JP5023558B2 - Acceleration sensor and manufacturing method thereof - Google Patents

Description

  The present invention relates to an acceleration sensor for detecting acceleration and a method for manufacturing the same.

A technique of an acceleration sensor that measures acceleration by using a transducer structure made of a semiconductor and detecting deflection with a piezoresistive element is disclosed (see Patent Document 1).
JP 2003-329702 A

Here, wiring is formed in the acceleration sensor for electrical connection with the piezoresistive element.
However, it has been found that there is a possibility that defects such as wire breakage may occur during the formation of the acceleration sensor.
In view of the above, it is an object of the present invention to provide an acceleration sensor and a method for manufacturing the same that reduce wiring defects.

  An acceleration sensor according to an aspect of the present invention includes a fixed portion having an opening, a displacement portion that is disposed in the opening and is displaced with respect to the fixed portion, and a connection that connects the fixed portion and the displacement portion. A first structure body integrally formed of a flat plate-like first semiconductor material, a weight portion joined to the displacement portion, and arranged to surround the weight portion, And a pedestal joined to the fixed part, and a second structure that is made of a second semiconductor material and is stacked on the first structure, and is arranged at the connection part. A piezoresistive element comprising a diffusion layer, a first insulating layer covering the diffusion layer, an interlayer connection conductor passing through the first insulating layer and connected to the diffusion layer, and the insulation A wiring disposed on the layer and connected to the interlayer connection conductor, and Al and N A conductor portion including bets, characterized by comprising, a second insulating layer covering the conductor portion.

  A method for manufacturing an acceleration sensor according to one embodiment of the present invention includes sequentially stacking a first layer made of a first semiconductor material, a second layer made of an oxide, and a third layer made of a second semiconductor material. Diffusing impurities in the first layer of the semiconductor substrate formed to form a diffusion layer as a piezoresistive element; forming a first insulating layer on the first layer; Forming an opening leading to the diffusion layer in the first insulating layer; an interlayer connection conductor passing through the first insulating layer and connected to the diffusion layer; and disposed on the insulating layer and A step of forming a conductor portion including Al and Nd, a step of forming a second insulating layer covering the conductor portion, and a wiring board connected to the interlayer connection conductor; Heat-treating the diffusion layer and the interlayer connection conductor Characterized by comprising the steps of contacting over beam properties, a.

  ADVANTAGE OF THE INVENTION According to this invention, the acceleration sensor which aimed at reduction of the defect of wiring and its manufacturing method can be provided.

Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings.
FIG. 1 is a perspective view showing an acceleration sensor 100 according to an embodiment of the present invention. FIG. 2 is an exploded perspective view showing a state in which the acceleration sensor 100 is disassembled. FIG. 3 is a top view illustrating a state in which the wiring on the connection portion (beam) of the acceleration sensor 100 is viewed from the top surface. 4 is a partial cross-sectional view showing a state in which the acceleration sensor 100 is cut along A1-A2 in FIG. Note that the illustration of the wiring is limited in FIGS. 1 to 3 in consideration of the visibility and the correspondence with FIG.

The acceleration sensor 100 includes a first structure 110, a joint 120, a second structure 130, and a base body 140 that are stacked on each other. In FIG. 2, the description of the joint 120 is omitted for easy viewing.
The outer periphery of the first structure 110, the joint 120, the second structure 130, and the base body 140 is, for example, a substantially square shape with a side of 1 mm, and the height thereof is, for example, 3 to 12 μm, 0.5 to 3 μm, 600 to 725 μm, and 600 μm.
The first structure 110, the junction 120, and the second structure 130 can be composed of silicon, silicon oxide, and silicon, respectively, and an SOI (Silicon On Insulator) having a three-layer structure of silicon / silicon oxide / silicon. It can be manufactured using a substrate. The base 140 can be made of, for example, a glass material.

  The first structure 110 has a substantially square outer shape, and includes a fixed portion 111, a displacement portion 112, and a connection portion 113, and a wiring structure 150 is disposed thereon. The first structure 110 can be formed by etching the semiconductor material film to form the opening 115.

The fixed portion 111 is a frame-shaped substrate whose outer periphery and inner periphery (opening) are both substantially square. The fixed portion 111 has a shape corresponding to a pedestal 131 described later, and is joined to the pedestal 131 by the joining portion 120.
The displacement part 112 is a substrate having a substantially square outer periphery, and is arranged in the vicinity of the center of the opening of the fixed part 111.
The connection part (beam) 113 is a substantially rectangular substrate, and connects the fixed part 111 and the displacement part 112 in four directions (X positive direction, X negative direction, Y positive direction, Y negative direction).

  The connecting portion 113 functions as a beam that can be bent. The displacement part 112 can be displaced with respect to the fixed part 111 by bending the connection part 113. Specifically, the displacement portion 112 is linearly displaced in the Z positive direction and the Z negative direction with respect to the fixed portion 111. Further, the displacement portion 112 can rotate positively and negatively with respect to the fixed portion 111 with the X axis and the Y axis as rotation axes. That is, the “displacement” here can include both movement and rotation (movement in the Z-axis direction, rotation in the X and Y axes).

By detecting the displacement (movement and rotation) of the displacement part 112, it is possible to measure the acceleration in the X, Y, and Z directions.
On the connecting portion 113, twelve piezoresistive elements R (Rx1 to Rx4, Ry1 to Ry4, Rz1 to Rz4) are arranged. The piezoresistive element R is for detecting the bending (or distortion) of the connecting portion 113 as a change in resistance, and consequently the displacement of the displacing portion 112. Details of this will be described later.

A wiring structure 150 is disposed on the first structure 110.
The wiring structure 150 has a layer structure of an insulating layer 151, a wiring layer 152, and a protective layer 153.
The insulating layer 151 is a layer for separating the first structure 110 and the wiring layer 152. A contact hole (opening) 154 for electrically connecting the piezoresistive element R and the wiring layer 152 is formed in the insulating layer 151. An interlayer connection conductor 155 is disposed in the contact hole 154.

In the wiring layer 152, patterns of wirings 156 and bonding pads 157 are arranged. The wiring 156 electrically connects the piezoresistive element R and the bonding pad 157 via the interlayer connection conductor 155.
The bonding pad 157 is a connection terminal for connecting the acceleration sensor 100 and an external circuit, for example, by wire bonding.
The interlayer connection conductor 155, the wiring 156, and the bonding pad 157 are made of the same material, that is, Al containing Nd. The reason why they are made of the same material is that they are formed by depositing and patterning this material.

The reason why this material is Nd-containing Al is to prevent hillocks from occurring in the interlayer connection conductor 155 and the wiring 156. As will be described later, in order to make the piezoresistive element R and the interlayer connection conductor 155 in ohmic contact (ohmic contact), the interlayer connection conductor 155 is annealed (heat treatment). By this annealing, hillocks are generated in the interlayer connection conductor 155 and the wiring 156, and the electrical connection between the piezoresistive element R and the bonding pad 157 may be poor. When the first and second structures 110 and 130 are formed, a defect such as disconnection may occur in the wiring 156 due to a hillock of the wiring 156.
By containing Nd in Al (1.5 to 10 at%), generation of hillocks in the interlayer connection conductor 155 and the wiring 156 can be prevented, and connection reliability can be improved. Details of this will be described later.

  The protective layer 153 is a kind of insulating layer for protecting the wiring layer 152 from the outside. A pad opening 158 is formed in the protective layer 153 corresponding to the bonding pad 157. This is for connection between an external circuit or the like and the bonding pad 157.

