JP4645673B2 - Variable capacitance element - Google Patents

Description

  The present invention relates to a mechanical variable capacitance element, and more particularly, to a small, high-performance, high-quality variable capacitance element and a method for forming the same in an electric circuit integrated with high density.

A conventional variable capacitance element will be described with reference to the micromechanical switch of FIG. 10 using a variable capacitance structure. FIG. 10 shows IEEE MTT-S Digest 1999, pp. It is a microwave switch introduced in 1923-1926. A gold contact portion 6 is provided at the lower end of the tip of the silicon cantilever 1 via an insulating layer 5, and a circuit terminal portion 7 that forms a closed circuit by contact with the contact portion is provided on a surface facing the contact portion 6. A driving electrode 8 is provided that applies an electrostatic force to the contact portion 6 to bend the silicon cantilever 1. The length of the silicon cantilever is about 200 μm, the width is about 20 μm, and the thickness is about 2.5 μm. The gap between the contact portion 6 and the circuit terminal portion 7 is set to 10 μm or less. When a voltage of 50 V or more is applied to the drive electrode 8, the beam 1 is bent and the contact portion 6 contacts the circuit terminal portion 7, and the contact is closed. It is done.
IEEE MTT-S Digest 1999, pp. 1923-1926

  However, since the voltage required to close the contact is as high as 50 V or more, it is necessary to mount a dedicated booster circuit, which hinders downsizing of the switch element. Further, the pad formed at the tip of the silicon beam has a larger area, so that the operation is slowed down due to the viscous resistance of atmospheric air when driven up and down, and a high-speed switching operation of several μs level becomes difficult.

  FIG. 11 shows a beam structure that can achieve low voltage driving and a switching speed of several μs level. The beam 1 has a width W = 2 μm, a thickness t = 2 μm, and a length L = 500 μm. An electrode 2 having a 0.1 μm thick insulating layer 5 formed on the surface of the beam 1 with a 0.6 μm gap is disposed on the substrate 4. When a voltage V is applied between the beam 1 and the electrode 2, the beam bends in the −z direction by an electrostatic force. When the pull-in voltage (pull-in voltage) or higher is reached, the electrostatic force is overwhelmingly increased rather than the restoring force of the beam, so that the beam 1 is rapidly adsorbed on the insulating layer 5. When the voltage is further increased, the beam 1 gradually increases the capacitance between the beam and the electrode while increasing the contact area with the insulating layer 5. In this way, by increasing the length of the beam, the spring property of the beam is weakened, and by reducing the width of the beam, the viscous resistance of the air is reduced to achieve low voltage drive and a switching speed of several μs level. it can. When aluminum having a Young's modulus of 77 GPa was used as the material of the beam, the pull-in voltage was 0.25 V when the beam was cantilevered and 1.72 V when the beam was both supported.

  However, the problems that become prominent with such an elongated beam shape are (1) residual stress, (2) thermal expansion, and (3) stiction.

  The first problem of residual stress will be described. For forming a minute beam, a thin film structure using a semiconductor process, a joining structure of a thin rolled material, or the like is used. In either case, residual stress inside the beam becomes a problem. There are two types of residual stress, one is compressive / tensile stress in the length direction of the beam, and the other is a stress gradient in the thickness direction of the beam.

  For example, if the beam in FIG. 11 is a doubly-supported beam, if excessive compressive stress remains in the x and y directions in the figure, the stress release in the y direction does not cause a significant change in the beam shape, but the beam end face is restrained. With respect to the x direction, the beam is buckled in an attempt to release the stress, and the beam bends regardless of the application of electrostatic force. Conversely, when tensile stress remains, the beam 1 does not seem to change, but as shown in the graph of FIG. 12, the pull-in voltage increases as the residual tensile stress increases, and the driving characteristics of the beam change significantly. End up. That is, it is ideal that the beam is generated with a residual stress of 0. However, if the internal stress cannot be accurately reproduced in the beam manufacturing process, buckling and pull-in voltage variations are caused, and the quality of the element deteriorates. It should be noted that this kind of stress is released for the cantilever beam, so that no buckling or pull-in voltage variation occurs.

