JP3424371B2 - Method of manufacturing acceleration sensor - Google Patents

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an acceleration sensor,
In particular, so-called 3 having sensitivity in the x-axis, y-axis, and z-axis directions
The present invention relates to an axial acceleration sensor.

[0002]

2. Description of the Related Art Generally, as an acceleration sensor, as shown in FIG. 4, there are a cantilever type and a double-supported beam type. In the figure, (a) is an explanatory view showing a schematic structure of a cantilever type acceleration sensor and (b) a double-ended beam type acceleration sensor. Figure 4 (a)
Reference numeral 1 denotes a cantilever, one end of which is fixed to a vertical support wall 2, and the cantilever 1 is fixed so as to project from the support wall 2 in a substantially horizontal direction. A weight 3 having a substantially rectangular shape in a side view is attached to the lower surface side of the other end of the cantilever 1.
In FIG. 4 (b), 4 is a double-supported beam, both ends of which are fixed to vertical support walls 5 and 6, respectively.
Is fixed substantially horizontally by the support walls 5 and 6. A weight 7 having a substantially rectangular shape in a side view is attached to the lower surface side of the central portion of the both-supported beam 4. As a 3-axis acceleration sensor,
Among them, the double-supported beam method as shown in FIG. 4 (b) is adopted. In FIG. 4B, the length of the doubly supported beam 4 is L, the width is W, the thickness is h, the Young's modulus is E, the horizontal length of the weight 7 is 2c, the mass is M, and the gravitational acceleration is g. Then, as follows
To become.

That is , in order to improve the sensitivity D,
Increase the mass M of the weight 7 or increase the length L of the doubly supported beam 4.
It suffices to increase (the distance from the support walls 5, 6 to the weight 7). However, when a silicon substrate is used as the weight 7, the density of silicon is relatively small. Therefore, in order to improve the sensitivity, the horizontal length 2c of the weight 7 is lengthened to increase the mass M of the weight 7. Let or doubly supported beam 4
It was necessary to increase the length L of. However, since the chip size is increased, an acceleration sensor having a heavy weight and a double-supported beam structure having a long beam without increasing the chip size is desired. An example of an acceleration sensor that realizes these is an acceleration sensor described in Japanese Patent Application No. 6-25677.

FIG. 5 shows a perspective view of the acceleration sensor described in Japanese Patent Application No. 6-25677. In FIG. 5, (a) is a perspective view seen obliquely from above, and (b) is a perspective view seen obliquely from below. In FIGS. 5A and 5B, 8 is a substantially flat plate-shaped support portion including a hollow portion 8a having a substantially rectangular shape in plan view, and 9 is formed in an opening on the surface side of the hollow portion 8a of the support portion 8. It is a diaphragm-shaped flexure that corresponds to a doubly supported beam. 10
Is a substantially quadrangular prism-shaped weight portion that is disposed inside the hollow portion 8a and is fixed to the bending portion 9.

In the acceleration sensor shown in FIG. 5, a part of the joint between the bending portion 9 and the weight portion 10 is removed, and the bending portion 9 and the weight portion 10 are joined only at the substantially central portion of the upper surface of the weight portion 10. Is. With such a configuration, the length of the doubly supported beam is almost unchanged as compared with the shape in which a part of the joint portion between the bending portion 9 and the weight portion 10 is not removed, and the shape and mass of the weight portion 10 are hardly changed. The distance from the side surface of the hollow portion 8a to the joint portion between the bending portion 9 and the weight portion 10 can be secured, and the above-mentioned demand for improved sensitivity is satisfied.

The acceleration sensor shown in FIG. 5 has an x-axis, a y-axis,
The bending portion 9 is deformed with respect to the acceleration in each direction of the z axis.
FIG. 6A is a cross-sectional view showing a state where acceleration in the z-axis direction perpendicular to the surface of the substantially flat plate-shaped support portion 8 acts on the acceleration sensor. In this case, the beam around the weight portion 10 is substantially formed. Similarly,
The bending portion 9 and the weight portion 10 bend substantially symmetrically around the joint portion. FIG. 6B is a cross-sectional view showing a state where acceleration in the x-axis direction parallel to the surface of the substantially flat support portion 8 acts on the acceleration sensor, and the deflection of the beam around the weight portion 10 is
It depends on the position in the x-axis direction. By grasping such a bending state of the beam, the acceleration sensor shown in FIG. 5 functions as a highly sensitive triaxial acceleration sensor.