  The second structure 130 has a substantially square outer diameter, and includes a pedestal 131, a weight part 132 (132 a to 132 e), and a protruding part 134. The second structure body 130 can be formed by etching the substrate of semiconductor material to form the opening 133. The pedestal 131 and the weight part 132 are substantially equal in height to each other, are separated by the opening 133, and are relatively movable.

The pedestal 131 is a frame-shaped substrate whose outer periphery and inner periphery (opening 133) are both substantially square. The pedestal 131 has a shape corresponding to the fixed portion 111 and is connected to the fixed portion 111 by the joint portion 120.
The weight part 132 has a mass and functions as a weight that receives a force due to acceleration or an action body. That is, when acceleration is applied, a force acts on the center of gravity of the weight portion 132.
The weight part 132 is divided into substantially rectangular parallelepiped weight parts 132a to 133e. The weight parts 132b to 132e are connected to the weight part 132a arranged at the center from four directions, and can be displaced (moved and rotated) integrally as a whole. That is, the weight part 132a functions as a connection part for connecting the weight parts 132b to 132e.

  The weight part 132 a has a substantially square cross-sectional shape corresponding to the displacement part 112, and is joined to the displacement part 112 by the joining part 120. As a result, the displacement portion 112 is displaced according to the acceleration applied to the weight portion 132, and as a result, the acceleration can be measured.

  Each of the weight portions 132b to 132e is disposed corresponding to the opening 115 of the first structure 110. This is to prevent the weight parts 132b to 132e from coming into contact with the connection part 113 when the weight part 132 is displaced (when the weight parts 132b to 132e come into contact with the connection part 113, detection of acceleration is hindered).

  The reason why the weight part 132 is constituted by the weight parts 132a to 133e is to achieve both miniaturization and high sensitivity of the acceleration sensor 100. When the acceleration sensor 100 is reduced in size (smaller capacity), the capacity of the weight part 132 is reduced and the mass thereof is reduced, so that sensitivity to acceleration is also reduced. The weight parts 132b to 132e are distributed and arranged so as not to hinder the bending of the connection part 113, thereby securing the mass of the weight part 132. As a result, the acceleration sensor 100 can be both downsized and highly sensitive.

The protruding portion 134 is for securing a gap (gap) between the weight portion 132 and the base 140 and enabling the weight portion 132 to be displaced.
The projecting portion 134 is configured integrally with the pedestal 131 and is a frame-shaped substrate whose outer periphery and inner periphery are both substantially square. The outer periphery of the protrusion 134 coincides with the outer periphery of the pedestal 131, and the inner periphery of the protrusion 134 is larger than the inner periphery of the pedestal 131.

The joint 120 connects the first and second structures 110 and 130 as described above. The joint 120 is divided into a joint 121 connecting the fixed part 111 and the pedestal 131, and a joint 122 connecting the displacement part 112 and the weight part 132a. The joint 120 does not connect the first and second structures 110 and 130 at other portions. This is because the connecting portion 113 can be bent and the weight portions 132b to 132e can be displaced.
The junctions 121 and 122 can be configured by etching a silicon oxide film.

The base body 140 is bonded to the protruding portion 134 of the second structure 130 and supports the first and second structures 110 and 130, and a bonding prevention layer 141 is disposed on the upper surface thereof.
The base 140 is made of, for example, a glass material and has a substantially rectangular parallelepiped outer shape.
The base body 140 and the protruding portion 134 are connected by, for example, anodic bonding. Bonding is performed by applying a voltage between the base 140 and the protrusion 134 in contact with each other and heating.

The bonding prevention layer 141 is for preventing the weight part 132 and the base body 140 from being bonded to each other. At the time of the above-described anodic bonding, the weight part 132 comes into contact with the base body 140, and these parts are bonded, and the acceleration sensor 100 may malfunction.
In the region corresponding to the lower surface of the protrusion 134, the bonding prevention layer 141 is not disposed. For example, Cr can be used as the constituent material of the bonding prevention layer 141 excluding the protrusions 134.

(Modification)
In the acceleration sensor 100 described above, the wiring layer 152 in the wiring structure 150 is a single layer. By preventing the wirings 156 from crossing each other, the wiring layer 152 can be a single layer.
On the other hand, the wiring layer can be multi-layered.
FIG. 6 is a partial cross-sectional view showing a state in which the acceleration sensor 100a according to the modification of the present invention is cut, and corresponds to FIG.
The wiring structure 150a of the acceleration sensor 100a has a layer structure of an insulating layer 151, a wiring layer 152, an insulating layer 151a, a wiring layer 152a, and a protective layer 153. That is, the insulating layer 151a and the wiring layer 152a are added to the wiring structure 150 of the acceleration sensor 100, and the wiring layers 151 and 151a and the wiring are two layers.

  The insulating layer 151a is a layer for separating the wiring layers 152 and 152a. A contact hole (opening) for electrically connecting the wiring layers 152 and 152a is formed in the insulating layer 151a as necessary. An interlayer connection conductor is disposed in the contact hole.

In the wiring layer 152a, a wiring 156a and a pattern of bonding pads are arranged. In FIG. 6, bonding pads 157 are arranged on the wiring layer 152 for convenience. However, it is possible to arrange the bonding pads on the wiring layer 152a without arranging them on the wiring layer 152. In this case, the piezoresistive element R and the bonding pad are electrically connected via the wirings 156 and 156a.
As described above, when the wirings are multi-layered, the wirings 156 and 156a can be allowed to intersect on a plane, and the degree of freedom in wiring design is improved.

  Even when the wiring is multi-layered in this way, it is preferable that the wiring 156a itself, the interlayer connection conductor with the wiring 156, and the bonding pad are made of the same material, that is, Al containing Nd. This is for preventing the occurrence of hillocks on the wiring 156a and the like.

(Adjustment of the crossing angle θ of the wirings 156 and 156a)
Here, it is conceivable to shift the crossing angle θ of the wirings 156 and 156a from 90 °. When Nd is added to Al, the wirings 156 and 156a become hard, and there is a possibility of disconnection. By shifting the crossing angle θ from 90 °, the possibility of disconnection of the wirings 156 and 156a can be reduced. The details will be described below.

FIG. 6A is a top view illustrating a state in which the intersecting wirings 156 and 156a are viewed from above. 6B is a cross-sectional view illustrating a state in which the wirings 156 and 156a are cut along B1-B2 in FIG. 6A.
An insulating layer 151a and a wiring 156a are disposed on the wiring 156, and the wirings 156 and 156a intersect at an intersection angle θ1 = 90 °. Since the wiring 156 is disposed across the wiring 156, the wiring 156a has a step, and there is a possibility of disconnection at the boundary of the step, that is, the edge portion EG. During the manufacturing process and use of the acceleration sensor 100, thermal stress is applied to the wiring 156a, and there is a possibility that the edge portion EG will break. This thermal stress is caused by a difference in thermal expansion between the wiring 156a and the first structure 110.

  When pure Al is used for the wiring 156a, the possibility of disconnection of the wiring 156a is small because Al is a relatively flexible material. However, when Nd is added to Al, the wiring 156a becomes hard, and the possibility of disconnection of the wiring 156a increases.

  By reducing the thickness D of the wiring 156, the possibility of disconnection of the wiring 156a can be reduced. However, the acceleration sensor 100 has a certain limit in reducing the thickness D of the wiring 156. It is necessary to reduce the resistance of the wiring 156a to some extent. For this purpose, it is necessary to increase either the film thickness D or the width of the wiring 156a to some extent. However, since the wiring 156a passes over the narrow connection part (beam) 113, there is a restriction on its width, and the film thickness D must be increased to some extent.