  However, when the beam 1 of FIG. 11 is a cantilever beam, if a stress gradient exists in the z direction, that is, the thickness direction of the beam, the beam is warped by stress release. For example, if a positive stress gradient of 2 MPa / μm exists along the z direction inside the beam, the tip of the beam 1 warps by about 2 μm. If this stress gradient value cannot be accurately reproduced in the beam manufacturing process, the degree of warpage will vary, and variations in capacity reduction and increase in pull-in voltage due to an increase in the distance between the beam 1 and the electrode 2 can be suppressed. Disappear. For example, the pull-in voltage when the stress gradient is 0 and there is no warp is 0.25 V, while the pull-in voltage when the tip is raised by 2 μm increases to 1.2 V.

  It is very difficult to control the compressive / tensile stress in the length direction of the beam and the stress gradient in the thickness direction in the manufacturing process. There is "Yananami" as a stress relaxation method in the manufacturing process, but in order to expose the element to a high temperature, the temperature is changed to various members other than the beam constituting the element, for example, the metal of the electrode or the bridge to the bridge structure. Therefore, since the sacrificial layer material or the like that is temporarily provided under the beam and is finally etched is limited to a temperature that does not change the material properties, the stress cannot be completely relieved.

  The second problem of thermal expansion will be described. Although the beam thermally expands in the length direction due to an increase in the ambient temperature of the element, the beam buckles in a double-supported beam structure in which both ends are constrained, and the beam bends regardless of the application of electrostatic force.

  Describe the third problem of stiction. FIG. 13 shows a configuration in which the beam 1 of FIG. 11 is a double-supported type, and shows the relationship between the voltage and the capacity when the residual stress is suppressed to approximately zero. When a voltage is applied, pull-in occurs at 1.72 V, and when a voltage higher than that is applied, the beam 1 and the electrode 2 come into contact with each other through the insulating layer 5 to increase the contact area and increase the capacity. Conversely, when the voltage is lowered, the contact between the beam and the electrode is not eliminated even if the voltage is lowered to 0.64V. This is because the spring property of the beam, that is, the restoring force of the spring is weak. This means that even if the voltage is returned to 0 V, if there is an adsorption force through water molecules in the atmosphere, an adsorption force due to a slight residual charge, or a van der Waals force in the contact region, the beam is in the initial state. This means that there is a high possibility that it will not be possible to return to. In order to avoid this, it is necessary to adopt a complicated structure such as a mechanism for forcibly driving the beam in a direction away from the electrode, for example, an electrode for pulling back the upper surface of the beam 1 in FIG. .

  It is an object of the present invention to implement a high-quality variable capacitance function with a simple configuration, for example, as a micromechanical switch, in such a mechanical variable capacitance element capable of low voltage and high speed driving.

  The present invention is composed of a flexible beam and an electrode installed in the vicinity of the beam to form a capacitor between the beam, and an electrostatic force is applied to the beam by applying a voltage between the beam and the electrode. The variable capacitance element is characterized in that the capacitance between the two is changed by bending the electrode with the electrode, and the bent beam and the electrode are brought into contact with each other via an insulating layer formed on at least one surface. The capacitance is changed by changing the contact area and the electrode is divided into a plurality of parts. In addition, a recess is formed in the insulating layer in which a part of the electrode is lower than the height of the other electrode surface, a voltage is applied between the beam of the recess and the other electrode, and the beam is drawn into the recess to A force is generated that separates the opposite part of the beam from the electrode, with the step generation part as a fulcrum. Therefore, by dividing the electrode that bends the beam by electrostatic force into a plurality of parts, each of which has a function to support the beam tip, a function as an AC signal line, and a function to eliminate stiction, with a simple configuration Degradation of quality and performance due to residual stress, thermal expansion, and stiction can be suppressed.

