JPH0480920A - Reactive ion etching method - Google Patents

Abstract

PURPOSE:To make it possible to reduce the speed of etching performed on one of layers without reducing the speed of etching performed on other layers by a method wherein a high frequency power source is applied to the electrode on the side of a substrate, and inert gas is mixed into reaction gas. CONSTITUTION:The mixed gas 43, obtained by mixing inert gas into reaction gas, is introduced into a plasma chamber 44, high density plasma 49 is formed in a plasma chamber 44, and it is led out into an etching chamber 45 by the divergent magnetic field of a coil 42. Rf power 47 is applied to a sample stand 48, self-bias voltage is generated between the sample stand 48 and the plasma 49, the ions in the plasma 49 is accelerated toward the sample stand 48, they are made incident on the surface of the sample 46, and the etching is accelerated by sputtering the non-volatile reaction product on the surface. On the other hand, neutral active species reach the surface of the sample 46, a compound with the active species is formed, and the etching operation makes progress by the impulse of the active species and the ions.

Description

[Detailed Description of the Invention] [Industrial Application Field] The present invention relates to a reactive ion etching method, and more specifically, the present invention relates to a reactive ion etching method, and more specifically, to a method using electron cyclotron resonance (hereinafter referred to as ECR), which enables highly accurate etching. The present invention relates to the reactive ion etching method used.

[Prior Art] Conventionally, for example, III-V group compound semiconductors such as GaAs and InP, or AfLGaAs have been used.

For etching mixed crystal semiconductors such as InGaAs and InAuAs, etching is generally used. That is, etching using a mixture of hydrogen peroxide or a mixture of nitric acid and hydrogen peroxide is used.

However, since this type of wet etching is so-called isotropic etching in which etching is performed equally in each direction, undercutting or side etching occurs, and the etching rate has a significant dependence on crystal orientation. Vertical machining and fine and deep etching (so-called etching with a high asbestos ratio) were difficult.

Therefore, when performing etching with a high aspect ratio, reactive ion etching is performed that allows anisotropic etching (etching only in a specific direction). gas, halogen-based gas containing halogen elements, etc.) is turned into plasma, and highly chemically reactive active species (e.g., radicals) are generated along with ions in the plasma, resulting in physical etching by the ions and chemical etching by the above active species. Etching is performed using the following methods.

In particular, when this reactive etching is performed using a plasma generation source using an electron cyclotron (ECR) and applying radio frequency (RF) power to the substrate,
Can generate high-density plasma at low pressure,
Further, it has the feature that the ion energy incident on the substrate can be optimized.

That is, compared to the conventional reactive ion etching method using parallel plate type RF plasma excitation, this technology has the characteristic that a high etching rate can be obtained with low ion energy. 10
x 10-'Torr, which is a low pressure of 1/10 to 1/100 of that of the normal parallel plate reactive ion etching method. Therefore, nonvolatile reaction products are less likely to remain than in the case of other methods, and a high etching rate can be obtained.

Therefore, for example, for GaAs, a high etching rate of 4000 etching per minute or more can be obtained. Also, In
InGaAs, including.

For InAuAs and InP, an etching rate of about 400 people/min can be obtained despite the low vapor pressure of In chloride.

However, the etching method using this technique has the following problems, for example.

The problems are described in detail below.

In heterojunction field effect transistors and heterobipolar transistors, which have been actively researched and developed in recent years, in order to reduce the ohmic contact resistance between the GaAs layer and the Aj2GaAs layer, for example, as shown in FIG. Then, InGaA is formed on the GaAs layer 32.
An s layer 31 is formed. When separating such a layer structure by etching, the etching rate of InGaAs is about 1710 that of GaAs as described above, so the I
When attempting to etch the nGaAs layer 31, Ga
Excessive etching of the As layer 32 will occur.

That is, in the layer structure shown in FIG.
Assuming that the thickness of the s layer 31 is 1000 layers and the thickness of the GaAs layer 32 is 1000 layers, as mentioned above, InGaA
Since the etching speed of s is 400 people/min, I n
The etching time required to etch the entire GaAs layer 31 is 2 minutes and 30 seconds.