Generally, increasing the film thickness D tends to increase the irregularities (surface roughness) on the surface of the wiring 156 due to hillocks and the like. As a result, by increasing the film thickness D, the insulating layer 151a is not sufficiently covered on the wiring 156, and there is a possibility that the wirings 156 and 156a are energized.
By adding an appropriate amount of Nd, generation of hillocks and, in turn, generation of irregularities on the surface of the wiring 156 can be avoided. Note that the thickness of the insulating layer 151a can be increased to prevent energization between the wirings 156 and 156a, but the productivity may decrease.

By shifting the crossing angle θ of the wirings 156 and 156a from 90 °, the possibility of disconnection of the wiring 156a can be reduced.
FIGS. 7A and 7B are top views showing a state in which the wirings 156 and 156a with the crossing angle θ shifted from 90 ° are viewed from above corresponding to FIG. 6A.
The crossing angles θ1 to θ3 in FIGS. 6A, 7A, and 7B are gradually reduced as 90 °>θ1>θ2> θ3. As a result, the lengths L1 to L3 of the edge portions EG of the wirings in FIGS. 6A, 7A, and 7B gradually become longer as L1 <L2 <L3. Since the edge portion EG is disposed so as to cross the wiring 156a, the longer the length L, the lower the possibility of disconnection of the wiring 156a. This is because even if a part of the edge portion EG is damaged, it is possible to avoid the wiring 156a from being completely disconnected unless the entire length L is damaged.

  On the other hand, shifting the crossing angle θ of the wirings 156, 156a from 90 ° increases the area of the region A where the wirings 156, 156a overlap. This means that the risk of short-circuiting the wirings 156, 156a due to hillocks increases. However, this risk can be reduced by adding Nd to Al.

As described above, by shifting the crossing angle θ from 90 °, the possibility of disconnection of the wiring 156a can be reduced. By shifting the crossing angle θ from 90 °, the risk of short-circuiting of the wirings 156 and 156a increases, but this short-circuit can be prevented by adding Nd to Al.
The crossing angle θ is preferably smaller than 90 °, and it is desirable to make the angle low within a range that does not hinder the design.
When the level difference of the wiring 156a is large, disconnection is likely to occur between the level differences (that is, the edge portion EG). As the distance between steps (that is, the length L of the edge portion EG) is increased, the risk of disconnection is reduced. When a hard material (Cr, Ni, Co, Mo, Ta, Al—Nd (high concentration Nd of 2% or more, etc.)) is used for the wiring 156 as compared with pure Al having flexibility, this tendency is greater. Become.

FIG. 8A to FIG. 8C are top views showing a state where the intersecting wirings 156 and 156a are viewed from above.
In FIG. 8A, the entire wiring 156 has a single linear shape, but the wiring 156a is divided into three linear partial wirings, of which the partial wiring CS1a intersects the wiring 156 (intersection angle). θ = 90 °).
8B and 8C, the crossing angle θ is smaller than 90 °. Further, both the wirings 156 and 156a are divided into three linear partial wirings, and the partial wirings CS2 and CS2a and the partial wirings CS3 and CS3a intersect each other (crossing angle θ <90 °). 8B and 8C, the wirings 156 and 156a are symmetrical with each other and the wiring lengths are almost equal. In FIG. 8B, the angles θ2 and θ2a formed by the partial wirings CS2 and CS2a are opposite to each other and the absolute values are equal (θ2 = −θ2a). This is a factor for making the wiring lengths of the wirings 156 and 156a equal. is there. This situation is the same in FIG. 8C (θ3 = −θ3a). For example, a difference of about 10 ° to 20 ° is allowed for the absolute values of the angles θ2 and θ2a.

As shown below, it is preferable from the characteristics of the acceleration sensor 100 that the wiring lengths of the wirings 156 and 156a are equal.
In the acceleration sensor 100, the acceleration in the triaxial direction is converted into an electric signal and transmitted on the wirings 156 and 156a. For this reason, it is desirable that the electrical signals of the accelerations in the three-axis directions are transmitted equally. If transmission of electrical signals of acceleration in the three axial directions is not equivalent, accurate detection of acceleration cannot be performed unless each electronic signal is corrected by some means.
In order to transmit the electrical signals equally, it is desirable that the resistances of the wirings 156 and 156a that transmit the electrical signals of the accelerations in the three-axis directions are equal. That is, it is desirable that the material, the wiring width, the wiring length, and the wiring film thickness of the wirings 156 and 156a are the same.
As shown in FIGS. 8B and 8C, by making the wiring lengths of the wirings 156 and 156a equal, electrical signals of acceleration in the three-axis directions are equally transmitted, and accurate detection of the acceleration becomes easy.

  In general, it is easy to make the material, wiring width, and wiring film thickness of the wirings 156 and 156a the same, but making the wiring lengths the same may cause design difficulties. For example, when using an existing IC chip, it is necessary to arrange the bonding pad 157 of the acceleration sensor 100 in accordance with the extraction electrode of the IC chip. As shown in FIGS. 8B and 8C, it is easier to make the wiring lengths of the wirings 156 and 156a equal by providing symmetry to the arrangement of the wirings 156 and 156a.

FIG. 9A is a top view showing a state in which the intersecting portions of the intersecting wirings 156 and 156a are viewed from above. FIG. 9B is a top view showing only the wiring 156 of FIG. 9A.
Here, each of the wirings 156, 156a has triangular protrusions T1, T1a. In other words, the wirings 156 and 156a protrude from the original width at the locations where the wirings 156 and 156a intersect. The arrangement of the projecting portions T1 and T1a on the wirings 156 and 156a can be applied separately from making the angle of the crossing angle θ smaller than 90 °.

Since the wiring 156 has the protruding portion T1, the risk of disconnection of the wiring 156a can be reduced. This is because the wiring 156a is hardly cracked for the following reasons (1) and (2).
(1) Since the protruding portion T1 has a tapered shape (the width is narrower toward the tip thereof), a taper is formed at the tip of the protruding portion T2 due to the effect of the etching solution when the wiring 156 is formed (during etching). (Thickness is gradually reduced), the step is relaxed. The relief of the step is caused by the presence of the protruding portion T1, and the protruding portion T1a is basically not involved.
9C is a cross-sectional view illustrating a state in which the wirings 156 and 156a are cut at C1-C2 in FIG. 9A. The protruding portion T1 of the wiring 156 has a taper, and the wiring 156a is disposed on the wiring 156 with an inclination. As described above, the effect of the etching solution wraps around the tip of the protrusion T1, and the film thickness of the wiring 156 decreases. As a result, the wiring 156a is hardly cracked.

  Here, the projecting portion T1 may have a tapered shape as a whole, and is a portion having the same width and partially reversed in width (in part, the width on the root side is narrow and the width on the tip side is wide). Is recognized. If a narrower portion than the base side is present on the tip side, the protruding portion T1 can be recognized as a tapered shape. Even in this case, the step can be reduced.

(2) The shape of the intersecting portion of the wirings 156 and 156a is complicated by the protruding portions T1 and T1a, and the length of the intersecting portion (the length L of the edge portion EG) is increased. As a result, when a crack is generated in a part of the wiring 156a, the possibility that the wiring 156a is disconnected is reduced.
Here, if the protrusions T1 and T1a have the same shape, it is easy to make the electric resistances of the wirings 156 and 156a the same, which is preferable.

