JPH0684279B2 - Surface improvement method during epitaxial growth in liquid phase epitaxial method - Google Patents

Description

Description: TECHNICAL FIELD The present invention relates to an improvement in a liquid phase epitaxial method that does not require surface polishing in liquid phase epitaxial growth of compound semiconductors.

There are many combinations of compound semiconductors such as GaAs, GaP, InP and GaAlAs.

Epitaxy is the growth of a single crystal thin film on a single crystal substrate.

Liquid phase epitaxy is a method of growing while maintaining thermal equilibrium, so that a dense film with few defects can be obtained. For this reason, liquid phase epitaxy is most suitable for manufacturing a light emitting device that has to pass a relatively large current.

Vapor phase epitaxy (VPE), molecular beam epitaxy (MB
E) and various types of epitaxy methods such as metalorganic pyrolysis (MOCVD, OMVPE) are being actively developed.

However, liquid phase epitaxy has few defects,
It has unrivaled advantages such as excellent crystallinity. It's a classic technique, but still useful.

However, the thin film formed by liquid phase epitaxy has a drawback that irregularities appear in liquid form on the surface.

The wavelength of the wave is about 1 μm to several μm.

Liquid phase epitaxial growth surface (GaAs / GaAs) with electron microscope
When you see, you can see a group of parallel irregularities that resemble the waves rolling in the sea. Such patterns are sometimes called terrace patterns or wave patterns.

The liquid phase epitaxial growth surface having such a wave pattern is called morphology (Morphoogy).

(A) Liquid phase epitaxy and wave pattern Wave pattern appears widely in liquid phase epitaxial growth method. First, we must emphasize this.

In addition to GaAs / GaAs, it also appears when homoepitaxial growth of the same on a GaP or InP substrate. GdTe etc.
It also appears in II-VI liquid phase epitaxy.

At any time with liquid phase epitaxy, there are several different methods.

There are taping method (tilt method), dateing method, slide boat method, and the like.

In the taping method, a substrate and a solution are put in a container, and the container is tilted to the left and right to bring the substrate and the solution into contact with each other and separate them.

In the dipping method, the substrate is grasped in the solution from above and immersed in the vertical direction. It will be raised when the growth is over. Still used in the production of issue diodes.

In the slide boat method, a concave plate for a substrate is formed in a slide plate of a graphite, a boat having some bottomless raw material solution reservoirs is placed on the concave portion, and they are moved relative to each other, and different epitaxies are sequentially performed. This is the mainstream of liquid phase epitaxy. Like a semiconductor laser, several epitaxial growth films with different compositions can be grown continuously.

The wave pattern described above appears in any of these three methods.

That is a common problem for liquid phase epitaxy.

However, why does the wave pattern appear? I don't know for sure about that. There are various theories. But I can't trust either.

(C) Vertical liquid phase epitaxial growth method However, the problem of the present invention is not any of the above-mentioned three methods.

In the liquid phase epitaxial growth method of the present invention, a wafer is placed in a shallow bottomed circular carbon dish, and a large number of the wafers are vertically stacked.
Liquid phase epitaxy is carried out by immersing this in a solution.

This will be described with reference to FIG.

A circular wafer 1 is placed in a jig 2 made of carbon. Many such jigs are stacked. Such an assembly as a whole is housed in the quartz cassette 4.

The solution 3 containing the solute is contained in this.

These are surrounded by the heater 5 and heated by this. The solution 3 also fills the inside of the jig 2 and contacts the upper surface of the wafer 1.

If the temperature is lowered, it will become supersaturated, and the compound will be deposited on the wafer.

This method is a method developed by the applicant. For example, JP-B-62-35998 (S62.8.5), JP-A-59-128298 (S5
(Published 9.7.25).

It is called vertical liquid phase epitaxy because wafers are stacked vertically.

This jig 2 is made of carbon. Thickness is about
2 mm, the depth of the concave and convex in which the wafer is inserted is about 1 mm, and the thickness of the wafer is about 0.3 mm. These are all examples.

Then, the thickness of the solution is 0.7 mm. Extremely thin. It can be thicker, of course, but doing so will reduce productivity.

By such a method, for example, a GaAs thin film doped with impurities can be grown on a GaAs wafer. In this case, the solution is Ga and the solute is GaAs. For example, Si is added as an impurity.

Of course, a terrace pattern is generated on the surface of the epitaxial thin film thus grown.

FIGS. 5 (a) and 5 (b) show an example of measuring the unevenness of the surface of the GaAs / GaAs epitaxial thin film. The horizontal axis is the position on the wafer surface at right angles. The vertical axis is the height of the surface, and one scale is 5μ.
m.

