JPH10135224A - Bump-forming method for semiconductor element - Google Patents

Abstract

PROBLEM TO BE SOLVED: To easily form bumps as high as possible with little dispersion in their heights. SOLUTION: A pad electrode, having a specified shape and area is formed on a semiconductor element, and N approximately spherical metal grains of approximately const. radius (N is natural no. 2 or greater) are mounted and fused on the pad electrode to form a bump for the semiconductor element. The min. area SN of the pad electrode for mounting the N metal grains is obtd., then an occupied area TN (TN=SN/N) per metal grain is obtd. from the min. area SN and N, the no. of the grains and no. Ni of metal grains and min. areas SNi+1 , SNi of the pad electrode for a max. change ratio ΔT (TN+1 -TN) of the occupied area are obtd. Based on the result, the no. N of the metal grains for forming a bump is set to Ni and area SN of the electrode, so that SNi <=SN<SNi+1 .

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a method for forming a bump on a pad electrode in a flip-chip type semiconductor device or the like.
The present invention relates to a bump forming method for a semiconductor element in which the formed bumps are high and the height does not vary.

[0002]

2. Description of the Related Art In recent years, as the number of electrodes has been increased and the electrode pitch and dimensions have been miniaturized, many connection techniques using bumps have been widely used for connection between a semiconductor element and a circuit board. , TAB system, and the like.

[0003] Among these, as a bump forming method in the flip-chip method, a vapor deposition method, an electrolytic plating method, a stud bump method and the like have been conventionally employed. However, the vapor deposition method has recently increased the diameter of a wafer, a bump pitch and a shape. However, the stud bump method is used only on a trial basis because of its high cost.In terms of mass production, in recent years, the electrolytic plating method has become mainstream instead of the vapor deposition method. is there.

[0004]

However, this method of forming solder bumps by electrolytic plating also has the following problems. First, the equipment used in the photoresist formation process and the base metal film formation process is expensive and the size of the wafer to be handled is limited. Could not handle. For this reason, this method increases the cost and makes it impossible to form bumps in units of chips other than wafers.

Second, when bonding the solder bumps to the substrate, the solder bumps should be about 20 μm in order to prevent the solder bumps from being crushed and shorted to the side surface of the semiconductor element.
Is plated on the underlying metal film. For this reason, since copper is a rigid body and firmly adheres to the underlying metal film, the semiconductor element expands and contracts when a temperature change of heating and cooling is applied as the chip size increases, and the copper core becomes a silicon substrate. , And the interface is peeled, the connection is cut, and the silicon is cracked.

Third, there is a step of etching the base metal film. For example, when aluminum, chromium, or copper is used as the base metal film, it is difficult to etch the base metal film without attacking lead and tin. It was difficult to control the etching amount and time.

Therefore, the present applicant has proposed a new bump forming method which does not have the above-mentioned problem in Japanese Patent Application No. 8-47767. In this new bump forming method, an adhesive is applied onto a pad electrode of a semiconductor element, metal particles are attached to the adhesive, and then the metal particles are melted to form a bump.

What is important when forming bumps in such a method is to make the formed bumps as high as possible from the viewpoint of accuracy and to eliminate variations in the heights. Is easy.

Therefore, according to the present invention, when metal particles are mounted on a pad electrode to form a bump, the height of the bump is made as small as possible, and the height of the bump is small.
In addition, it is an object of the present invention to provide a bump forming method for a semiconductor device in which a bump can be easily formed.

The present invention is not limited to the method of adhering metal particles on a pad electrode using the above-mentioned pressure-sensitive adhesive, but may be any method for forming a bump on a pad electrode using metal particles. It can be applied to

[0011]

In view of the above, the inventors of the present invention have conducted intensive studies. First, in order to increase the height of the formed bump, it is necessary to mount a certain number of metal particles on a pad electrode. It is found that the minimum area can be determined unconditionally because the relationship between the number of metal particles and the minimum area differs depending on the shape of the pad electrode, although the area may be set to the minimum area that can be mounted.