  Here, as a preferred example of the aspect ratio (wiring direction length La: wiring vertical direction length Lb) of the protruding portion T1, a so-called golden ratio can be cited. The golden ratio is a ratio La: Lb when La: Lb = Lb: (La + Lb) is established, and La: Lb = 1: ((1 + √5) / 2) = 0.618: 1 It is.

By applying the golden ratio to the aspect ratio of the protrusion T1, the same shape is formed in the outer shape of the protrusion T1 (a Fibonacci sequence of aspect ratios), and by complicating the shape of the intersection, the length of the intersection is increased. You can lengthen the length. As a result, the risk of disconnection of the wiring 156a can be reduced.
Thus, by arranging the protruding portions T1 and T1a at the intersections of the wirings 156 and 156a, the risk of disconnection of the wiring 156a can be reduced.
At this time, the apex of the protruding portion T1 is arranged along the wiring direction of the wiring 156a. That is, the angle θ1 formed by the wiring direction of the wiring 156 and the arrangement direction of the apexes of the protrusion T1 is 90 °, which is equal to the intersection angle θ = 90 ° of the wirings 156 and 156a. Further, when the aspect ratio of the protrusion T1 is a golden ratio, the angle θ3 formed by the arrangement direction of the vertices of the protrusion T1 and the side of the protrusion T1 is 17.14 ° as shown in the following equation.
θ3 = tan ⁻¹ (La / (2 * Lb))
= Tan ⁻¹ (1 / (1 + √5)) = 17.14 °

  FIG. 9D is a top view illustrating the wiring 156 corresponding to FIG. 9B and having the curved protruding portion T2. Here, a case is considered in which each of the wirings 156 and 156a has curved protruding portions T2 and T2a at intersections.

The risk of disconnection of the wiring 156a can be further reduced by the curved protruding portion T2 compared to the protruding portion T1 on the straight line. This is because the wiring 156a is difficult to crack for the following three reasons.
(1) Due to the sharp tip of the projecting portion T2, a taper is formed at the tip of the projecting portion T2 due to the effect of the surrounding of the etching solution when forming the wiring 156 (during etching), and the step is alleviated. The

(2) The protruding portion T2 complicates the shape of the intersecting portions of the wirings 156 and 156a, and the length of the intersecting portions becomes long, so that the possibility of disconnection is reduced even if a crack occurs.
These reasons (1) and (2) correspond to the reasons (1) and (2) in the case of the linear protrusion T1, but the shape of the protrusion T2 is curved. The effect is greater.
On the other hand, the following reason (3) is inherent to the curved protrusion T2.
(3) A crack may occur along the steps of the wirings 156 and 156a. Due to the curved protrusion T2, the step is not formed in one straight line, and cracks can be prevented.

FIG. 9E is a top view showing a state in which the intersecting portions of the intersecting wirings 156 and 156a are viewed from above. FIG. 9F is a top view showing only the wiring 156 of FIG. 9E.
As shown in the figure, the crossing angle of the wirings 156, 156a can be made different from 90 °, and the projecting portions T3, T3a can be arranged. In this way, the disconnection risk of the wiring 156a can be further reduced.

  At this time, the apex of the protruding portion T3 is arranged along the wiring direction of the wiring 156a. That is, the angle θ1 formed by the wiring direction of the wiring 156 and the arrangement direction of the apexes of the protrusion T1 is equal to the intersection angle θ of the wirings 156 and 156a. As described above, when the aspect ratio of the protruding portion T3 is a golden ratio, the angle θ3 formed by the arrangement direction of the vertices of the protruding portion T1 and the side of the protruding portion T1 is 17.14 °.

Here, an appropriate range of the angle θ4 formed by the side of the protruding portion T3 and the wiring direction of the wiring 156 is considered.
90 ° ≦ θ4 ≦ 90 ° + θ3
Here, if the angle θ4 is more acute than 90 °, etching at the time of forming the wiring 156 becomes difficult, so (90 ° ≦ θ4).
Further, (θ4 ≦ 90 ° + θ3) is established from the following two equations.
θ4 = θ1 + θ3
θ1 ≦ 90 °
The former of these two formulas is based on the fact that the shape of the protrusion T3 is a triangle and the sum of the inner angles is 180 °.
When the angle θ3 is 17.14 °, a preferable range of the angle θ4 is expressed as follows.
90 ° ≦ θ4 ≦ 90 ° + θ3 = 107.14 °

Here, from (θ4 = θ1 + θ3), the following equation is established.
90 ° -θ3 ≦ θ1 ≦ 90 °
When the angle θ3 is 17.14 °, a preferable range of the angle θ1 is expressed as follows.
72.86 ° ≦ θ1 ≦ 90 °
Note that the angle θ1 is not necessarily limited to this range, and may be 10 ° ≦ θ1 ≦ 90 °.

FIG. 9G is a top view illustrating a state in which the intersecting portions of the intersecting wirings 156 and 156a are viewed from above. FIG. 9H is a top view showing only the wiring 156 of FIG. 9H.
The side of the protruding portion T4 is constituted by a so-called Koch curve.
A Koch curve is a type of fractal curve, and is a curve that can be created by repeating infinitely adding a smaller triangle to the side of a triangle (by dividing a line segment into three equal parts, The limit of repeating infinitely the operation of replacing the part with a broken line). In other words, a triangle having a side that is one third of the length of the side is attached to the center of each side of the triangle. A triangle with a length of one third is added to each side of the figure. Repeat this process and add small triangles.

Theoretically, the Koch curve has an infinite number of vertices and the total length of the sides is infinite. For this reason, the shape of the protrusion T4 is complicated. For this reason, a taper is formed in the peripheral part (front-end | tip part) by the surrounding of etching liquid at the time of formation of protrusion part T4. As a result, disconnection hardly occurs in the wiring 156a.
Since the Koch curve can be easily formed by a computer program, mask formation for the wiring 156 and EB drawing are easy.

FIG. 9I is a top view illustrating a state where the intersecting portions of the intersecting wirings 156 and 156a are viewed from above.
The upper wiring 156 a has a partial wiring CS 0, and the wiring direction thereof substantially coincides with the wiring direction of the wiring 156. The partial wiring CS0 has an opening H (the partial wiring CS0 branches into two). Further, the width of the partial wiring CS0 is larger than the width of the wiring 156 overlapping therewith. The widths of the partial wiring CS0 and the wiring 156 and the size of the opening H are adjusted so that the electric resistances of the wiring 156 and the partial wiring CS0 are the same (matched in terms of area).

Since the width of the partial wiring CS0 is larger than the width of the wiring 156, even if the alignment of the wirings 156 and 156a is misaligned, any of the branched partial wirings CS0 is placed on the step of the wiring 156, making it difficult to disconnect. Further, since the partial wiring CS0 is branched into two, the probability of disconnection is reduced to ½. Furthermore, the possibility of disconnection is also reduced when the partial wiring CS0 and the wiring 156 are arranged in parallel and the stepped portion becomes long.
Even in this case, the possibility of disconnection can be reduced by introducing a fractal structure such as a Koch curve continuous to the stepped portion.
Since this modification has essentially no difference from the first embodiment except that the wiring is multilayered, other description is omitted.

(Operation of the acceleration sensor 100)
The principle of acceleration detection by the acceleration sensor 100 will be described. As described above, a total of twelve piezoresistive elements Rx1 to Rx4, Ry1 to Ry4, and Rz1 to Rz4 are arranged in the connecting portion 113.
Each of these piezoresistive elements can be constituted by a P-type or N-type impurity doped region (diffusion layer 116) formed in the vicinity of the upper surface of the connection portion 113 made of silicon.