[0007]

As described above, the acceleration sensor described in Japanese Patent Application No. 6-25677 has an excellent structure that satisfies the above requirements, but has the following problems in manufacturing. Have That is, a method has been used in which a sacrifice layer made of polysilicon or the like is formed in a part (removed portion) of a joint between a beam and a weight portion, the weight portion is formed, and then the sacrifice layer is removed by etching. However, according to this method, stable conditions cannot be established in the steps of flattening polysilicon for forming a sacrificial layer, joining polysilicon and silicon, etc., and a sufficient yield cannot be obtained at present. Is.

As a method for solving the above problems,
There is USP4882933. In the manufacturing method described in USP4882933, the first conductivity type semiconductor substrate in which the reverse conductivity type semiconductor layers are stacked is bonded to a substrate that becomes a weight, and then the first conductivity type semiconductor substrate is removed by etching to form a reverse conductivity type semiconductor layer. In this method, since the first conductivity type semiconductor substrate and the substrate are bonded to each other on the wafer and etched, the semiconductor element is formed on the opposite conductivity type semiconductor layer to be the beam portion, Problems such as warpage of the beam have occurred.

The present invention has been made in view of the above problems,
An object of the invention is to provide a structure of an acceleration sensor and a manufacturing method thereof, which can improve sensitivity and yield.

[0010]

In order to achieve the above object, in a method of manufacturing an acceleration sensor according to a first aspect of the present invention, a substrate, a beam formed on the substrate, and one surface of the beam are fixed. A method of manufacturing an acceleration sensor, comprising: a weight portion; and a surface of a first conductivity type semiconductor substrate that is the substrate.
A step of growing a second conductivity type semiconductor layer to be the beam
And the second conductive layer from the back surface of the first conductive type semiconductor substrate.
-Type semiconductor layer is electrolytically etched to reach the cavity
A step of forming a sacrificial layer around the connection between the beam and the weight portion on the back surface of the beam above the cavity, and the sacrifice around the connection between the beam and the weight portion. A step of forming a metal thin film that becomes a part of the weight portion on the surface and side surfaces of the layer, a step of plating the surface of the metal thin film with a metal, and a step of removing the sacrificial layer. It is what

A method of manufacturing an acceleration sensor according to a second aspect is the method of manufacturing an acceleration sensor according to the first aspect, wherein the sacrifice is performed.
The sacrificial layer is composed of a resist .

[0012]

[0013]

[0014]

[0015]

In the method of manufacturing the acceleration sensor according to the first or second aspect of the present invention, the weight supported by the beam is formed by metal plating without using the direct bonding of the wafers, so that a highly sensitive triaxial acceleration is obtained. This is to realize a sensor structure. Further, in the conventional technique, the surface on which the sacrificial layer is formed cannot be flattened, and it is difficult to polish the attached wafer to a diaphragm that becomes a beam. However, according to the method of manufacturing an acceleration sensor of the present invention. , Does not require planarization of the sacrificial layer. Specifically, instead of the conventional bare silicon wafer,
Using a substrate (wafer) in which the second conductivity type semiconductor layer is formed on the surface of the first conductivity type semiconductor substrate, a cavity reaching the second conductivity type semiconductor layer is formed by electrolytic etching from the back side of the substrate, and inside the cavity. The weight supported by the beam is formed by forming the weight portion by metal plating. As described above, in the beam formation, which has been difficult in the past, by using the second conductivity type semiconductor layer for the beam, it is possible to leave only the second conductivity type semiconductor layer epitaxially grown by electrolytic etching, and to make the beam uniform. Also, it can be formed with good reproducibility. Further, since it is formed by one wafer, the cost problem is solved. Furthermore, by forming the weight portion by metal plating and using a material having a larger specific gravity than silicon, it is possible to form a highly sensitive acceleration sensor in the case of the weight portion having the same size.