Second, in order to reduce the variation in the height of the bump, the number of metal particles used for forming the bump may be increased to a plurality. It was found that the accuracy was dramatically improved as compared with. This is because the volume of the bump is the cube of the radius of the metal particle. Even if the accuracy of the same metal particle is more than one, the accuracy is better in the case of a plurality of metal particles (FIG. 4). See Figure 1). Note that the dispersion accuracy of the metal particles is ±
Although it is not impossible to make it 5% or less, the cost becomes very high and it is not practical. At present, metal particles having an accuracy of about ± 10% have to be used in terms of cost.

On the other hand, in order to easily form a bump, it is considered that a predetermined number of metal particles can be mounted on a pad electrode with the most margin, that is, a case where the exclusive area of the pad electrode for one metal particle is the widest. . But,
In a situation where miniaturization of the pad electrode is increasingly promoted, the area of the pad electrode cannot be increased without limit. Therefore, within the allowable area of the pad electrode,
After studying how to obtain the maximum occupied area, it was found that when the number of metal particles to be mounted was changed, the rate of change of the occupied area per metal particle was the largest. This is the third finding.

The present invention has been made based on each of these findings. Metal particles having a substantially spherical shape and a substantially constant radius are formed on a pad electrode formed on a semiconductor element and having a predetermined shape and area. N is a natural number of 2 or more)
And a method for forming a bump of a semiconductor element by melting and forming a bump, wherein a minimum area S _N of the pad electrode on which the N metal particles can be mounted is determined, and then the minimum area S _N and the number of metal particles are determined. The occupation area T _N (T _N = S _N / N) per metal particle is determined from N and the metal when the change rate ΔT (T _{N + 1} −T _N ) of the occupation area is the maximum. determining the minimum area S _{Ni + 1,} S _Ni of the number N _i and the pad electrode of the particles, based on this result, the number N of the metal particles in forming a bump with a N _i number, the area of the pad electrode S _{N is set so} that S _Ni ≦ S _N <S _{Ni + 1} . This makes it possible to easily form bumps that are high and have a small variation in the height.

[0015] In the above bump forming method, the outer shape of the pad electrode can be circular, and the number of metal particles is preferably four at this time. This makes it possible to form bumps having a relatively high height and a circular outer shape easily and with little variation in height.

In the above bump forming method, the outer shape of the pad electrode may be square, and the number of metal particles is preferably five at this time. This makes it possible to form bumps having a relatively high height and a square outer shape easily and with little variation in height.

Further, in the above bump forming method, the outer shape of the pad electrode can be a regular triangle, and the number of metal particles is preferably six at this time. Furthermore, when the outer shape of the pad electrode is an equilateral triangle,
It is preferable to arrange such that one side is parallel to the outer periphery of the semiconductor element, and when there are a plurality of equilateral triangular pad electrodes, it is preferable to arrange so that opposing sides of adjacent pad electrodes are parallel. preferable. With this configuration, not only can a bump having a relatively high triangular shape and an outer shape be easily formed with a small variation in height, but also a bump having a regular triangular shape can be formed on a semiconductor element at a high density. .

In the above bump forming method, the outer shape of the pad electrode can be rectangular, and the number of metal particles is preferably six at this time. further,
When the outer shape of the pad electrode is rectangular, it is preferable to arrange the pad electrode so that its short side is parallel to the outer periphery of the semiconductor element. It is preferable that the sides are arranged in parallel.

In the above bump forming method, the outer shape of the pad electrode can be rhombic, and the number of metal particles is preferably six at this time. Further, when the outer shape of the pad electrode is rhombic, it is preferable to arrange the short diagonal of the diagonal lines so as to be parallel to the outer periphery of the semiconductor element. It is preferable that the pad electrodes are arranged such that opposing sides thereof are parallel to each other.

Even when the pad electrodes are rectangular or diamond-shaped as described above, the bumps can be formed easily and with little variation in height, and the bumps can be arranged with high density.

Preferably, the corners of the pad electrodes are chamfered. By doing so, it is possible to prevent interference between adjacent bumps, to facilitate the formation of the bumps, and to increase the density of the arrangement of the bumps.