However, when considering the in-plane distribution of layer thickness, etc., it is certain that I n
In order to completely remove the GaAs layer 31, the film thickness must be reduced by 10
% (15 seconds) is usually added. At this time, the surface of the GaAs layer 32 is etched for 15 seconds in areas where the InGaAs layer 31 has already been etched (that is, where the surface of the GaAs layer 32 is exposed). Since the etching rate of GaAs is 4000 etching per minute, the amount of etching of the GaAs layer 32 during this over-etching is 1000 etching, resulting in a problem that the GaAs layer 32 is completely etched. If the thickness of the GaAs layer 32 is less than 1000 Å, there is a risk that the layer below the GaAs layer 32 (substrate 33 in FIG. 3) may also be etched depending on the material.

Therefore, in order to control the amount of etching of the GaAs layer 32 with high precision, it is necessary to reduce the etching rate of GaAs. Moreover, the important thing is that InGaAs
This means that it is necessary to reduce the etching rate of GaAs without reducing the etching rate of GaAs.

However, the technology to achieve this does not currently exist.

Note that this problem does not only occur between InGaAs and GaAs, but also when a plurality of thin films having different compositions are formed on a substrate, the etching rates of the plurality of thin films are different, and one of the thin films has different etching rates. This often occurs when it is desired to etch only a layer.

[Problems to be Solved by the Invention] An object of the present invention is to provide a reactive ion etching method that can reduce the etching rate of other films without reducing the etching rate of the film to be etched. shall be.

[Means for Solving the Problems] The gist of the present invention for solving the above problems is to provide a reactive ion etching method for etching a thin film formed on a substrate by converting a reactive gas into plasma by electron cyclotron resonance. In the following, the effects of the present invention will be described. Based on the knowledge obtained when making the invention, it will be explained along with the detailed structure.

Note that, here, InGaAs and GaAs will be explained as examples.

First, the present inventors discovered that InGaA
The idea was that the reason why the etching rates of S and GaAs are different is that the factors that control the etching rates are different.

However, the rate-determining factor when applying RF power using ECR has not been elucidated in the past.

Therefore, the present inventor first varied the RF power applied to the substrate-side electrode and investigated the effect of each on the etching rate as a clue to determining the rate-determining factor. In addition, in this case, 100% as reactive gas
Chlorine gas was used.

An example of the results is shown in FIG.

As shown in FIG. 2, in the case of InGaAs, RF[
As the power increases, the etching rate (line 22) increases, but in the case of GaAs, even if the RF power increases, the etching rate (line 21) saturates or decreases slightly above 50 W. Ta.

By the way, when RF power is applied to the substrate-side electrode, a self-bias voltage is generated on the substrate surface.

Since the ions in the plasma have charges, they are accelerated by this self-bias voltage, gain a large amount of energy, and enter the substrate surface. Increasing the RF power also increases the self-bias voltage and therefore increases the energy of the ions incident on the substrate surface. On the other hand, since active species are neutral, the incident energy is not affected by the self-bias voltage.

Therefore, as shown in FIG. 2, the etching rate of GaAs (line 21) does not increase in correlation even if the RF power is increased, which means that the etching rate of GaAs (line 21) depends on the ion energy. It is considered that the rate of supply of active species (for example, chlorine and chlorine radical cu', etc. when chlorine gas is used as the reactive gas) is not sufficient.

In contrast, the etching rate of InGaAs (line 22
) increases with RF power, it is thought that the etching rate (line 22) is controlled by ion bombardment, and in the case of InGaAs, it is thought to be rate-limiting of ion supply.

We have considered why the factors that determine the etching rates of InGaAs and GaAs are different in this way, and the reason is believed to be as follows.