[0016]

[0017]

[0018]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS The method of manufacturing an acceleration sensor of the present invention, which will be described below, is formed from one substrate (wafer) and has a low concentration on the surface of a first conductivity type silicon wafer (first conductivity type semiconductor substrate). As the impurities, a substrate on which the second conductivity type semiconductor layer is grown by the thickness of the beam is electrolytically etched to form a cavity reaching from the back surface to the second conductivity type semiconductor layer which is the low concentration impurity layer, and the cavity A beam is formed above the base plate and a weight portion is formed in the cavity by metal plating, thereby forming a structure in which the weight portion is supported only by the beam.

An embodiment of a method of manufacturing an acceleration sensor according to the present invention will be described with reference to FIG. FIG. 1 is a sectional view showing a manufacturing process. (A) is a sectional view showing an epitaxial growth step, in which a second conductivity type semiconductor layer 12 is formed on a first conductivity type semiconductor substrate 11 (P type silicon substrate) which is a substrate.
A substrate 13 is formed by epitaxially growing (N-type single crystal silicon) to a thickness of about 10 μm. Since the second conductivity type semiconductor layer 12 (N-type single crystal silicon) serves as a beam that supports the weight portion, when changing the thickness of the beam, the thickness of the second conductivity type semiconductor layer 12 should be changed. Good.

Next, as shown in (b), a resist mask 14 to be a piezoresistive mask is formed. However, this mask is not limited to the resist mask, and a thin film such as a silicon oxide film may be used as the mask.

Next, as shown in (c), a piezoresistor 15 is formed at a predetermined location of the second conductivity type semiconductor layer 12. An impurity resistor is used as the piezoresistor 15. The impurity concentration for zeroing the temperature characteristic of the sensitivity of the piezoresistor 15 is about 10 × (18) / cm ³ and about 10 × (20) / c in surface concentration.
is a m ^3, the resistance value, from the viewpoint of sensitivity, approximately 10 × (18) pieces / cm ³
Is desirable. In reality, boron (B) or the like is formed by ion implantation on the second conductivity type semiconductor layer 12. It may be formed not only by ion implantation but also by diffusion.

Next, as shown in (d), the piezoresistor 1
A silicon oxide film 16 having an opening at the upper portion of 5 is formed, and a metal wiring 17 that contacts the piezoresistor 15 is formed at the opening. A wire material such as gold is desirable for the metal wiring 17. It is formed by sputtering or the like and patterned by ion milling, RIE or the like. Patterning may be performed by wet etching with aqua regia.

Further, the step (e) is a step of forming an etching mask when the cavity on the back surface of the substrate 13 is formed by electrolytic etching. Silicon nitride films (Si ₃ N ₄ films) 8a and 8b are formed on the front and back surfaces of the substrate 13 to a thickness of 1000 Å. At this time, in order to protect the metal wiring 17 on the front surface side of the substrate 13 during etching, the silicon nitride film 8a is also formed on the front surface side of the substrate 13. And
The silicon nitride film 8b formed in the cavity forming portion on the back surface of the substrate 13 is removed to expose the back surface of the first conductivity type semiconductor substrate 11. The etching of the silicon nitride film 8b is performed by ion milling, plasma etching, or
Etch with hot phosphoric acid solution.

Next, as shown in (f), the substrate 13 (first conductivity type semiconductor substrate 11) is dug out from the back surface with a silicon etching solution such as KOH to remove the second conductivity type semiconductor layer 1.
Stop the etching at 2. Here, selectively the second
In order to leave the conductivity type semiconductor layer 12, etching is performed using electrolytic etching using the pn junction as an etch stopper.