In the example of (a), there is unevenness of 5 μm in height from peak to peak. In the example of (b), there is unevenness of 10 μm.

(D) Prior art An insulating film and a metal electrode must be further attached on the epitaxial thin film by a wafer process.

According to the photoetching method, in this case, if the surface of the thin film is uneven, the adhesion to the photomask is poor. Therefore, the pattern is blurred. Of course, it is impossible to make thin lead patterns with good reproducibility. When the chip is completed, put it in a package and perform automatic wire bonding.
If there are irregularities on the surface, it will be
Misunderstand the bond position.

A wafer as-grown by liquid phase epitaxial growth is referred to herein as as-grown.

Since the as-grown wafer has many irregularities, it cannot be directly used in the subsequent wafer process.

Therefore, conventionally, liquid phase epitaxy was performed thickly, and the epitaxial surface was polished to be flat.

If the surface is flat, there will be no problems in adhesion of the photomask, positioning of automatic wire bonding, and the like.

(E) Problems to be solved by the invention However, by polishing the surface of the epitaxial thin film, the following new problems arise.

(1) The total thickness of the wafer surface and the thickness of the epitaxial layer are not uniform. Since the thickness is not uniform, the light emission output varies depending on the position on the wafer surface where the chip is attached when the diode is used.

(2) The surface carrier density is not uniform. This is because the thickness of the epitaxial layer is non-uniform. Therefore, in the case of a light emitting diode, the voltage Vf with respect to the reference current during light emission varies depending on the chip.

(3) Acceleration of deterioration due to stress on the PN junction. In the PN junction, P-type impurities and N-type impurities are doped in close proximity. Even if the lattice constants are not different, effective diameters of impurities are different. If a strong pressing force is applied here, impurities may deviate from the regular site. Then, impurity scattering increases and luminous efficiency also decreases.

(4) Increase in man-hours in device manufacturing process and increase in defective rate Since polishing is performed, man-hours increase. Since policing has success or failure, it becomes a factor of increasing the defect rate.

(F) Purpose Poor policing has a large negative effect. I want to omit policing.

It is sufficient to carry out liquid phase epitaxy which gives an epitaxial thin film which can be used for a wafer process as it is without polishing.

It is an object of the present invention to provide a vertical liquid phase epitaxial growth method capable of forming a thin film having a flat surface without showing a terrace pattern.

(G) Problem of top surface of carbon jig The present inventor studied a carbon jig for containing a wafer and a solution. It is unclear why the terrace pattern occurs, but the carbon jig may have some influence.

FIG. 2 shows an enlarged sectional view of the jig.

If there is a protrusion or the like on the upper surface 8 of the jig, this causes the wafer to be distorted and the light emission output is extremely reduced. Therefore, the applicant has already proposed that the upper surface should be flattened by removing the projections on the upper surface.

This is described in JP-A-62-223093 (published on S62.10.1).

This means that the protrusions deform the wafer when it softens.

This time, the problem is the back surface 7 of the jig.

(H) Method of the present invention The present inventor considered that the unevenness on the back surface of the jig had a strong correlation with the unevenness of the epitaxial film.

The wafer has a diameter of, for example, 3 inches or 2 inches.
However, since the jig is shallow, the solution 3 is extremely thin. 0.7m
It is about m.

The wafer faces the back surface of the carbon jig directly above. The epitaxial layer may have a thickness of several tens of μm. If the back surface of the carbon jig has unevenness, it may affect the flatness of the film of the epitaxial layer.
I thought.

Therefore, the flatness of the back surface of the jig immediately above the (carbon dish) jig used to prepare the epitaxial thin film shown in FIGS. 5 (a) and 5 (b) was measured. The results are shown in FIGS. 6 (a) and 6 (b). The horizontal axis is the diametrical position.
The vertical axis is the height of the surface. One scale is 10 μm. From peak to peak, there is a height difference of 10 μm even in a narrow range. There is a height difference of 20 μm as a whole.

Since the thickness of the solution is about 700 μm, the height difference of 20 μm is slight, but there may still be some correlation.

The inventor then decided to polish the back surface of the jig to make it flat.

The method of polishing is arbitrary. For example, # 3000 Al ₂ O ₃
The back surface 7 of the jig was polished using the above as abrasive grains. And
The flatness was set to 1 μm or less.

3 (a) and 3 (b) show the results of measuring the height of the back surface of such a jig. In this way, vertical liquid phase epitaxy is performed using a jig having a flatness of 1 μm or less.