That is, the reaction products generated during etching of GaAs are a compound of Ga and the constituent elements of the reactive gas and As.
and a constituent element of the reactive gas (a chlorine compound if the reactive gas is, for example, chlorine gas). However, since the vapor pressure of these reaction products is high, it is thought that they easily evaporate from the substrate surface and etching progresses. Therefore, it is believed that what primarily determines the etching rate of GaAs (line 21) is the number of active species (chlorine, chlorine radicals, etc.) incident on the surface.

On the other hand, the reaction products generated during etching of InGaAs are compounds of Ga and the constituent elements of the reactive gas;
They are considered to be a compound of As and a constituent element of the reactive gas, and a compound of In and a constituent element of the reactive gas (a chlorine compound if the reactive gas is, for example, chlorine gas). Of these, In
Since the compound has a low vapor pressure, it remains as a non-volatile product, covering the substrate surface and inhibiting etching. For this reason, for example, the etching rate of InGaAs (
It is thought that the line 22) is about 1710, which is smaller than that of GaAs.

It is thought that the detachment of the In compound covering the surface occurs due to ion bombardment with the etched surface at room temperature, and therefore the etching rate of InGaAs (line 22
) is thought to be primarily determined by the number or energy of ions incident on the surface.

Based on the above knowledge, we have earnestly searched for solutions to the problems of the conventional techniques and have found that the above problems can be solved by mixing an inert gas with the reactive gas.

Generally speaking, when an inert gas is mixed with a reactive gas, the partial pressure of the reactive gas naturally decreases, and even if the etching rate of GaAs decreases, it is expected that the etching rate of InGaAs will also decrease. be done. but,
Contrary to such general expectations, the present inventors have discovered that only the etching rate of GaAs decreases, but the etching rate of InGaAs does not decrease, and has led to the present invention.

In GaAs etching rate is not reduced.
The reason why only the etching speed of s decreases is considered in conjunction with the above findings. It can be inferred as follows.

That is, when a mixed gas containing a reactive gas and an inert gas is used, the number of active species in the plasma is reduced, and the number of incident active species radicals is also reduced. However, inert gas ions also occur in the plasma, and the overall number of ions does not decrease after all. As mentioned above, InGa
In the case of As, the rate of ion supply is considered to be the limiting factor, so unless the number of incident ions decreases, the etching rate of InGaAs will not decrease much.

In contrast, the etching rate of GaAs decreases in proportion to the concentration of the reactive gas. Therefore, it is thought that this is because when the reactive gas is diluted with an inert gas, only the etching rate of GaAs decreases, and the etching rate of InGaAs does not change.

Note that the above explanation uses InGaAs and GaA as thin films.
Although the explanation takes I n as an example, the present invention simply provides I n
It is applicable not only when two GaAs layers are formed, but also when a layer that generates a reaction product with a difference in vapor pressure is formed (in such cases, the conventional method It is clear from the above description that it can be generally applied to cases where a difference occurs.

For example, InP/GaAs, InGaAs/AJIGa
It can also be applied to etching thin films made of compound semiconductors such as As, InP/AfGaAs, mixed crystal semiconductors, and other thin films.

In the present invention, the reactive gas may be appropriately determined depending on the material of the thin film to be etched, and for example, halogen gas may be used. Among the halogen gases, chlorine gas, bromine gas, and iodine gas are preferred. Moreover, two or more kinds of these may be mixed.

On the other hand, as an inert gas to be mixed with the above reactive gas,
For example, helium gas, neon gas, argon gas, krypton gas, xenon gas, nitrogen gas (in this specification, nitrogen gas is also included in the inert gas), etc. Furthermore, using chlorine gas as the reactive gas, It is preferable to use argon gas as the inert gas.

The mixing ratio of reactive gas and inert gas is determined by determining the relationship between reactive gas concentration and etching rate as shown in Figure 1 based on the specific combination of reactive gas and inert gas. The etching rate can be determined on a case-by-case basis depending on the allowable amount of etching of the film that does not produce reaction products, and the etching rate can be varied over a wide range by changing the mixing ratio (reactive gas concentration). can.