It is preferable that the adhesive is applied to each of the pad electrodes in advance, and then the metal particles are mounted. This ensures that the metal particles adhere to the pad electrode, thereby preventing the metal particles from dropping in a later step, facilitating the bump formation and improving the accuracy of the bump height.

[0023]

Embodiments of the present invention will be described below. First, an example of a method of mounting a plurality of spherical metal particles on a pad electrode formed on a semiconductor element and then melting the metal particles to form a bump will be described.

FIG. 1 is a process diagram, and FIGS. 2A to 2F are cross-sectional views of an electrode portion in main steps. In the pretreatment step S1 in FIG. 1, the oxide film of the aluminum pad electrode 2 of the semiconductor element completed on the wafer 1 is removed as shown in FIG. Next, activation processing step S
In step 2, an activation process for selectively electrolessly plating nickel on the pad electrode 2 is performed. Then, in the electroless nickel plating step S3, a nickel film 11 as a metal is electrolessly plated on the pad electrode 2 as shown in FIG. Reference numeral 3 denotes a passivation film.

Next, in the pressure-sensitive adhesive treatment step S4, FIG.
As shown in (1), the wafer 1 is immersed in a chemical bath to selectively apply the adhesive 4 to the metal portion exposed on the wafer 1, and then dried. Then, in the solder particle attaching step S5, the solder particles 12 are sprinkled on the wafer 1 as shown in FIG. Then, in a post-processing step S6, after performing a heat treatment for temporarily bonding the solder particles to the electrode portions, as shown in FIG. 2 (e), the solder particles 12 other than the portion where the adhesive 4 is applied are removed. Gently brush and remove. As a result, only the solder particles 12 in the portion where the adhesive 4 is applied remain.

After the plurality of solder particles 12 are fixed to the metal portion of the wafer 1 in this manner, a flux is applied on the surface of the wafer 1 in a flux application step S7.
Next, in the solder reflow step S8, the molten solder 14 is rolled into a ball shape by surface tension as shown in FIG. Thereafter, after cleaning and drying, inspection is performed in an inspection step S9 to complete the solder bumps 10 on the wafer. If the thickness of the solder is insufficient after the solder reflow step S8, the process returns to the adhesive processing step S4, and the subsequent steps are repeated.

Here, the solder particles include Pb / Sn, Ag / Sn / Zn, Zn / Sn, Sn / Cu,
Various things such as Sn / Ag / Bi and Sn / In can be applied. As the metal particles, in addition to the above solder particles,
A metal core when a core is put in the bump is also included.

Next, one embodiment of the bump forming method of the present invention will be described. FIG. 3 is a view showing the relationship between the number of metal particles and the mounting state on the pad electrode when the external shape of the pad electrode is circular, square, triangular, rectangular, or rhombic. For each external shape of the pad electrode, a pad electrode area (minimum area) _SN required at a minimum when N metal particles are mounted, for example, in the state shown in FIG. 3 is obtained. This can be determined theoretically and / or experimentally from the size of the metal particles and the degree of variation in the size.

From the minimum area S _N of the pad electrode obtained in this way and the number N of mountable metal particles, the minimum required area (exclusive area) T _N (T _N = T _N = one metal particle)
S _N / N). FIG. 5 (Chart 2) and FIG. 6 (Chart 3) show the minimum area and the amount of change of the occupied area when the external shape of the pad electrode is circular, square, equilateral triangle, rectangular, and rhombic. .

According to FIGS. 2 and 3, it can be understood that the minimum area and the occupied area when the number of metal particles is N are different depending on the outer shape of the pad electrode. Further, it can be understood that even in the pad electrodes having the same external shape, the change rate of the occupied area varies depending on the number of metal particles mounted.

As described above, the size of the pad electrode at the time of forming the bump is desirably the smallest occupied area in order to make the bump highest, while the metal particles are stably deposited on the pad electrode. It is desirable that the occupied area be as large as possible for mounting. In order to satisfy these two contradictory conditions, the number of metal particles must be N _i while maintaining the minimum area.
Change rate ΔT of the occupied area when changing from N to N _{i + 1}
(T _{N + 1} -T N) so that it may be selected number of the most larger metal particles.

That is, according to Chart 3 of FIG. 6, when the outer shape of the pad electrode is triangular, the minimum occupied area required for mounting six metal particles is about 1100 μm ² ,
The minimum occupied area required to mount seven units is about 3700
μm ² , during which the occupied area of the pad electrode is the largest. Therefore, as can be understood from the chart 2 of FIG. 5, if the six metal particles are mounted on the pad electrode within the area of less than about 26000 μm ² and about 6600 μm ² or more, the above two conditions are satisfied. It can be in the state of having done.

In this case, the minimum area of the pad electrode is about 66
00Myuemu ² bumps the closer (the footprint about 1100 .mu.m ²⁾ the value of the higher (ease of bump formation is slightly lower), the minimum area of about 26000Myuemu ² to a value smaller than (about 3700Myuemu ² the occupied area) The closer the bump is, the easier it is to form the bump (the height of the bump is slightly lower). An arbitrary minimum area (exclusive area) within the above range may be selected in accordance with (1).

When the pad electrode has a circular outer shape, the change rate of the occupied area becomes the largest when the number of metal particles to be mounted becomes four to five (about 1400 μm).
² to about 2000 μm ² ), four metal particles of about 5600
If mounted on the [mu] m ² or more at about 10200Myuemu ² smaller range of pad electrodes on consisting of area, given more than maintaining a certain height of the bump, and it becomes possible to facilitate the bump formation.

Similarly, according to FIGS. 2 and 3, when the outer shape of the pad electrode is square, the number of metal particles becomes from five to six, and when the outer shape of the pad electrode is rectangular, four metal particles are obtained. When the outer shape of the pad electrode is rhombic, the occupied area changes the most when the number of metal particles changes from six to seven. Therefore, the number of metal particles at this time is set to square: 5, rectangular: 4, and diamond: 6,
Each square area of the pad electrode when the shape: about 6500Myuemu ² or more at about 11500Myuemu ² smaller ranges, rectangle: about 7300Myuemu ² or about 10000μm
Range smaller than ² , diamond: about 7800 μm ² or more and about 1
When the thickness is smaller than 3700 μm ² , the height of the bumps can be kept at a certain level or more, and the bumps can be easily formed.

The outer shape of the pad electrode is not limited to the above shape, but may be, for example, an ellipse or a polygon. Even in the case of these shapes, the minimum area and the occupied area of the pad electrode can be theoretically and / or experimentally determined according to the number of metal particles.

Next, the arrangement of the pad electrodes according to the external shape of the pad electrodes will be described with reference to FIGS. That is, when the pad electrode 2 is triangular, as shown in FIG. 7A, the side 2a is arranged so as to be parallel to the side 20a of the semiconductor element 20, and two or more triangular pad electrodes 2 are formed. When adjacently arranged, it is preferable that the opposing sides 2b, 2b are parallel. This makes it possible to arrange the pad electrodes at a high density by narrowing the gap between the pad electrodes.

Similarly, when the pad electrode 2 is rectangular,
As shown in FIG. 7B, the short side 2a is
0 is arranged parallel to one side 20a, and when the pad electrode 2 is rhombic, the short diagonal 2a of the diagonal is parallel to the side 20a of the semiconductor element 20, as shown in FIG. It is preferable to arrange them so that Also in such an arrangement, the pad electrodes are arranged neatly, and the pad electrodes can be arranged at a high density.

When the outer shape of the pad electrode is square, it is preferable to chamfer the corner as shown in FIGS. 7 (a) to 7 (c). This facilitates the formation of the pad electrode, and can eliminate a corner portion that does not substantially function as an electrode, which is effective in preventing interference between the pad electrodes.

The present invention is not limited to the above embodiment, but includes various modified embodiments within the scope of the invention. For example, the present invention can be applied to a case where an adhesive is not applied to a pad electrode in advance.

