JP2003347337A - Semiconductor device and its manufacturing method - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To accurately measure electric characteristics of a semiconductor device by a probe needle and reliably form a bump electrode on the semiconductor device after measuring the electric characteristics. <P>SOLUTION: An electrode pad 2 is formed on a semiconductor substrate 1. The probe needle is contacted with a gold layer 5 formed on the electrode pad 2 to measure the electric characteristics of the semiconductor device. Thereafter, a bump electrode such as a gold stud bump 81 or the like is formed on the gold layer 5 to which the probe needle 6 is contacted. When a solder ball is used as a bump electrode, a barrier metal and a gold layer are sequentially formed on the electrode pad, and thereafter, the probe needle is contacted with the gold layer to measure the electric characteristics, and then, the solder ball is formed on the gold layer. <P>COPYRIGHT: (C)2004,JPO

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

The present invention relates to a semiconductor device having metal bumps formed on electrode pads of a semiconductor substrate, so-called
More particularly, the present invention relates to a flip chip structure, measurement of electrical characteristics thereof, and a method of forming metal bumps.

[0002]

2. Description of the Related Art A semiconductor device realizes a desired circuit function by connecting circuit elements such as transistors and resistors to a semiconductor substrate through a metal wiring layer. In order to supply a power supply voltage and various signals to the semiconductor device, an electrode pad is formed on the semiconductor substrate. A semiconductor device in which metal bumps are formed on the electrode pads is generally called a flip chip. The metal bumps in the flip chip electrically and mechanically connect the flip chip and the printed circuit board on which the flip chip is mounted.

Before the flip chip is mounted on the printed circuit board, it is a matter of course that an electrical characteristic test of the circuit function must be performed. Such a test is performed by bringing a probe needle into contact with a metal bump. However, a needle mark of the probe needle, that is, a depression is generated on the surface of the metal bump contacted with the probe needle. When the flip chip is mounted on the printed board, the depression on the surface of the metal bump causes a reduction in the reliability of the connection between the two.

Further, since metal bump debris adheres to the probe needle in contact with the metal bump, it is necessary to clean the probe needle. The adhesion of this metal bump scrap to the probe needle
This is particularly noticeable when the metal bumps are solder balls.

In view of this, it is conceivable that a probe needle is brought into contact with the electrode pad to conduct an electrical characteristic test, and then a metal bump is formed. In this method, the above-described problem of the depressions on the surface of the electrode bumps and adhesion of metal bump debris to the probe needle is solved. However, this time, a needle trace is generated on the electrode pad due to the contact of the probe needle, and this needle trace lowers the reliability of the connection between the metal bump and the bump electrode.

A technique for solving such a problem is disclosed in JP-A-2
001-93933. According to such a technique, the electrode pad of the semiconductor element is covered with a conductive paste, the paste is cured, a probe needle is brought into contact with the conductive paste thus formed, a test is performed, and then solder bumps are formed.

[0007]

However, this technique has several problems. The first is due to the conductivity of the conductive paste. When testing the electrical characteristics, the probe needle is electrically connected to the aluminum pad via a conductive paste. Since the conductive paste contains a resin as its component, the conductivity is smaller than that of the metal layer. As a result, this is equivalent to a state in which an electrical resistance is inserted between the probe needle and the aluminum pad. This electrical resistance causes an error in the measured value, and accurate characteristic evaluation cannot be expected.
Further, the problem of electrical resistance of the conductive paste is a concern even after solder bumps are formed and mounted on a printed circuit board as a flip chip.

The second problem relates to the solder wettability of the conductive paste. When solder balls are employed as the metal bumps, the reliability of the connection between the solder bumps and the conductive paste decreases due to the problem of solder wettability of the conductive paste. In addition, the electrical connection resistance between the two increases.

An object of the present invention is to provide a method of manufacturing a semiconductor device having a bump electrode that can solve problems that may occur during the execution of a test with a probe needle.

[0010]

A manufacturing method according to the present invention comprises a step of forming a gold layer on an electrode pad formed on a semiconductor substrate, a step of bringing a probe needle into contact with the gold layer, And a step of forming metal bumps on the contacted gold layer.

That is, the present invention is characterized in that the electrode pad of the semiconductor element is covered with a gold layer, a probe needle is brought into contact with the gold layer, a test is performed, and then a bump electrode is formed.