The relationship between the etching rate and the concentration of the reactive gas for each type of thin film material is determined as shown in Figure 1, and if the concentration is selected to be less than or equal to the concentration of the reactive gas at the point where each line intersects, the etching rate is equal. Alternatively, lower etching rates can be obtained. For example, the above-mentioned InGa
In the example of As and GaAs, straight lines 11 and 12 intersect at a reactive gas concentration of about 10%, so if the reactive gas concentration is set to below about 10%, the etching rate of GaAs will be lower than that of InGaAs. The etching speed can be lower than that.

[Example code] The present invention will be explained in detail below.

(7/7S1 Example) FIG. 4 shows a conceptual diagram of the ECR etching apparatus used in this example.

Well, let me explain this device.

In this apparatus, a mixed gas 43 obtained by mixing a reactive gas with an inert gas is introduced into a plasma chamber 44, and a high-density plasma 49 is formed in the plasma chamber 44.

This plasma 49 is drawn into the etching chamber 45 by the divergent magnetic field of the coil 42. In the etching chamber 45, RF power 47 is applied to the sample stage 48, and the sample stage 48
According to the pressure near 8, the sample stage 48 and the plasma 49
A self-bias voltage is generated between Due to this self-bias voltage, ions in the plasma 49 are accelerated toward the sample stage 48 and are incident on the surface of the sample (substrate) 46, sputtering nonvolatile reaction products on the surface, or
Etching is promoted by applying energy equivalent to the heat of vaporization to the reaction product and evaporating it.

On the other hand, neutral active species (for example, chlorine and chlorine radicals) reach the surface of the sample 46 with an areal density according to the etching pressure and form a compound with the active species.

As described above, in the apparatus shown in FIG. 4, etching progresses due to the impact of active species and ions.

An actual etching test was conducted using this device.

Prior to etching the layered structure shown in FIG. 3, the dependence of the etching rate on the reactive gas concentration for the materials constituting each layer shown in FIG. 3 was determined.

The etching conditions at that time are as follows.

Layer constituent material: GaAs, InGaAsRF power...
・50W RF frequency...13.56MHz Substrate temperature...Room temperature Reactive gas...Chlorine gas Inert gas...Argon gas Reactive gas concentration...0 to 100% (partial pressure ratio) However, no Active gas concentration (%) = (reactive gas)/(inert gas + reactive gas).

As a result of this test, the results shown in FIG. 1 were obtained regarding the relationship between the etching rate of GaAs and InGaAs and the dependence on chlorine gas concentration.

As shown in Figure 1, the etching rate of GaAs (
Line 11) indicates that at 100% chlorine gas concentration, 4000 people/
minutes, but gradually decreased as the chlorine gas concentration decreased, and for example, at about 10%, it was 400 people/minute.

On the other hand, the etching rate of InGaAs (line 12) increases as the chlorine gas concentration increases from 0 to about 5%, but when the concentration exceeds 5%, the etching rate begins to saturate at 400 ppm/min.

That is, from 5% to 100%, there was almost no change at 400 people/min.

Then, line 11 (etching rate of GaAs) and line 12
(InGaAs etching rate) intersects at a point where the chlorine gas concentration is approximately 10% (note that the etching rate for both at the intersection is 400 people/min), therefore, the chlorine gas concentration is less than 10% ( Preferably 5-10%)
It was confirmed that the etching rate of GaAs can be made lower than the etching rate of InGaAS by performing etching as follows.

Next, by setting the chlorine gas concentration to 10%, a layer structure as shown in FIG. 3, that is, an InGaAs layer 31 (thickness:
A layered structure in which a GaAs layer 32 (thickness: 1,000 layers) and a GaAs layer 32 (thickness: 1,000 layers) were sequentially laminated on a substrate 33 was etched.

Note that the etching conditions other than the chlorine gas concentration were the same as those used when FIG. 1 was obtained.

First, the InGaAs layer 31 was etched for 2 minutes and 30 seconds. At this time, a part of the GaAs layer 32 was exposed on the surface.

Then, for 15 seconds (the thickness of the InGaAs layer 31 is 10%).
(equivalent to ).