  As described above, according to the present invention, the electrode is divided into a plurality of parts, and each of them has a function of supporting the beam tip, a function as an AC signal line, and a function of eliminating stiction. Thus, it is possible to suppress deterioration of quality and performance due to residual stress, thermal expansion, and stiction. As a result, a small and high quality variable capacitor capable of low voltage and high speed driving and an RF switch using the same can be realized.

  Hereinafter, embodiments of the present invention will be described with reference to FIGS.

(Embodiment 1)
FIG. 1 is a schematic diagram of a variable capacitance element according to Embodiment 1 of the present invention. The beam 1 is a cantilever whose one end is fixed on the substrate 4 by the anchor portion 3 and has a thickness t = 2 μm, a width W = 2 μm, and a length L = 500 μm. Here, aluminum having a Young's modulus of 77 GPa is used as the material of the beam. The electrodes 2 a and 2 b are fixed on the substrate 4 so as to be parallel to the side surface of the beam 1. The length la of the electrode 2 a is about 50 μm, the length of the electrode 2 b is about 450 μm, and the insulating layer 5 is opposed to the beam of the electrode 2 in order to avoid an electrical short circuit between the beam 1 and the electrode 2. About 0.1 μm is formed on the surface. Here, the height h2 of the upper surface of the electrode 2 is set higher than the height h1 of the upper surface of the beam 1.

  Since the beam 1 is a cantilever beam, the tensile / compressive stress in the x direction and the y direction and the stress gradient in the z direction can be released. FIG. 2 shows the actual beam shape after stress release. Since the stress gradient in the z direction remains about 2 MPa / μm, the tip warps in the z direction by about 2 μm. The height of the electrode 2 has a height h2 enough to compensate for it. Further, the facing area between the beam 1 and the electrode 2 does not change, and therefore the capacitance does not decrease. When the beam 1 is grounded and a voltage is applied to the electrode 2 as shown in FIG. 2, the beam is pulled in the y direction to cause pull-in at 0.25 V, and the voltage is slightly returned to 0.20 V, so that the top view of FIG. As shown, the beam tip is fixedly supported on the electrode 2a. Since cantilever beams are easily affected by external vibrations, the both ends fixed support state shown in Fig. 3 is normally maintained, the voltage is released as needed to return to the cantilever state, and internal stress is generated due to thermal expansion. It is possible to perform a refresh operation that alleviates this.

  If h2 is equal to h1 in FIG. 1, the opposing area between the beam 1 and the electrode 2 decreases due to warpage of the beam 1 in the z direction due to stress release, and the pull-in voltage becomes as high as 0.42V. From this, it can be seen that an effect of suppressing an increase in the pull-in voltage can be obtained by setting h2 higher than h1.

  In the configuration of FIG. 2, the electrode 2 for bending the beam 1 by applying an electrostatic force is divided, and the function as an AC signal line in which the input port Pi and the output port Po are provided on the electrode 2b which is a part of the electrode 2b. An AC signal circuit is formed. Although it is possible to consider the configuration of FIG. 2 as a simple actuator and to provide a movable member connected to the beam 1 and to form a variable capacitor dedicated to an AC signal with this movable member, the structure becomes complicated and the beam Since the mass of the movable member including 1 increases, there is a problem that the operation speed, particularly the switching speed when used as a switch is lowered. Therefore, in the embodiment of the present invention, as shown in FIG. 2, a structure in which a part of the electrode 2 is also used as an AC signal line is adopted, the configuration is simplified, and high-speed switching is achieved.