Three sets of piezoresistive elements Rx1 to Rx4, Ry1 to Ry4, and Rz1 to Rz4 are arranged on the connecting portion 113 so as to be aligned in a straight line in the X axis direction, the Y axis direction, and the X axis direction.
The piezoresistive elements Rx1 to Rx4 and Rz1 to Rz4 are arranged differently depending on the connecting portion 113. This is to make the detection of the bending of the connecting portion 113 by the piezoresistive element R more accurate.

  The three sets of piezoresistive elements Rx1 to Rx4, Ry1 to Ry4, and Rz1 to Rz4 function as X, Y, and Z axis direction component displacement detectors that detect the displacement of the weight part 132 in the X, Y, and Z axis directions, respectively To do. The four piezoresistive elements Rz1 to Rz4 are not necessarily arranged in the X-axis direction, and may be arranged in the Y-axis direction.

The direction and amount of acceleration can be detected from the combination of the expansion (+) and contraction (−) of the piezoresistive element R and the amount of expansion / contraction. The expansion and contraction of the piezoresistive element R can be detected as a change in the resistance of the piezoresistive element R.
For example, consider a case where the crystal plane index of the constituent material of the connection portion 113 is {100} and the crystal direction in the longitudinal direction of the piezoresistive element R is <110>. Here, it is assumed that each piezoresistive element R is constituted by P-type impurity doping into silicon. At this time, the resistance value in the longitudinal direction of the piezoresistive element R increases when a stress in the expansion direction is applied, and decreases when a stress in the contraction direction is applied.
Note that when the piezoresistive element R is configured by doping N-type impurities into silicon, the increase and decrease of the resistance value is reversed.

  10A to 10C are circuit diagrams showing configuration examples of detection circuits for detecting accelerations in the X, Y, and Z axial directions from the resistance of the piezoresistive element R, respectively. In this detection circuit, in order to detect acceleration components in the X, Y, and Z axial directions, a bridge circuit including four sets of piezoresistive elements is formed, and the bridge voltage is detected.

In these bridge circuits, the relationship of the output voltage Vout (Vx_out, Vy_out, Vz_out) with respect to each of the input voltages Vin (Vx_in, Vy_in, Vz_in) is expressed by the following equations (1) to (3).
Vx_out / Vx_in =
[Rx4 / (Rx1 + Rx4) −Rx3 / (Rx2 + Rx3)] (1)
Vy_out / Vy_in =
[Ry4 / (Ry1 + Ry4) −Ry3 / (Ry2 + Ry3)] (2)
Vz_out / Vz_in =
[Rz3 / (Rz1 + Rz3) −Rz4 / (Rz2 + Rz4)] (3)

  The amount of expansion and contraction of the piezoresistor R is proportional to the acceleration, and the amount of expansion and contraction of the piezoresistive element R is proportional to the change in the resistance value R. As a result, the ratio of the output voltage to the input voltage (Vxout / Vxin, Vyout / Vyin, Vzout / Vzin) is proportional to the acceleration, and the acceleration on the X, Y, and Z axes can be measured separately. .

(Creation of acceleration sensor 100)
The production process of the acceleration sensor 100 will be described.
FIG. 11 is a flowchart showing an example of a procedure for creating the acceleration sensor 100. 12A to 12O correspond to FIG. 4 and are cross-sectional views showing the state of the acceleration sensor 100 in the creation procedure of FIG.

(1) Preparation of semiconductor substrate W (step S11 and FIG. 12A)
As shown in FIG. 12A, a semiconductor substrate W is prepared by laminating three layers of first, second and third layers 11, 12 and 13.

The first, second, and third layers 11, 12, and 13 are layers for forming the first structure 110, the joint 120, and the second structure 130, respectively. A layer made of silicon or silicon.
A semiconductor substrate W having a three-layer structure of silicon / silicon oxide / silicon is obtained by bonding a silicon substrate with a silicon oxide film on a silicon substrate and then polishing the latter silicon substrate thinly. It can be created (so-called SOI substrate). The semiconductor substrate W can also be created by sequentially laminating a silicon oxide film and a silicon film on a silicon substrate.

The reason why the second layer 12 is made of a material different from that of the first and third layers 11 and 13 is that the etching characteristics are different from those of the first and third layers 11 and 13 and the etching stopper layer is used. It is for use. The second layer 12 functions as an etching stopper layer in both the etching from the upper surface of the first layer 11 and the etching from the lower surface of the third layer 13.
Here, the first layer 11 and the third layer 13 are made of the same material (silicon), but the first, second, and third layers 11, 12, and 13 are all made of different materials. You may comprise by.

(2) Formation of diffusion mask 14 (step S12 and FIG. 12B)
A diffusion mask 14 is formed on the first layer 11. This is because the diffusion layer 116 of the piezoresistive element R is formed in the first layer 11.
For example, a SiN film is laminated on the first layer 11 by low pressure CVD (Low Pressure-Chemical Vapor Deposition), and an opening is formed by RIE (Reactive Ion Etching) using a resist as a mask. In this way, a film having an opening 15 on the first layer 11, that is, a diffusion mask 14 is formed.

(3) Formation of diffusion layer 116 (step S13 and FIG. 12C)
A pattern of the diffusion layer 116 is formed using the diffusion mask 14. This formation is performed as in the following ac.
a. An impurity layer, for example, a layer containing B is formed on the diffusion mask 14. For example, a layer containing B can be formed by spin coating.
b. By the heat treatment, impurities contained in the impurity layer, for example, B are diffused into the first layer 11 to form a diffusion layer 116. For example, B is thermally diffused by heat treatment at 1000 ° C.
c. The impurity layer on the diffusion mask 14 is removed. For example, the B impurity layer is etched using hydrofluoric acid.
d. In the above a to c, the diffusion layer 116 is formed using thermal diffusion. On the other hand, the diffusion layer 116 may be formed by means other than thermal diffusion, for example, ion implantation.

(4) Removal of diffusion mask 14 (step S14 and FIG. 12D)
The diffusion mask 14 is removed. When the constituent material of the diffusion mask 14 is SiN, it can be etched and removed by hot phosphoric acid. As a result, the first layer 11 is exposed.

(5) Formation of insulating layer 151 (step S15 and FIG. 12E)
An insulating layer 151 is formed over the first layer 11. For example, a layer of SiO ₂ can be formed by thermally oxidizing the surface of the first layer 11.
A contact hole (opening) 154 is formed in the insulating layer 151. For example, a pattern of contact holes (openings) 154 can be formed in the insulating layer 151 by RIE using a resist as a mask.

(6) Formation of wiring 156 (step S16 and FIG. 12F)
A wiring 156 is formed over the insulating layer 151. This formation is performed as follows a and b.
a. An Al layer containing Nd is formed on the first layer 11. For example, Al containing Nd can be deposited by sputtering. As a result of this deposition, a wiring layer 152 is formed on the first layer 11, and an interlayer connection conductor 155 is formed in the contact hole 154.
The reason why Al containing Nd is used as the constituent material of this film is to prevent the generation of hillocks during the subsequent heat treatment.

  b. The wiring layer 152 is patterned to form patterns of wiring 156 and bonding pads 157. For example, the pattern of the wiring 156 and the bonding pad 157 can be formed by wet etching using a resist as a mask.

c. Here, as shown in FIG. 6, the wiring layer can be multi-layered. In this case, the following steps are added.
An insulating layer 151 a is formed on the wiring layer 152. For example, a SiO 2 layer is formed on the wiring layer 152 by CVD.
A second contact hole pattern is formed in the insulating layer 151a. This is for conduction between the upper and lower wiring layers 152, 152a.
An Al layer (wiring layer 152a) containing Nd is formed on the insulating layer 151a. At this time, a second interlayer connection conductor is formed in the second contact hole. Even when the number of wiring layers is two or more, using Al containing Nd as the constituent material is effective in preventing the occurrence of hillocks.
The wiring layer 152a is patterned to form the wiring 156a. The wiring 156a is connected to the wiring 156 by the second interlayer connection conductor.
・ The above is the case where there are two wiring layers. If necessary, the third insulating layer and the third wiring can be added to form three or more wiring layers.