When the amount of etching of the GaAs layer 32 during this over-etching was investigated, the amount of etching was suppressed to 100 layers at most. That is, it was reduced to about 1/1O compared to the conventional one.

In this way, highly accurate etching was achieved in this example.

(Second Example) In this example, instead of the GaAs layer 32 in the first example,
The layer 32 was an A, Q GaAs layer, and the layer 31 was an <InGaAs layer as in the first embodiment.

In this example, as in the first example, first, Al2G
The relationship between the aAs etching rate, the InGaAs etching rate, and the reactive gas concentration was determined. Note that the etching conditions at this time were the same as those used when obtaining FIG. 1 in the first example, except for the layer-constituting material.

As a result, for Al1GaAs, the chlorine concentration was 100
In terms of %, the etching speed is 4000 people/min and Ga
Although it was the same as As, the etching rate was 500λ/min at a chlorine gas concentration of about 10%.

On the other hand, the same results as in Example 1 were obtained for InGaAs.

Then, the line showing the etching rate of AρGaAs and the line I
A line showing the etching rate of nGaAs (line 12 in Figure 1)
) is the point at which the chlorine gas concentration is approximately 8% (etching rate 40%).
0 people/min).

Next, by setting the chlorine gas concentration to 8%, the layer structure as shown in FIG.
00 layers) and Al2GaAs layer 32 (thickness: 1000 layers)
Etching was performed on a laminated structure consisting of: Note that the etching conditions other than the chlorine gas concentration were the same as those in the first embodiment.

As a result, the same results as in the first example were obtained.

That is, the AllGaAs layer 3 during overetching
The amount of etching for 2 was limited to 100 people at most.

In this way, highly accurate etching was achieved in this example as well.

(Third Example) In this example, W: 31 is an InP layer, and layer 32 is an A 12
G a As W7.

In this example, as in the first example, first, Al1G
The relationship between the aAs etching rate, the InP etching rate, and the reactive gas concentration was determined.

Note that the etching conditions at this time were the same as those used when obtaining FIG. 1 in the first example, except for the layer-constituting material.

As a result, for Al1GaAs, the same results as in the second example were obtained (that is, at a chlorine gas concentration of 100%, the etching rate was 4000 people/min, and at a chlorine gas concentration of 10%, the etching rate was 500 people/min). people/minute).

On the other hand, for InP, in the case of InGaAs (the first
The same result as line 12) in the figure was obtained.

Then, a line indicating the etching rate of AfGaAs and I
The line indicating the nP etching rate is at a chlorine gas concentration of approximately 8
% point (etching speed 400 people/min).

Next, with the chlorine gas concentration set at 8%, the layer structure as shown in FIG.
A laminated structure consisting of 32 (thickness: 1000 layers) and Al1GaAs layer 32 was etched. Note that the other etching conditions were the same as those in the first embodiment.

As a result, the same results as in the first example were obtained.

That is, the amount of etching of the AffiGaAs layer 32 during over-etching was suppressed to 100 at most.

In this way, highly accurate etching was achieved in this example as well.

(Fourth Example) In this example, the layer constituent materials were the same as those in the first example. However, in this example, nitrogen gas was used instead of argon gas as the inert gas.

In this example, as in the first example, first, the relationship between the etching rate of GaAs, the etching rate of InGaAs, and the reactive gas concentration was determined. Note that the etching conditions at this time were the same as those used for obtaining FIG. 1 in the first embodiment, except for the inert gas.

As a result, the etching rate of GaAs was 4000 etching/min at 100% chlorine gas concentration, gradually decreased as the chlorine gas concentration decreased, and at 10%, 400 etching/min.
It was a minute. That is, the same results as in the first example were obtained. In other words, there was no difference in the etching rate of GaAs whether argon gas or nitrogen gas was used as the inert gas.