  When a voltage is further applied from the state of FIG. 3, the contact area between the beam 1 and the electrode 2b is increased through the insulating layer 5, and the capacitance is increased. Since the beam 1 and the electrode 2b form a shunt-type switch, when a voltage of 3.8 V is applied as shown in FIG. 4, 80% of the opposed area of the beam 1 and the electrode 2 is in a contact state, and Pi− A state in which the AC transmission line between Pos is shunted to zero potential can be created.

  In the state shown in FIG. 3, that is, when the shunt is not grounded, if the capacitance in the vicinity of the part A where the beam 1 and the electrode 2b are close to each other cannot be ignored as the parasitic capacitance, the configuration shown in FIG. Can do. Here, the electrode 2a is brought closer to the beam 1 than 2b, and the electrode 2b is divided into 2b 'and 2b "so that the AC signal passes through the electrode 2b" located farther from the part A. Since the distance between the beam 1 and the electrode 2b is increased by the configuration of FIG. 5, the parasitic capacitance component can be suppressed. At this time, the electrode 2b 'cannot be omitted because it plays a role of driving the beam.

  Next, the manufacturing process used to construct FIG. 1 which is a schematic diagram of the present embodiment will be described. FIG. 6 is a process sectional view using the A-A ′ section of FIG. 1. By thermally oxidizing the high resistance silicon substrate 9, the silicon oxide film 10 is formed with a thickness of 300 nm. Thereafter, a silicon nitride film 11 is deposited with a film thickness of 200 nm using a low pressure CVD method. Further, a silicon oxide film 12 is deposited with a thickness of 50 nm by using a low pressure CVD method. See Figure 6a).

  Thereafter, a sacrificial layer made of a photoresist is spin-coated, exposed and developed on the silicon oxide film 12 at a film thickness of 2 μm, and then the sacrificial layer 13 is formed by baking at 140 ° C. for 10 minutes on a hot plate. See Figure 6b).

  Thereafter, as shown in FIG. 6c), aluminum 14 is deposited on the entire surface of the substrate to a thickness of 2 μm by sputtering, and a pattern 15a is formed by photoresist so that the resist remains in a predetermined region, and the resist pattern 15a is penetrated. The height of a part of the aluminum 14 is lowered by etching from the hole.

  Next, as shown in FIG. 6d), a photoresist pattern 15b is formed again.

  Next, the beam 16 is formed by performing dry etching of aluminum using the photoresist pattern 15b as a mask, and the photoresist pattern 15b and the sacrificial layer 13 are removed by oxygen plasma. As a result, a beam having the substrate surface and the gap 17 is formed. See Figure 6e).

  Furthermore, as shown in FIG. 6f), a silicon nitride film 18 is deposited on the entire surface by plasma CVD to a thickness of 50 nm, thereby forming a silicon nitride film 18 on the silicon oxide film 12 on the substrate surface and around the beam 16.

  Finally, as shown in FIG. 6g), the silicon nitride film is etched back by a dry etching method having anisotropy at a film thickness equal to or greater than the deposited film thickness, for example, 100 nm, under a condition having a selectivity with the silicon oxide film. A beam 19 having no silicon nitride film on the upper surface and a nitride film remaining on the side surface is formed, and electrical conduction can be obtained from the upper surface of the beam.

  Note that although a high-resistance silicon substrate is used as the substrate in this embodiment mode, a normal silicon substrate, a compound semiconductor substrate, or an insulating material substrate may be used.

  Further, although the silicon oxide film 10, the silicon nitride film 11, and the silicon oxide film 12 are formed as insulating films on the high resistance silicon substrate 9, the formation of these insulating films may be omitted if the resistance of the substrate is sufficiently high. . In addition, the silicon oxide film 10, the silicon nitride film 11, and the silicon oxide film 12 are formed on the silicon substrate as an insulating film having a three-layer structure, but the silicon nitride film 11 has a thickness of the silicon nitride film deposited on the beam. If the film thickness is sufficiently thick, that is, a film thickness that does not disappear even after a so-called etch-back process, the silicon oxide film 12 forming process can be omitted.