A method of electrolytic etching will be described with reference to FIG. However, illustration of masks and the like is omitted. Figure 2
As shown in, the N-type second conductivity type semiconductor layer 12 to be a beam
Is applied by the voltage source 18 so as to be positive with respect to the P-type first conductivity type semiconductor substrate 11. This allows
Since a reverse voltage is applied to the pn junction, almost no current flows. In this state, the substrate 13 is treated with a KOH aqueous solution 20.
insert. When the P-type silicon forming the first conductivity type semiconductor substrate 11 is etched and reaches the N-type second conductivity type semiconductor layer 12, a voltage is directly applied to the N-type silicon, so that a passivation film is formed and etching is stopped. Becomes Further, the piezoresistor is not etched because it is a P ⁺ layer which is doped with impurities such as boron at a high concentration and serves as an etch stop layer. As the etching solution, KOH aqueous solution or CsOH aqueous solution is used. At the time of etching, it is necessary to protect the metal wiring 17 formed on the surface of the substrate 13, but the metal wiring 17 can be protected by leaving the silicon nitride film 8a in the previous step not removed. . Instead of forming a protective film, the jig 10
There is also a method in which the above is applied to the surface of the substrate 13 and the etching is performed without touching the etching solution. Through the etching process described above, the cavity 21 and the beam 22 are formed as shown in FIG. Then, the silicon nitride films 8a and 8b are removed.

Next, as shown in FIG. 1G, a resist pattern 23 is formed around the connecting portion between the beam 22 and the weight portion. The resist pattern 23 becomes a sacrificial layer that is removed after forming the weight portion. Resist pattern 23
As the resist constituting the above, a thick coating resist having a high viscosity, which can be applied to a thickness of about 10 μm, is used, and the entire surface is applied to the back surface of the substrate 13, leaving only the joint portion between the weight portion and the beam 22. It is general that poly-Si, SiO ₂ is deposited on the sacrificial layer by low pressure CVD, plasma CVD or the like, and is removed by etching in a later step. In this method, the sacrificial layer is partially formed in an island shape in the substrate. In order to form it, a so-called photolithography process of forming a resist pattern and leaving only a sacrificial layer portion by etching is necessary.
In the present invention, the resist pattern 23 is used instead of using the thin film as the sacrificial layer. Therefore, the film forming process which has been conventionally required to form the sacrificial layer is unnecessary.

Further, since chemical etching is conventionally used to remove the sacrificial layer, the effect on the metal wiring 17 and the like cannot be ignored, but when a resist is used as the sacrificial layer, development or nitric acid-based etching is performed. Since it can be easily removed by the resist stripper and has no effect on the metal wiring 17 and the like, the yield of the device can be improved.

Next, as shown in (h), a metal thin film 24 for the frame of the weight portion is formed in the cavity portion 21 of the substrate 13. As the metal thin film 24, aluminum or chromium is formed by a vapor deposition apparatus of about 1000 to 2000 liters.

Further, as shown in (i), the metal thin film 2
The surface of 4 is plated with metal to form the weight portion 25.
An advantage of forming the weight portion 25 by metal plating is that a heavy weight portion can be formed by using a material having a larger specific gravity than silicon as compared with the case where the weight portion of the same size is formed of silicon. In the acceleration sensor, the sensitivity is determined by how heavy the weight portion is and how thin the beam can be made. Therefore, according to this method, it is possible to form a high-sensitivity chip having a conventional chip size.

The metal coating method includes 1) electroplating method and 2)
Although there are melt immersion plating method, 3) metal spraying method, etc., electroplating method is preferable. In this electroplating method, the metal thin film 24 formed in the previous step is used as a cathode to immerse it in an aqueous solution of a metal salt for electroplating. When electrolysis is performed using a metal having good corrosion resistance as an anode, only the portion of the cathode where the metal thin film 24 is formed is plated with metal. Ni, Cr, Cd, Sn, Zn, brass, etc. are used as the material of this metal plating. Since the metal wiring 17 and the like are formed on the surface of the substrate 13, it is desirable to protect the surface with a resist or the like so that plating is not formed.

Finally, as shown in (j), the sacrificial layer resist pattern 23 is removed. As previously mentioned,
Since a resist was used as the material for forming the sacrificial layer,
There is no need to remove it by etching. It can be removed with a developing solution or a nitric acid-based resist stripping solution. By this step, the weight portion 25 made of metal plating is separated from the peripheral portion of the first conductivity type semiconductor substrate 11 made of silicon and is supported only by the beam 22.

A reference example of the method for manufacturing a semiconductor device of the present invention will be described with reference to FIG. The acceleration sensor of this reference example is formed by joining a substrate and a first conductivity type semiconductor substrate. The substrate is a substrate serving as a weight portion, and the first conductivity type semiconductor substrate is a substrate serving as a beam portion. The substrate and the first conductivity type semiconductor substrate are bonded together by direct bonding, the first conductivity type semiconductor substrate is etched into a diaphragm shape by electrolytic etching or anisotropic etching, and the substrate has a weight portion by anisotropic etching. A structure separated from the surrounding silicon substrate portion and supported by the beam portion is created.

First, as shown in (a), a silicon nitride film 27 (Si ₃ N ₄ ) serving as a mask for anisotropic etching with KOH or the like is formed on the entire surface of the substrate 26 by low pressure CVD to 1000Å.
It is formed, and an opening is formed around the portion that becomes the joint surface between the beam portion and the weight portion. At this time, both surfaces of the substrate 26 are simultaneously patterned. Since the anisotropic etching in the next step is short-time etching, a silicon oxide film (SiO ₂ ) may be used as a mask. The silicon nitride film 27 is patterned by RIE, ion milling or the like. The reason why both surfaces of the substrate 26 are patterned at the same time is to facilitate the mask pattern alignment after direct bonding.

Next, as shown in (b), both surfaces of the substrate 26 are etched to a depth of about 10 μm by an etching solution such as KOH to form a recess 26a, and then, as shown in (c), Only the silicon nitride film 27 on the bonding surface of 26 is etched by ion milling, plasma etching, or hot phosphoric acid solution. Since the silicon nitride film on the back surface serves as a mask for anisotropic etching in a later step, an alkali-resistant thick coating resist of 4 μm or more is applied and etching is performed while protecting the back surface.

Next, a manufacturing method up to the direct bonding step of the first conductivity type semiconductor substrate to be the beam portion will be described. First,
As shown in (d), an opposite conductivity type semiconductor layer 29 is formed on one surface of the first conductivity type semiconductor substrate 28. In particular,
N-type single crystal silicon is grown to a thickness of about 10 μm on a P-type silicon substrate. Since the opposite conductivity type semiconductor layer 29 which is the N type epitaxial layer serves as a beam portion that supports the weight portion,
When changing the thickness of the beam portion, the thickness of the opposite conductivity type semiconductor layer 29 may be changed.

Next, as shown in (e), a resist mask 30 is formed on the surface of the semiconductor layer 29 of the opposite conductivity type, and a piezoresistive forming portion which is a semiconductor element is opened. The mask is not limited to the resist mask, and a thin film such as a silicon oxide film may be used as the mask.

Next, as shown in (f), a semiconductor element is formed on the opposite conductivity type semiconductor layer 29 at the opening of the resist mask 30. As this semiconductor element, a piezoresistor 31
To use. Impurity resistance is used as the piezo resistance. When the impurity resistance is used, the impurity concentration that makes the temperature characteristic of sensitivity zero is 10 × (18) pieces / cm ³ and 10 × (2
0) pieces / cm ³ , but 10 × (18) pieces in terms of resistance and sensitivity
/ cm ³ is preferable. In practice, boron (B) or the like is formed by ion implantation on the reverse conductivity type semiconductor layer 29 formed of N-type single crystal silicon. It may be formed not only by ion implantation but also by diffusion.

Next, as shown in (g), the substrate 26 and the first-conductivity-type semiconductor substrate 28 are bonded to each other so that the opposite-conductivity-type semiconductor layer 29 is in contact with the substrate 26, and directly in a furnace at 1200 ° C. or higher. To join.

Further, as shown in (h), the first conductivity type semiconductor substrate 28 is removed by electrolytic etching. The electrolytic etching is performed in the same manner as the method shown in FIG. Electrolytic etching uses a pn junction as an etch stopper, and can form a low-concentration n-type silicon thin film without the problem of internal stress that occurs when high-concentration boron is added. Therefore, piezoresistance cannot be formed in this region. It is possible. As the etching solution, KOH aqueous solution or CsOH aqueous solution is used. During etching, the back surface of the substrate 26 may be protected by a silicon nitride film or the like, leaving only the first conductivity type semiconductor substrate 28 to be removed by etching.

