiroshi Watanabe - Figure 1 Figure 2 Flute 3 Figure 9C 20A 20B Figure 5 W-0A20B Figure 7 Figure 8

Claims (1)

[Claims] When bonding a substrate having a photoelectric conversion portion and a substrate having a charge injection processing portion having a function of injecting and processing charges generated in the photoelectric conversion portion, a bottom layer and a top layer are formed around the periphery of at least one of the substrates. is a positive resist film and at least one negative resist film is interposed between the spacers, and then by applying pressure to bumps made of soft metal on the surface of the substrate. A method for manufacturing a semiconductor device, comprising the steps of bonding substrates together and then removing the spacer.