  Note that although aluminum is used as a material for forming the beam in this embodiment mode, other metal materials Mo, Ti, Au, Cu, and a semiconductor material into which impurities are introduced at a high concentration, for example, amorphous silicon, a highly conductive material Molecular materials or the like may be used. Further, although sputtering is used as a film forming method, it may be formed using a CVD method, a plating method, or the like.

(Embodiment 2)
FIG. 7 shows the structure of FIG. 11 in which the beam 1 is a double-supported beam, and the voltage is released after the beam 1 is once brought into contact with the insulating layer 5 by pull-in, but the central portion of the beam cannot be detached from the insulating layer 5 by stiction. Is shown. An embodiment of the present invention for solving this is shown in FIG. FIG. 8A is an enlarged view of the central portion of the beam 1 causing stiction. The beam 1 is in surface contact with the insulating layer 5 on the electrode 2. Adsorption force via water molecules in the atmosphere, electrostatic force due to residual charge, van del Walls force, etc. act on this contact surface. The electrode 2 ′ forms a concave portion farther from the beam than the electrode 2 and does not contact the beam. Here, when a voltage is applied to the electrode 2 ′, the beam 1 is drawn into the concave portion as shown in FIG. 8B, and a force that separates the attachment of the beam 1 to the electrode 2 with the stepped portion B serving as a fulcrum is exerted. Can be eliminated. FIG. 9 shows a structure in which the surfaces of both the electrodes 2 and 2 ′ are in the same plane, but a recess is provided only in the insulating layer, and the effect of eliminating stiction can be obtained similarly.

  In addition, since the stiction due to water molecules in the atmosphere can be reduced by vacuum-sealing the variable capacitance element of this embodiment, the stiction elimination function of the structure shown in this embodiment is more effective. .

External view of variable capacitance element in first embodiment Diagram showing beam shape change due to release of internal stress of beam The top view showing the state which supported the beam tip in the 1st example by electrostatic force The top view showing the state which gave the shunt type switch to the AC signal line in the first embodiment The top view showing the structure which reduces the parasitic capacitance of a signal line in a 1st Example Manufacturing process diagram by the thin film manufacturing technique having the configuration in the first embodiment A diagram showing the state where stiction occurred on both sides of the beam Principle diagram of stiction elimination structure in the second embodiment The figure showing the structure which formed the recessed part with the level | step difference only of an insulating layer in 2nd Example. The figure which shows the structure of the microwave switch which applied the conventional variable capacity structure Figure showing a conventional variable capacitance structure capable of low-voltage and high-speed operation 10 shows the relationship between the internal residual stress in the x and y directions inside the beam and the pull-in voltage in the structure of FIG. A diagram showing the relationship between applied voltage and capacitance change in the structure of FIG.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 Beam 2 Electrode 3 Anchor part 4 Board | substrate 5 Insulation layer 6 Contact part 7 Circuit terminal part 8 Drive electrode 9 Silicon substrate 10 Silicon oxide film 11 Silicon nitride film 12 Silicon oxide film 13 Sacrificial layer 14 Aluminum 15 Resist pattern 16 Beam 17 Beam- Gap between substrates 18 Silicon nitride film 19 Beam from which nitride film on top surface is removed

Claims (3)

A beam having flexibility, and an electrode disposed in proximity to the beam to form a capacitor between the beam and applying a voltage between the beam and the electrode to A variable capacitance element that changes capacitance between the two by bending with electric power, wherein the electrode is divided into a plurality of parts, and a part of the electrode adsorbs the free end of the beam with electrostatic force. The distance between the beam that supports the free end of the beam with an electrostatic force and the beam is arranged closer to the beam than other electrodes, and a part of the plurality of electrodes is exchanged A variable capacitance element characterized by being a signal line. A variable capacitance element obtained by vacuum-sealing the variable capacitance element according to claim 1. A switch using the variable capacitance element according to claim 1.

Images