The electrical resistance of gold is extremely small, and it is very excellent in the connectivity and wettability with solder bumps and gold bumps exclusively used as bump electrodes. Furthermore, a substantial natural oxide film is not formed on the surface of the gold layer. This means that even if the pressure applied to the probe needle when the probe needle is brought into contact with the gold layer is reduced, the contact resistance between the two is kept sufficiently small. Therefore, the probe needle trace remaining on the gold layer surface is very small, and the electrical connection and the wettability with the bump electrode are excellent. can get.
Of course, the test can be performed without generating a depression in the bump electrode, and the pressure applied to the probe needle can be reduced, so that the probe needle does not break through the gold layer and reach the electrode pad to damage it.

When forming a solder bump as the metal bump, it is preferable to form a barrier metal between the electrode pad and the gold layer, that is, to cover the electrode pad with the gold layer via the barrier layer. This barrier layer prevents diffusion of solder components of the solder bumps to the electrode pads. A metal layer mainly composed of nickel or copper is suitable as the barrier metal. Of course, gold bump electrodes may be used.

When the electrode pad is directly covered with a gold layer, stud bumps made of gold wires are preferably formed as metal bumps.

Further, a step of pressing the gold layer surface by pressing means having a flat tip shape may be added before forming the metal bumps.

[0016]

FIG. 1 shows a manufacturing process of a semiconductor device having metal bumps according to a first embodiment of the present invention. An electrode pad 2 made of aluminum, an alloy containing aluminum as a main component, copper, or an alloy containing copper as a main component is formed on a semiconductor substrate 1 on which circuit elements are formed. Thereafter, a surface protective film 3 made of an insulating film such as a silicon oxide film is formed on the surface of the semiconductor element, and an opening 4 exposing the electrode pad 2 is formed in the surface protective film 3. The opening 4 is filled with a gold layer 5. In this embodiment, the gold layer 5 is formed on the electrode pad 2 by electroless plating, and the surface thereof is the surface protective film 3.
It was formed so as to be almost the same height as the surface of. Through the manufacturing steps so far, the electrode pad portion having the cross-sectional shape of FIG.

After the gold layer 5 is formed on the electrode pad 2, the gold layer 5
In addition, for example, a probe needle 6 made of tungsten (W) is brought into contact. As a result, the probe needle 6 has the gold layer 5
It is electrically connected to the electrode pad 2 via
Since the electrical conductivity of gold is orders of magnitude greater than that of the conductive paste, the electrical resistance of the gold layer 5 existing between the probe needle 6 and the electrode pad 2 is not a problem at all, and the electrical characteristics of the semiconductor device are accurate. Can be measured.

FIG. 1B is a sectional view of the electrode pad portion in a state where the probe needle 6 is separated from the gold layer 5 after the electrical measurement of the semiconductor device is completed. On the surface of the gold layer 5, a needle mark 7 is formed by contact with the probe needle. In the drawing, the size of the needle trace 7 is exaggerated, but in reality, a substantial natural oxide film is not formed on the surface of the gold layer 5, so even if the applied pressure of the probe needle 6 is reduced. The contact resistance between the gold layer 5 and the needle 6 can be reduced. As a result, the needle trace 7 does not interfere with the subsequent bump electrode formation, and of course, the electrode pad 2
Never reach.

The pressure applied to the probe needle will be described in more detail. When the probe needle is brought into contact with the electrode pad, normally, until the tip of the probe needle reaches the inside of the metal layer constituting the electrode pad, in other words, the electrode Press the probe needle against the electrode pad to the extent that the needle mark is on the surface of the pad. This is for breaking through the oxide film formed on the surface of the metal layer to be the electrode pad and lowering the contact resistance between the probe needle and the electrode pad.

In particular, when the electrode pad is aluminum,
The aluminum oxide film formed on the surface has higher hardness than aluminum. Therefore, it is necessary to bring the probe needle into contact with the electrode pad at a pressure higher than that required to bring the probe needle into contact with the aluminum electrode pad.
As a result, once the tip of the probe needle breaks through the oxide film on the aluminum surface constituting the electrode pad, the needle trace formed on the electrode pad becomes wide and deep enough to expose the semiconductor substrate on which the electrode pad is formed. There is a case. Such a deep needle trace results in a decrease in the reliability of the electrode pad.

In the method of forming a semiconductor device having metal bumps according to the first embodiment of the present invention, the electrode pad 2
A gold layer 5 is formed thereon. Since gold does not form an oxide film on its surface and its hardness is lower than that of the probe needle 6, the pressure required for bringing the gold layer 5 and the probe needle 6 into contact with each other is minimal. As a result, the probe needle 6 can be brought into contact without damaging the electrode pad 2.