(7) Formation of protective layer 153 (step S17 and FIG. 12G)
A protective layer 153 is formed on the wiring layer 152. For example, a SiN layer is deposited by low pressure CVD. At this time, for example, the wiring layer 152 may be heated to about 300 ° C. or 350 ° C., and hillocks may occur in the wiring 156. By forming the constituent material of the wiring layer 152 as Nd-containing Al, generation of hillocks is prevented. Details of this will be described later.

(8) Heat treatment (step S18 and FIG. 12G)
The semiconductor substrate W is heat-treated. This is to make ohmic contact between the diffusion layer 116 and the interlayer connection conductor 155. At this time, for example, the wiring layer 152 may be heated to about 380 ° C. or 400 ° C., and hillocks may be generated in the wiring 156. By forming the constituent material of the wiring layer 152 as Nd-containing Al, generation of hillocks is prevented. Details of this will be described later.

(9) Formation of pad opening 158 (step S19 and FIG. 12H)
A pad opening 158 is formed in the protective layer 153. The pad opening 158 can be formed by etching the protective layer 153 by RIE using a resist as a mask.

(10) Creation of first structure 110 (etching of first layer 11, step S20, and FIG. 12I)
By etching the first layer 11, the opening 115 is formed and the first structure 110 is formed. That is, by using an etching method that has erosion with respect to the first layer 11 and does not have erosion with respect to the second layer 12, a predetermined region (opening 115) of the first layer 11 is used. Then, etching is performed in the thickness direction until the upper surface of the second layer 12 is exposed.

A resist layer having a pattern corresponding to the first structure 110 is formed on the upper surface of the first layer 11, and an exposed portion not covered with the resist layer is eroded vertically downward. In this etching step, since the second layer 12 is not eroded, only a predetermined region (opening 115) of the first layer 11 is removed.
FIG. 12I shows a state where the first structure 110 is formed by etching the first layer 11 as described above.

(11) Creation of second structure 130 (etching of third layer 13, step S21, and FIGS. 12J and 12K)
The second structure 130 is created in two stages.
1) Formation of protrusion 134 (FIG. 12J)
A resist layer having a pattern corresponding to the protrusion 134 is formed on the lower surface of the third layer 13, and the exposed portion not covered with the resist layer is eroded vertically upward. As a result, a recess (concave portion) 21 is formed on the lower surface of the third layer 13. The outer periphery of the recess 21 is a protrusion 134.

2) Formation of pedestal 131 and weight part 132 (FIG. 12K)
By further etching the recess 21 of the third layer 13, an opening 133 is formed, and the second structure 130 is formed. That is, with respect to a predetermined region (opening 133) of the third layer 13 by an etching method that has erosion with respect to the third layer 13 and does not have erosion with respect to the second layer 12. Etching in the thickness direction is performed until the lower surface of the second layer 12 is exposed.

  A resist layer having a pattern corresponding to the second structure 130 is formed on the lower surface of the third layer 13. The exposed portion not covered with the resist layer in the recess 21 is eroded vertically upward. In this etching process, since the second layer 12 is not eroded, only a predetermined region (opening 133) of the third layer 13 is removed.

  FIG. 12K shows a state where the second structure 130 is formed by etching the third layer 13 as described above.

  The order of the etching process (step S20) for the first layer 11 and the etching process (step S21) for the third layer 13 described above can be interchanged. Either etching process may be performed first or at the same time.

(12) Creation of joint 120 (etching of second layer 12, step S22, and FIG. 12L)
Etching the second layer 12 forms the joint 120. That is, the second layer 12 is erodible with respect to the second layer 12, and the second layer 12 is etched with respect to the first layer 11 and the third layer 13 by an etching method without erosion. Etching is performed in the thickness direction and the layer direction from the exposed portion.

In the above manufacturing process, it is necessary to perform an etching method that satisfies the following two conditions in the step of forming the first structure 110 (step S20) and the step of forming the second structure 130 (step S21). There is.
The first condition is to have directionality in the thickness direction of each layer. The second condition is erosive to the silicon layer but not erodible to the silicon oxide layer. That is. The first condition is a condition necessary for forming an opening or a groove having a predetermined size, and the second condition is for using the second layer 12 made of silicon oxide as an etching stopper layer. This is a necessary condition.

As an etching method that satisfies the first condition, an inductively coupled plasma etching method (ICP etching method) can be given. This etching method is effective when digging deep grooves in the vertical direction, and is a kind of etching method generally called DRIE (Deep Reactive Ion Etching).
In this method, an etching stage in which the material layer is dug while being eroded in the thickness direction and a deposition stage in which a polymer wall is formed on the side surface of the dug hole are alternately repeated. Since the side walls of the holes that have been dug are protected by the formation of polymer walls, it is possible to advance erosion almost only in the thickness direction.

On the other hand, in order to perform etching satisfying the second condition, an etching material having etching selectivity between silicon oxide and silicon may be used. For example, a mixed gas of SF ₆ gas and O ₂ gas may be used in the etching stage, and C ₄ F ₈ gas may be used in the deposition stage.

In the etching process for the second layer 12 (step S22), it is necessary to perform an etching method that satisfies the following two conditions. The first condition is to have a direction in the layer direction as well as the thickness direction, and the second condition is erosive to the silicon oxide layer but erodible to the silicon layer. Is not to.
The first condition is a condition necessary for preventing the silicon oxide layer from remaining in an unnecessary portion and preventing the degree of freedom of displacement of the weight portion 132. The second condition is a condition necessary to prevent erosion of the first structure 110 and the second structure 130 made of silicon that has already been processed into a predetermined shape.

As an etching method that satisfies the first and second conditions, wet etching using buffered hydrofluoric acid (for example, a mixed aqueous solution of HF = 5.5 wt% and NH ₄ F = 20 wt%) can be given.

(13) Bonding of base 140 (step S23, and FIGS. 12M and 12N)
1) Formation of anti-bonding layer 141 on substrate 140 (FIG. 12M)
A bonding prevention layer 141 is formed on the base 140. For example, a Cr layer is formed on the upper surface of the substrate 140 by sputtering. Further, the outer periphery of this layer is removed by etching using a resist as a mask so as to correspond to the lower surface of the protrusion 134. This is to prevent the weight portion 132 and the base body 140 from being joined while securing the joint between the protruding portion 134 and the base body 140.

2) Bonding of the semiconductor substrate W and the base 140 (FIG. 12N)
The semiconductor substrate W and the base 140 are bonded. When the constituent materials of the base 140 and the protrusion 134 are glass and Si, anodic bonding (also referred to as electrostatic bonding) is possible.
A voltage is applied between the base 140 and the protrusion 134 in a state where the base 140 and the protrusion 134 are heated. The glass of the substrate 140 is softened by heating. In addition, a sodium-deficient layer is generated on the glass of the substrate 140 by the movement of movable ions (for example, Na ions) contained in the glass. Specifically, mobile ions move in the glass in the direction opposite to the bonding surface and precipitate on the glass surface, and a sodium-deficient layer is formed near the bonding surface in the glass. As a result, an electric double layer is generated between the base 140 and the protrusion 134, and these are joined by the electrostatic attractive force.
At this time, the bonding prevention layer 141 restricts the movement of ions between the base 140 and the weight part 132. As a result, bonding between the base body 140 and the weight portion 132 is prevented.