FIG. 4 shows the results of measuring the flatness of the surface of a GaAs wafer by liquid phase epitaxial growth of a GaAs thin film. The vertical scale is 5 μm as in FIG. It can be seen that the surface roughness of the epitaxial thin film is 1 μm or less. Fourth
Comparing FIG. 5 with FIG. 5 shows that the flatness of the epitaxially grown thin film according to the method of the present invention is excellent.

After all, according to the method of the present invention, a large number of wafers placed in a shallow jig are stacked, placed in a solution in which a solute having a saturated concentration is dissolved, and gradually cooled to form a thin film on the wafer. In the vertical liquid phase epitaxial growth method for growing, the back surface of the jig is polished in advance to increase the flatness.

The unevenness on the back surface of the jig is initially 10 μm to 20 μm. This one
If the thickness is less than μm, the unevenness of the epitaxial thin film is 1 μm.
It becomes the following.

The flatness of the back surface of the jig is preferably 1 μm or less,
If the thickness is about 3 μm or less, the flatness of the epitaxial thin film could be set to 2 μm or less.

(K) Work In the vertical liquid phase epitaxial growth method, since the back surface of the jig was made flat, the flatness of the epitaxial surface was improved. Since the flatness is excellent, it is not necessary to polish the epitaxial layer thereafter.

As mentioned at the beginning, in liquid phase epitaxy, a terrace pattern appears by any method. Why does the terrace pattern appear? Not even that clear.

Therefore, in the vertical liquid phase epitaxial growth method, it is unknown why the epitaxial thin film becomes flat when the back surface of the carbon jig is flattened.

The present inventors presume as follows.

In liquid phase epitaxy, the difference in temperature between the solution and the wafer is extremely small. Even within the wafer plane, the temperatures are not exactly the same. There is a slight rise and fall in temperature.

Since it is oversaturated, crystal growth starts from the point of the minimum temperature in the wafer surface. There are several points of minimum temperature. Crystal growth starts independently from these points. It creates irregularities.

However, the minute temperature distribution in the wafer surface is strongly affected by the back surface of the carbon jig directly above it.

If the back surface of the carbon jig is uneven, a minute temperature distribution is generated on the wafer surface directly below the jig.
This creates a terrace.

When the back surface of the carbon jig is made flat as in the present invention, the temperature distribution does not occur on the wafer surface immediately below it. The temperature becomes uniform. For this reason, uniform film formation is performed subsequent to uniform nucleation.

This is conceivable, but this is a speculation. I don't know what I'm attached to.

However, all that is required is a clear causal relationship.

(C) Effect According to the present invention, the flatness of the liquid phase epitaxial growth thin film can be increased. Therefore, it is not necessary to polish the epitaxial growth surface.

Wafer processes can be performed without polishing. Can be used as-grown.

Therefore, the drawbacks associated with the policy can be solved. That is, due to the polishing, (1) non-uniform total thickness in the plane and epi thickness, resulting non-uniform emission output (2) non-uniform Vf during device fabrication due to non-uniform surface carrier concentration (3) PN Deterioration due to stress applied to the junction (4) It is possible to eliminate defects such as increased man-hours during device fabrication and increased defect rate.

[Brief description of drawings]

FIG. 1 is a principle diagram of a known vertical liquid phase epitaxial growth method developed by the present applicant. FIG. 2 is a sectional view of only the jig. FIG. 3 is a graph showing the results of measuring the flatness of the back surface of the jig when the back surface of the jig was polished, using a surface roughness meter. FIG. 4 is a graph showing the results of measuring the flatness of the surface of a GaAs / GaAs thin film epitaxially grown by the method of the present invention using a polished jig with a surface roughness meter. FIG. 5 is a graph showing the results of measuring the flatness of the surface of a GaAs / GaAs thin film epitaxially grown by using a jig without polishing, using a surface roughness meter. FIG. 6 is a graph showing the results of measuring the unevenness of a jig that is not polished with a surface roughness meter. 1 ... Wafer 2 ... Jig 3 ... Solution 4 ... Quartz cassette 5 ... Heater 7 ... Jig surface 8 ... Jig top surface

Claims (1)

[Claims] 1. A large number of dish-shaped jigs 2 each containing a semiconductor wafer 1 are vertically stacked, immersed in a high-temperature solution in which a raw material to be epitaxially grown is dissolved, and a saturated solution is placed in the dish-shaped jig 2. In the method of introducing liquid crystal No. 3 into contact with the surface of the semiconductor wafer 1 and lowering the temperature of the solution to grow a thin film on the semiconductor wafer 1 by liquid phase epitaxial growth, the flatness of the back surface is 3 μm or less. 2. A method for improving a surface during epi-growth in a liquid phase epitaxy method, characterized by using 2.