On the other hand, the etching rate of InGaAs increased as the chlorine gas concentration increased from 0 to 5%, but from 5% the etching rate reached saturation at 300 ppm/min. That is, 5% to 100
% remained almost unchanged at 300 people/minute. As a result, the etching rate at saturation was lower than in Example 1 in which argon gas was used as the inert gas.

Then, the etching rate line of GaAs and InGaA
The etching speed line of s intersected at a point where the chlorine gas concentration was 8% (note that the etching speed of both at the intersection was 300 people/min).

Therefore, the chlorine gas concentration should be 8% or less (preferably 5 to 5%).
It was confirmed that the etching rate of GaAs can be made lower than the etching rate of InGaAs by performing etching at a concentration of 8%).

Next, by setting the chlorine concentration to 8%, a layer structure as shown in FIG. 3, that is, an InGaAs layer 31 (thickness: 1000
A laminated structure consisting of a GaAs layer 32 (thickness: 1,000 layers) and a GaAs layer 32 (thickness: 1,000 layers) was etched. Note that the etching conditions other than the type of inert gas and the chlorine gas concentration were the same as those in the first embodiment.

First, the InGaAs layer 31 was etched for 3 minutes and 20 seconds. At this time, a part of the GaAs layer 32 was exposed on the surface.

Then, for 20 seconds (the thickness of the InGaAs layer 31 is 10%)
(equivalent to) was over-etched.

When the amount of etching of the GaAs layer 32 during this over-etching was investigated, the amount of etching was suppressed to 100 layers at most.

In this way, highly accurate etching was achieved in this example.

In the above example, the reactive gas concentration was set to 8% or 10% as an example in which the etching rate of the two layers should be the same.
Although the case has been described above, there is no need to be particular about such a reactive gas concentration, and it goes without saying that the reactive gas concentration can be appropriately selected within a range of less than 100%.

What can be said from the first to fourth examples is that even if the layer constituent materials and the type of inert gas are changed,
For each layer-constituting material, the relationship between reactive gas concentration and etching rate as shown in Figure 1 is determined, and etching is performed at the reactive gas concentration at the intersection of the lines corresponding to the combination of layer-constituting materials. 'The lower layer of the layer structure shown in Figure M3 (
32) can be kept constant (100 people in the first to fourth embodiments).

[Effect of the invention] According to the present invention, for example, the following effects can be obtained.

According to the conventional method, when etching two or more layers with different etching speeds, it is possible to reduce the etching speed of the other layer to an arbitrary speed without reducing the etching speed of one layer. If the layer has a laminated structure, the etching accuracy of the laminated structure can be improved and the limitations on processing of the laminated structure are reduced.

[Brief explanation of drawings]

FIG. 1 is a graph showing the dependence of etching rate on chlorine concentration when chlorine gas is diluted with Ar. Figure 2 shows GaAs and In by 100% chlorine gas.
3 is a graph showing the dependence of GaAs etching rate on RF power. FIG. 3 is an example of a layer structure of a semiconductor to be etched according to an embodiment of the present invention. ! FIG. 4 is a conceptual diagram of an ECR etching apparatus used in an embodiment of the present invention. (Explanation of symbols) 11.21...Etching rate of GaAs 12.22
...I nGaAs etching rate 31-I n
G a As layer (InP layer) 32-・-G a A
S layer (AuGaAs layer) 33...Base (substrate) 41...Microwave 42...Coil 43...Mixed gas 44...Plasma chamber 45...Etching chamber 46...Sample (substrate) ) 47...RF power 48...Sample stage 49...Plasma diagram C12/(C12+Ar) (%) (Partial pressure ratio) 11...GaAs 12-1nGaAs Diagram 2+ ・=GaAs 22 ・=InGaAs Fig. 33... Base 4 46... Sample (substrate) 47... RF power source 48... Sample stage 49... Plasma

Claims (1)

[Claims] (1) In a reactive ion etching method in which a thin film formed on a substrate is etched by converting a reactive gas into plasma through electron cyclotron resonance, high-frequency power is applied to the substrate-side electrode and at the same time the reactive gas is A reactive ion etching method characterized by mixing an active gas.