FIG. 1C is a cross-sectional view of an electrode pad portion in which a gold stud bump 81 is formed on the semiconductor substrate 1 after measuring the electrical characteristics of the semiconductor device. Gold stud bump 81
Is a known technique, in which a gold wire is bonded to the gold layer 5 (not shown) with a bonding tool, and only the wire head portion is left as a metal bump. Since the gold stud bump 81 and the electrode pad 2 are connected via the gold layer 5, it goes without saying that the electrical resistance value between them is extremely small.

In this embodiment, 2 to 3 u are formed on the electrode pad 2.
In order to form the gold layer 5 having a film thickness of m, metal bumps were formed with a metal other than the solder bumps. This is because the solder bump formed on the thick gold layer has a brittle interface with the gold layer and cannot maintain the mechanical strength.

FIG. 2 shows a manufacturing process of a semiconductor device having metal bumps according to the second embodiment of the present invention. This embodiment particularly shows an example suitable for the case where solder bumps are used as bump electrodes, but gold bumps can be similarly applied. In addition, the same number is attached | subjected to the structure same as FIG.

FIG. 2A shows the cross-sectional structure of the electrode bump portion immediately before measuring the electrical characteristics of the semiconductor device. Forming an opening 4 for exposing the electrode pad 2 in the surface protective film 3;
The opening 4 is filled with a two-layer structure of a barrier metal layer 51 and a gold layer 52. As the barrier metal layer 51, nickel (Ni) or copper (Cu), or a metal layer containing them as a main component is suitable. In this embodiment, nickel is formed with a thickness of 2 to 3 μm by electroless plating. A gold layer 51 was formed on the barrier metal layer 51 to a thickness of 0.01 to 0.1 μm by electroless plating.

Solder bumps are formed by using lead (Pb) and tin (S
After forming a solder ball having n) as a main component or a solder plating layer by electrolytic or electroless plating on the electrode pad 2, the semiconductor device is heated and bonded to the electrode pad 2. When the electrode pad is made of aluminum or an alloy mainly composed of aluminum, lead and tin do not form an alloy with aluminum, so that the solder bump is easily peeled off from the electrode pad. In order to solve this problem, a barrier metal layer 51 that prevents lead and tin from reaching the aluminum electrode pad is interposed between the gold layer 52 and the electrode pad 2.

Also, the solder bumps require good wetting with the lower layer. The gold layer 52 also exhibits very good characteristics in this respect.

FIG. 2B shows a cross-sectional shape after the barrier metal layer 51 and the gold layer 52 are sequentially formed on the electrode pad 2 and the surfaces thereof are contacted by the probe needle 7. Similar to the first embodiment, in this embodiment, the probe needle 7 has the gold layer 52.
Can be performed with the minimum necessary contact pressure.

In this embodiment, a barrier metal layer 51 made of nickel having a hardness higher than that of gold is formed under the gold layer 52. Therefore, even if the probe needle 6 penetrates the gold layer 52 as a result of forming the gold layer 52 very thin, for example, 0.01 to 0.1 μm,
The lower barrier metal layer 51 can prevent the probe needle 6 from reaching the pad electrode 2.

FIG. 2C is a cross-sectional view after bonding the solder balls 82 onto the semiconductor substrate 1 after measuring the electrical characteristics of the semiconductor device. After the solder balls 82 are temporarily bonded to the gold layer 52 by flux, heat treatment (reflow) is performed to mechanically and electrically connect the solder balls 82 to the electrode pads 2. The solder ball is melted by this heat treatment, and the barrier metal 51 on the electrode pad 2 and the solder ball are joined.

On the other hand, during this heat treatment, the gold layer 52 and tin contained in the solder ball form an Au-Sn compound. Since the solder becomes brittle when this gold and tin compound is present in a large amount in the solder, there is an upper limit to the film thickness of the gold layer 52,
In this embodiment, the maximum is 0.1 μm. On the other hand, there is a lower limit to the film thickness of the gold layer 52 in order to ensure solder wettability and good contact with the probe needle 6, and in this embodiment, the minimum thickness is 0.01 μm.

As shown in FIG. 2C, after the solder balls 82 are temporarily bonded onto the gold layer 52 on which the needle marks 7 are formed, the semiconductor substrate 1 is heat-treated. As a result, the needle trace 7 becomes a solder ball 82.
However, there is no problem in the mechanical connection strength between the solder ball 82 and the gold layer 52 due to the heat treatment. Gold layer 52
This is because the size (volume) of the needle trace 7 is small because the thickness is thin.