Next, as shown in (i), an insulating film 32 is formed on the semiconductor layer 29 of the opposite conductivity type, and an opening above the piezoresistor 31 is opened to form a metal wiring 33. Gold or the like is desirable for the metal wiring. A uniform layer is formed by sputtering or the like and patterned by ion milling, RIE or the like.
Patterning may be performed by wet etching with aqua regia.

Finally, as shown in (j), anisotropic etching is performed from the back surface of the substrate 26 with KOH or the like. At this time, the surface side on which the piezoresistor 31 is formed is protected by a silicon nitride film or a jig. The etching groove reaches the recess formed in the joint surface, the weight portion 334 is separated from the substrate 26, and the triaxial acceleration sensor supported by the beam portion 35 is produced.

[0042]

As described above, according to the present invention, according to claim 1 to manufacture how the acceleration sensor according to claim 2, by constructing the weight portion with a metal having a larger specific gravity than that of silicon, the same size of the In the case of the weight portion, a highly sensitive acceleration sensor can be formed.

Further, according to the acceleration sensor manufacturing method of the first or second aspect, the direct bonding of Si-Si and the flattening of the bonding surface with a low yield are unnecessary.

In the method of manufacturing an acceleration sensor according to the first aspect of the present invention, since the process which conventionally required two wafers can be formed by one wafer, the cost can be reduced and the yield can be lowered in the conventional process. In the beam forming step, it is possible to improve the yield by forming the beam by electrolytic etching. Also,
This makes it possible to simplify the complicated process.

The method for producing an acceleration sensor according to the second aspect, because it can form a process that required the conventional two wafers in a single wafer, it is possible to reduce the cost, conventionally, it has lowered the yield in the process Further, it is possible to improve the yield by using the sacrificial layer forming step and the beam forming step as a material for forming the sacrificial layer and forming the beam by electrolytic etching. In addition, the complicated process can be simplified.

[0046]

[Brief description of drawings]

FIG. 1 is a cross-sectional view showing an embodiment of a method for manufacturing an acceleration sensor of the present invention.

FIG. 2 is an explanatory diagram showing a method of electrolytic etching.

FIG. 3 is a cross-sectional view showing a reference example of the method for manufacturing the acceleration sensor of the present invention.

FIG. 4 is an explanatory diagram showing a method of an acceleration sensor.

FIG. 5 is a perspective view showing an example of a conventional acceleration sensor.

FIG. 6 is a sectional view showing an example of a conventional acceleration sensor.

[Explanation of symbols]

11 First conductivity type semiconductor substrate (substrate) 12 Second conductivity type semiconductor layer 21 Cavity 22 beams 23 Resist pattern (sacrificial layer) 24 Metal thin film 25 Weight 26 Substrate 28 First conductivity type semiconductor substrate 29 Reverse conductivity type semiconductor layer 31 Piezoresistor (semiconductor element)

─────────────────────────────────────────────────── ─── Continuation of the front page (58) Fields surveyed (Int.Cl. ⁷ , DB name) H01L 29/84 G01P 15/12

Claims (2)

(57) [Claims] 1. A substrate and a beam formed on the substrate,
A method of manufacturing an acceleration sensor, comprising: a weight portion fixed to one surface of the beam, wherein a second conductivity type semiconductor layer serving as the beam is formed on a surface of a first conductivity type semiconductor substrate which is the substrate. growing a, forming a cavity by performing electrolytic etching from the back surface of the first conductivity type semiconductor substrate to reach the second conductivity type semiconductor layer, the rear surface of the cavity above the beam, before Forming a sacrificial layer around the connecting portion between the beam and the weight portion, and forming a part of the weight portion on the surface and the side surface of the sacrificial layer around the connecting portion between the beam and the weight portion. A method of manufacturing an acceleration sensor, comprising: a step of forming a thin film; a step of plating the surface of the metal thin film with metal; and a step of removing the sacrificial layer. Wherein, wherein the sacrificial layer is composed of a resist, the production method of an acceleration sensor according to claim 1, wherein.