[0041]

As described above, according to the method for forming a bump of a semiconductor device of the present invention, the bump is formed as high as possible, the height thereof is small, and the bump can be formed easily. Can be

[Brief description of the drawings]

FIG. 1 is a process chart showing one embodiment of a bump forming method for a semiconductor device of the present invention.

2 (a) to 2 (f) are cross-sectional views of an electrode portion in main steps of the steps in FIG.

FIG. 3 is a plan view showing a mounting state of metal particles for each shape of a pad electrode.

FIG. 4 is a chart 1 showing a relationship between the variation accuracy of one solder particle and the variation accuracy of a bump composed of N solder particles.

FIG. 5 is a chart 2 showing the relationship between the number of solder particles and the minimum area of the pad electrode for each shape of the pad electrode.

FIG. 6 is a chart 3 showing an occupied area per solder particle for each shape of the pad electrode.

FIGS. 7A to 7C are plan views showing the arrangement of the pad electrodes for each shape of the pad electrodes.

[Explanation of symbols]

 2 Pad electrode 10 Solder bump 12 Solder particle 20 Semiconductor element

Claims (18)

[Claims] An N-shaped (N is a natural number not less than 2) metal particle having a substantially spherical shape and a substantially constant radius is mounted on a pad electrode having a predetermined shape and area formed on a semiconductor element.
And forming a bump by melting the semiconductor element, wherein a minimum area S _N of the pad electrode on which the N metal particles can be mounted is determined. Then, the minimum area S _N and the number of metal particles are determined. The occupation area T _N (T _N = S _N / N) per one metal particle is obtained from N and the metal when the rate of change ΔT (T _{N + 1} −T _N ) of the occupation area is maximized. The number N _{i of} particles and the minimum areas S _{Ni + 1} and S _Ni of the pad electrode are obtained. Based on the results, the number N of metal particles when forming a bump is set to N _i and the area of the pad electrode is determined. S
_A method for forming a bump of a semiconductor device, wherein _{N is set} to S _Ni ≦ S _N <S _{Ni + 1} . 2. The method according to claim 1, wherein the outer shape of the pad electrode is circular. 3. The method according to claim 2, wherein the number N of the metal particles to be mounted is four. 4. The method according to claim 1, wherein the outer shape of the pad electrode is square. 5. The method according to claim 4, wherein the number N of the metal particles to be mounted is set to five. 6. The method according to claim 1, wherein the outer shape of the pad electrode is an equilateral triangle. 7. The bump forming method for a semiconductor device according to claim 6, wherein the regular triangular pad electrode is arranged so that one side thereof is parallel to the outer periphery of the semiconductor device. 8. The bump forming method for a semiconductor element according to claim 6, wherein the regular triangular pad electrodes are arranged such that opposing sides of adjacent regular triangles are parallel to each other. 9. The method according to claim 6, wherein the number N of the metal particles to be mounted is six. 10. The method according to claim 1, wherein the outer shape of the pad electrode is rectangular. 11. The bump forming method for a semiconductor device according to claim 10, wherein the rectangular pad electrode is arranged such that a short side thereof is parallel to an outer periphery of the semiconductor device. 12. The method according to claim 10, wherein the number N of the metal particles to be mounted is four. 13. The method according to claim 1, wherein the outer shape of the pad electrode is rhombic. 14. The bump forming method for a semiconductor device according to claim 13, wherein the diamond-shaped pad electrode is disposed such that a short side of a diagonal line is parallel to an outer periphery of the semiconductor device. 15. The method according to claim 13, wherein the number N of the metal particles to be mounted is six. 16. The method for forming a bump of a semiconductor device according to claim 1, wherein a corner of the pad electrode is chamfered. 17. The method according to claim 1, wherein an adhesive is previously applied to the pad electrode, and then metal particles are mounted.
7. The method for forming a bump of a semiconductor device according to any one of 6. 18. The method for forming a bump of a semiconductor device according to claim 1, wherein solder balls are used as the metal particles.