(14) Dicing of the semiconductor substrate W (step S24 and FIG. 12O)
The semiconductor substrate W and the base body 140 bonded to each other are cut with a dicing saw or the like and separated into individual acceleration sensors 100.

(Details of preventing hillocks)
Hereinafter, details of prevention of hillock generation by using Nd-containing Al as the constituent material of the wiring layer 152 will be described.
The hillock here refers to, for example, a hemispherical protrusion formed on the wiring 156. Compressive stress is generated by heating the wiring 156 and the like. A hillock is generated by plastic deformation of the wiring 156 and the like by this compressive stress.

When a hillock occurs in the wiring 156, stress is applied to the protective layer 153 over the wiring 156. Therefore, the protective layer 153 may be destroyed or the strength may be reduced (for example, the protective layer 153 is cracked). Further, when the strength of the protective layer 153 is reduced, there is a possibility that the protection of the wiring 156 by the protective layer 153 may be insufficient during the subsequent etching for forming the bonding portion 120. In other words, the wiring 156 is etched and disconnected by the etching material of the joint 120, which may cause malfunction of the acceleration sensor 100. Note that the wiring 156 can be protected by forming a resist over the protective layer 153 when the first and second structures 110 and 130 are formed.
Even if the protective layer 153 is not destroyed or the strength is not lowered, the wiring 156 may be disconnected. That is, there is a possibility that the wiring 156 is greatly deformed due to the occurrence of hillocks and is disconnected.

By increasing the thickness of the protective layer 153, the strength can be increased and the influence of hillock generation can be reduced. However, when the protective layer 153 is thickened, the influence of the internal stress of the protective layer 153 on the acceleration sensor 100 also increases.
When the film thickness of the protective layer 153 is increased, the first structure 110 on which the protective layer 153 is formed, in particular, the connecting portion (beam) 113 is bent. This bending may be an obstacle to acceleration detection by the acceleration sensor 100.
Further, when the film thickness of the protective layer 153 is increased, the connecting portion (beam) 113 is substantially increased, and the distance between the surface of the connecting portion (beam) 113 and the piezoresistive element R is increased. These cause a decrease in stress applied to the piezoresistive element R and, consequently, a decrease in sensitivity of the acceleration sensor 10.

In the present embodiment, Nd-containing Al is used as the constituent material of the wiring layer 152, so that generation of hillocks on the wiring 156 is prevented, and an increase in the thickness of the protective layer 153 is unnecessary.
When the constituent material of the wiring 156 is pure Al, the internal stress in the wiring 156 changes from a tensile stress to a compressive stress due to a temperature rise due to heating. When this compressive stress becomes greater than a certain value (hillock generation stress), plastic deformation of the wiring 156 is started. The constituent atoms of the wiring 156 move due to the compressive stress (so-called creep), and gather to form hillocks.
This hillock stress is temperature dependent and decreases with increasing temperature. The hillock stress is larger than the so-called yield stress. That is, even if the compressive stress exceeds the yield stress, hillocks are not immediately formed.

When the constituent material of the wiring 156 is an AlNd alloy, Nd in the wiring 156 recrystallizes and precipitates when the temperature rises. By this deposition, the structure of the wiring 156 is reconfigured, and the internal stress in the wiring 156 is reduced. As a result, generation of hillocks due to compressive stress in the wiring 156 is prevented.
As described above, by adding Nd to Al, Nd is recrystallized when the temperature is raised, and internal stress is alleviated, so that generation of hillocks in the wiring layer 152 is prevented.

  The amount of Nd added to Al is preferably 1.5 to 10 at%, more preferably 2 to 3 at%. If the amount of Nd added is small, the internal stress is not sufficiently relaxed and hillocks may occur. On the other hand, if the amount of Nd added is large, the internal resistance and hardness of the wiring 156 increase, which is not preferable.

It is more effective to use an AlNd alloy as the constituent material of the wiring 156 when the wiring layer is multilayered.
If hillocks occur in the multi-layered wiring layer, the insulating film between the wiring layers and the protective film on the outermost periphery may be destroyed or the strength may be reduced. The decrease in the strength of the insulating film and the protective film may cause insufficient protection of the wiring during etching for forming the first and second structures 110 and 130, leading to disconnection of the wiring and poor insulation between the wirings. There is. For example, if a hillock occurs at a location where the upper and lower wirings intersect, there is a high possibility that an insulation failure will occur between the upper and lower wirings.
As described above, by using AlNd alloy as the constituent material of the wirings 156 and 156a, it is possible to effectively cope with the multilayered wiring layer.

(Example)
An experimental study was made on the constituent material of the wiring layer 152 and the presence or absence of hillocks.
Specifically, an Nd-containing Al film was formed on a silicon substrate, and the presence or absence of hillocks due to heat treatment was confirmed. The heat treatment conditions were a temperature of 400 ° C. and a treatment time of 1 hour. The presence or absence of hillocks was observed with an optical microscope.
When the Nd content was 0 at% (pure Al) and 1.3 at%, generation of hillocks was observed. On the other hand, the occurrence of hillocks was not observed when the Nd content was 2.0 at%. From this result, it is considered that hillocks can be reduced when the Nd content is about 1.5 at% or more.
When the Nd content was 2.0 at%, the treatment temperature was 30 minutes, and the temperature was changed to 200 ° C., 300 ° C., 400 ° C., and 430 ° C., no hillocks were observed at any temperature.

Based on the above knowledge, the acceleration sensor 100 was produced using an Al film having a Nd content of 2.0 at% as the wiring layer 152. As a result, a good result was obtained without disconnection or the like of the wiring 156 when the first and second structures 110 and 130 were formed. At this time, the thickness of the wiring layer 152 was 100 nm, the width of the wiring 156 was 6 μm, and the protective layer 153 was SiN having a thickness of 200 μm.
On the other hand, when the acceleration sensor 100 is formed using pure Al as the wiring layer 152, the wiring 156 may be disconnected.

(Other embodiments)
Embodiments of the present invention are not limited to the above-described embodiments, and can be expanded and modified. The expanded and modified embodiments are also included in the technical scope of the present invention.
For example, instead of Al—Nd, a material such as Al—Ta, Ti, Cr, or Mo can be used. That is, an alloy such as AlTa can be used as the wiring layer 152. As described above, Nd prevents the occurrence of hillocks in the wiring layer 152 by relieving internal stress due to recrystallization when the temperature is raised. The material capable of relaxing such internal stress is not limited to Nd.

It is a perspective view showing the acceleration sensor which concerns on one Embodiment of this invention. FIG. 2 is an exploded perspective view showing a state where the acceleration sensor of FIG. 1 is disassembled. It is a top view showing the state which looked at the acceleration sensor of Drawing 1 from the upper surface. It is a partial cross section figure showing the state which cut | disconnected the acceleration sensor of FIG. It is a partial cross section figure showing the acceleration sensor which concerns on the modification of this invention. It is a top view showing the state which looked at the wiring which cross | intersects from upper direction. It is sectional drawing showing the cross-sectional state of the wiring which cross | intersects. It is a top view showing the state which looked at the wiring which shifted crossing angle theta from 90 degrees from the upper part. It is a top view showing the state which looked at the wiring which shifted crossing angle theta from 90 degrees from the upper part. It is a top view showing the state which looked at the wiring which cross | intersects from the top. It is a top view showing the state which looked at the wiring which cross | intersects from the top. It is a top view showing the state which looked at the wiring which cross | intersects from the top. It is a top view showing the state which has the protrusion part and looked at the intersection part of the wiring which cross | intersects from the top. It is a top view showing the state which looked at only one side of wiring of Drawing 9A from the top. It is sectional drawing showing the state which cut | disconnected the wiring of FIG. 9A. It is a top view showing the state which looked at wiring which has a curve-like projection part from the top. It is a top view showing the state which looked at the intersection part of the wiring which cross | intersects from the top. FIG. 9B is a top view illustrating a state in which only one of the wirings in FIG. 9E is viewed from above. It is a top view showing the state which looked at the intersection part of the wiring which cross | intersects from the top. It is a top view showing the state which looked at only one side of wiring of Drawing 9G from the top. It is a top view showing the state which looked at the intersection part of the wiring which cross | intersects from the top. It is a circuit diagram which shows the structural example of the detection circuit for detecting the acceleration in a X-axis direction from the resistance of a piezoresistive element. It is a circuit diagram which shows the structural example of the detection circuit for detecting the acceleration in a Y-axis direction from the resistance of a piezoresistive element. It is a circuit diagram which shows the structural example of the detection circuit for detecting the acceleration in a Z-axis direction from the resistance of a piezoresistive element. It is a flowchart showing an example of the preparation procedure of an acceleration sensor. It is sectional drawing showing the state of the acceleration sensor in the preparation procedure of FIG. It is sectional drawing showing the state of the acceleration sensor in the preparation procedure of FIG. It is sectional drawing showing the state of the acceleration sensor in the preparation procedure of FIG. It is sectional drawing showing the state of the acceleration sensor in the preparation procedure of FIG. It is sectional drawing showing the state of the acceleration sensor in the preparation procedure of FIG. It is sectional drawing showing the state of the acceleration sensor in the preparation procedure of FIG. It is sectional drawing showing the state of the acceleration sensor in the preparation procedure of FIG. It is sectional drawing showing the state of the acceleration sensor in the preparation procedure of FIG. It is sectional drawing showing the state of the acceleration sensor in the preparation procedure of FIG. It is sectional drawing showing the state of the acceleration sensor in the preparation procedure of FIG. It is sectional drawing showing the state of the acceleration sensor in the preparation procedure of FIG. It is sectional drawing showing the state of the acceleration sensor in the preparation procedure of FIG. It is sectional drawing showing the state of the acceleration sensor in the preparation procedure of FIG. It is sectional drawing showing the state of the acceleration sensor in the preparation procedure of FIG. It is sectional drawing showing the state of the acceleration sensor in the preparation procedure of FIG.

Explanation of symbols

100 Acceleration sensor 110 First structure 111 Fixing part 112 Displacement part 113 Connection part 115 Opening part 116 Diffusion layer 120 Joining part 121 Joining part 122 Joining part 130 Second structure 131 Pedestal 132 (132a-133e) Weight part 133 Opening 134 Projection 140 Base 141 Bonding prevention layer 150 Wiring structure 151 Insulating layer 152 Wiring layer 153 Protective layer 154 Contact hole 155 Interlayer connection conductor 156 Wiring 157 Bonding pad 158 Pad opening Rx1-Rx4, Ry1-Ry4, Rz1-Rz4 Piezo Resistance element

Claims (12)

A fixed portion having an opening; a displacement portion disposed in the opening and displaced with respect to the fixed portion; and a connection portion connecting the fixed portion and the displacement portion; A first structure integrally formed of a first semiconductor material;
The first structure includes a weight part joined to the displacement part, a pedestal arranged so as to surround the weight part and joined to the fixed part, and made of a second semiconductor material. A second structure disposed in a stack on the body;
A piezoresistive element composed of a diffusion layer disposed in the connecting portion;
A first insulating layer covering the diffusion layer;
An interlayer connection conductor that penetrates the first insulating layer and is connected to the diffusion layer; and a first wiring that is disposed on the first insulating layer and connected to the interlayer connection conductor. And a conductor part including Al and Nd;
A second insulating layer covering the conductor portion;
A third insulating layer disposed between the conductor portion and the second insulating layer;
A second wiring disposed between the third insulating layer and the second insulating layer, intersecting the first wiring at an angle different from 90 ° and including Al and Nd;
An acceleration sensor comprising: The acceleration sensor according to claim 1, wherein the first wiring has a protrusion at a portion intersecting with the second wiring. The protrusion is
A first triangular shape having a first side corresponding to the wiring direction of the first wiring, and a second side and a third side not corresponding to the wiring direction of the first wiring;
A second triangular shape disposed on the second side and having fourth and fifth sides;
A third triangular shape disposed on the third side and having sixth and seventh sides;
The acceleration sensor according to claim 2, further comprising: fourth to seventh triangular shapes arranged on the fourth to seventh sides, respectively. The acceleration sensor according to any one of claims 1 to 3, wherein the Nd content of the conductor portion is 1.5 at% or more and 10 at% or less. The acceleration sensor according to any one of claims 1 to 4, wherein each of the first and second semiconductor materials is silicon. Impurities are introduced into the first layer of the semiconductor substrate in which a first layer made of a first semiconductor material, a second layer made of an oxide, and a third layer made of a second semiconductor material are sequentially stacked. Diffusing and forming a diffusion layer as a piezoresistive element;
Forming a first insulating layer on the first layer;
Forming an opening in the first insulating layer leading to the diffusion layer;
An interlayer connection conductor that penetrates the first insulating layer and is connected to the diffusion layer; and a first wiring that is disposed on the first insulating layer and connected to the interlayer connection conductor. And forming a conductor portion containing Al and Nd;
Forming a second insulating layer covering the conductor portion;
Heat-treating the semiconductor substrate to bring the diffusion layer and the interlayer connection conductor into ohmic contact;
Etching the first layer to fix a fixed portion having an opening, a displacement portion disposed in the opening and displaced relative to the fixed portion, and a connection portion connecting the fixed portion and the displacement portion When,
Forming a first structure having:
The second layer is etched to form a second structure having a weight part joined to the displacement part and a pedestal arranged to surround the weight part and joined to the fixed part. Steps to do,
Forming a second wiring on the second insulating layer that intersects the first wiring at an angle different from 90 ° and includes Al and Nd;
Forming a third insulating layer on the second wiring;
A method for manufacturing an acceleration sensor, comprising: The method of manufacturing an acceleration sensor according to claim 6, wherein the first wiring has a protruding portion at a portion intersecting with the second wiring. The protrusion is
A first triangular shape having a first side corresponding to the wiring direction of the first wiring, and a second side and a third side not corresponding to the wiring direction of the first wiring;
A second triangular shape disposed on the second side and having fourth and fifth sides;
A third triangular shape disposed on the third side and having sixth and seventh sides;
The method for manufacturing an acceleration sensor according to claim 7, further comprising: fourth to seventh triangular shapes arranged on the fourth to seventh sides, respectively. The method for manufacturing an acceleration sensor according to any one of claims 6 to 8, wherein the Nd content of the conductor portion is 1.5 at% or more. The method for manufacturing an acceleration sensor according to claim 6, wherein the heat treatment temperature is 380 ° C. or higher. 11. The method for manufacturing an acceleration sensor according to claim 6, wherein a temperature at the time of forming the third insulating layer is 300 ° C. or higher. The method of manufacturing an acceleration sensor according to any one of claims 6 to 11, wherein the first and second semiconductor materials are both silicon.

Images