In the second embodiment, the solder ball 82 is bonded to the semiconductor substrate 1 as an example of the metal bump. Solder may be formed as a metal bump by electrolytic or electroless plating. Alternatively, gold may be grown by electrolytic or electroless plating to form bumps. Further, the barrier metal layer 51 and the gold layer 52 may be formed by vapor deposition or sputtering.

FIG. 3 shows a manufacturing process of a semiconductor device having metal bumps according to the third embodiment of the present invention. In this embodiment, the gold layer 5 (first
Embodiment) or gold layer 52 (second embodiment)
The needle trace 7 formed in the above is flattened by the flattening jig 9.

The tip shape of the flattening jig 9 is smaller in length than one side of the corresponding opening, and the tip portion has a planar shape. The upper limit of the size of the tip is set such that the tip of the flattening jig 9 does not cover the surface protective film 3 in consideration of the mechanical alignment accuracy between the opening 4 and the flattening jig 9. The lower limit of the size takes into consideration the planar shape of the needle trace 7, and the lower limit is the size that allows the needle trace 7 to be flattened with a single press by the flattening jig 9.

The pressing operation of the electrode pad 2 by the flattening jig 9 may be performed for each electrode pad 2 or simultaneously by the plurality of flattening jigs 9 against the corresponding electrode pads 2 respectively. The same operation may be performed.

As described above, this embodiment adds the step of pressing the gold layer surface by pressing means having a flat tip shape after the test with the probe needle and before the formation of the metal bump. Yes. Therefore, the gold stud bump formation can be more stably performed by flattening the needle trace 7 of the gold layer 5 with the flattening jig 9 in the manufacturing method of the first embodiment. For the manufacturing method of the second embodiment, an effect of further improving the problem caused by the needle trace 7 during the heat treatment of the metal bumps can be obtained.

[0038]

As described above, according to the present invention,
The measurement accuracy of the electrical characteristics by the probe needle can be improved without damaging the electrode pad, and the formation of a highly reliable metal bump can be realized.

[Brief description of the drawings]

FIG. 1 is a cross-sectional view of an electrode pad portion showing a manufacturing process of a semiconductor device according to a first embodiment of the present invention.

FIG. 2 is a cross-sectional view of an electrode pad portion showing a manufacturing process of a semiconductor device according to a second embodiment of the present invention.

FIG. 3 is a sectional view of an electrode pad portion showing a manufacturing process of a semiconductor device according to a third embodiment of the present invention.

[Explanation of symbols]

1 Semiconductor substrate 2 electrode pads 3 Surface protective film 4 openings 5 gold layers 51 Barrier metal layer 52 gold layer 6 Probe needle 7 Needle marks 81 gold stud bump 82 Solder balls 9 Flattening jig

Claims (9)

[Claims] 1. A method of manufacturing a semiconductor device having metal bumps on an electrode pad, the step of forming the electrode pad on a semiconductor substrate, and forming a gold layer electrically connected to the electrode pad on the electrode pad. A method of manufacturing a semiconductor device, comprising: a step of bringing a probe needle into contact with the gold layer; and forming a metal bump on the gold layer in contact with the probe needle. 2. The metal bump is a solder bump, and includes a step of forming a barrier metal layer that suppresses diffusion of a solder component into the electrode pad between the electrode pad and the gold layer. A method for manufacturing a semiconductor device according to claim 1. 3. The semiconductor device according to claim 2, wherein the electrode pad is a metal layer mainly composed of aluminum, and the barrier metal layer is a metal layer mainly composed of nickel or copper. Production method. 4. The method of manufacturing a semiconductor device according to claim 1, wherein the metal bump is a stud bump made of a gold wire. 5. A step of pressing the surface of the gold layer by a pressing means having a flat tip shape between the step of bringing the probe needle into contact with the gold layer and the step of forming the metal bump. The method for manufacturing a semiconductor device according to claim 1, wherein: 6. An electrode pad of a semiconductor element is covered with a gold layer,
A method for manufacturing a semiconductor device, wherein a test is performed by bringing a probe needle into contact with the gold layer, and then a bump electrode is formed. 7. The electrode pad is directly covered with the gold layer,
7. The method of manufacturing a semiconductor device according to claim 6, wherein the bump electrode made of gold is formed after the test. 8. The method of manufacturing a semiconductor device according to claim 6, wherein the bump electrode is formed after the electrode pad is covered with the gold layer through a barrier layer and the test is performed. 9. A semiconductor device manufactured by the manufacturing method according to claim 1. Description: