JP2004312784A - Oscillator filter and method of manufacturing same - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide an oscillator filter that can achieve a high-performance characteristic through a simple structure and process. <P>SOLUTION: The oscillator filter comprises: a silicon substrate 40; an oscillator 41, which is provided by patterning a polycrystalline silicon layer which is deposited on the oscillator 41, in a rectangular pattern; an input terminal electrode 42; and an output terminal electrode 43. Sides of the oscillator 41 face the side of the input terminal electrode 42 and the side of the output terminal electrode 43 with minute gaps 47 therebetween. When a voltage is applied to the input terminal electrode 42, an electrostatic force acts to allow the oscillator 41 to oscillate horizontally. <P>COPYRIGHT: (C)2005,JPO&NCIPI

Description

  The present invention relates to a micro-mechanical device using surface micro-machine technology.

  Electrically controlled switch elements used in various electronic devices include semiconductor switches and reed relays having mechanical contacts. These have advantages and disadvantages from the viewpoint of an ideal switch.

  Semiconductor switches have advantages such as miniaturization, high speed and high reliability, and are easy to integrate as a switch array. For example, a PIN diode, a HEMT, a MOSFET, or the like is used as a switch for switching antennas for microwaves, millimeter waves, and the like. However, a semiconductor switch has a high on-impedance and a low off-impedance as compared with a switch that connects and opens a mechanical contact. There is also a large stray capacitance.

  On the other hand, reed relays have a higher on / off impedance ratio than semiconductor switches, can be designed to minimize insertion loss, and maintain signal fidelity. For this reason, they are widely used in, for example, semiconductor testers. However, the size is large and the switching speed is slow.

  On the other hand, recently, a micromechanical switch that has the advantages of a semiconductor switch and a reed relay has attracted attention. Above all, a micromechanical switch formed using surface micromachine technology and operated electrostatically can be formed using a semiconductor thin film process, and can be realized at low cost.

  FIG. 15 shows a plan view and a cross-sectional view taken along the line I-I 'of a conventionally proposed micromechanical switch. This switch has a source electrode 51 and a drain electrode 52 formed on a substrate 50 of silicon or the like, and a gate electrode 53 formed between the source electrode 51 and the drain electrode 52. A conductor beam 54 is formed on the gate electrode 53 in a state of floating with a predetermined gap.

  The conductor beam 55 becomes an anchor part 55 having one end fixed to the source electrode 51. The other end of the conductor beam 54 is an open end and forms a movable contact (contact chip) 56. When a voltage is applied to the gate electrode 53, the conductor beam 54 is displaced downward by electrostatic force, and the movable contact 56 contacts the drain electrode 52. When the gate voltage is removed, the conductor beam 54 returns to its initial position by restoring force.

  The result of analyzing the deflection of the conductor beam of this switch using a mechanical model was published by PM Zavracky. According to this, when a gate voltage is applied, the conductor beam 54 connected to the source electrode 51 is held at a position d (x) on the gate electrode 53 by electrostatic force as a longitudinal distance x from the source. . The gate voltage required to hold the conductor beam 54 in a flexed state increases monotonically with the flexure, and when the beam flexes beyond a certain degree, the gate voltage required to maintain the beam monotonically decreases. Becomes unstable, the beam is bent at a certain gate voltage (threshold voltage Vth), and the switch is closed.

  The threshold voltage Vth according to this model is expressed as Vth = (2/3) × d0 × √ (2 kd0 / 3ε0A). Here, d0 is the gap between the conductor beam 54 and the gate electrode 53 in the initial state, k is the effective spring constant of the conductor beam 5, and A is the area of the conductor beam 54 facing the gate electrode 53.

  As a result, the threshold voltage Vth increases the facing area A of the gate electrode 53 (increases the electrostatic force acting on the beam), reduces the spring constant k of the conductor beam, and reduces the gap d0 between the conductor beam and the gate electrode. It can be seen that it can be lowered. However, reducing the spring constant k reduces the maximum switching speed, and reducing the gap do increases the electrostatic coupling between the gate electrode and the signal line. Another method for reducing the threshold voltage Vth is to reduce the gap g between the movable contact 56 and the drain electrode 52 by increasing the amount of protrusion of the movable contact 56 downward. Thus, the switch can be closed before the unstable point is reached.

  From the above, it is indispensable to manufacture the gaps d0 and g with high accuracy in order to reduce the threshold voltage Vth, but to manufacture this micromechanical switch requires a complicated process. More specifically, a source electrode 51, a drain electrode 52, and a gate electrode 53 are first formed on a substrate by patterning. Next, a sacrificial layer such as a silicon oxide film is deposited on these electrodes. This sacrificial layer is patterned in two steps. In the first step, the sacrifice layer is partially etched in order to form the contact chip portion 56. In the second step, the sacrifice layer is etched to reach the source electrode 51 in order to form the anchor portion 55. Subsequently, a conductor layer is deposited on the sacrificial layer and is patterned. Finally, the sacrificial layer is etched away to separate the conductor beam 54 from the substrate.

The following four lithography steps (mask steps) are required for the above manufacturing steps.
(1) Patterning of the source electrode etc. (2) Patterning of the contact tip portion of the sacrificial layer (3) Patterning of the anchor portion of the sacrificial layer (4) Patterning of the conductor layer And a bandpass filter of about 100 MHz has been manufactured (for example, Non-Patent Document 1). The advantages of the mechanical oscillator filter are that the Q value is extremely high as compared with the electric LC filter, and the size can be extremely small as compared with the dielectric filter and the SAW filter.

  FIG. 16 is a plan view showing a unit configuration of such a resonator filter and a cross-sectional view taken along the line I-I '. A vibrator 61, an input terminal 62 and an output terminal 63 are formed on a substrate 60 by a micromachine technique. The vibrator 61 is integrally formed of polycrystalline silicon together with the four support beams 64a to 64d, and ends of the support beams 64a to 64d are fixed to anchors 65a, 65b, 65c, and the vibrator 61 is held in a floating state. Have been.

  The input terminal 62 is formed of the same polycrystalline silicon film as the vibrator 61, but a base metal extends to immediately below the vibrator 61 to form a gate electrode (drive electrode) 66. The output terminal 63 and the vibrator 61 are formed on a common metal electrode 67. Actually, by connecting a plurality of such unit oscillator filters in parallel, a mechanical filter having a predetermined pass bandwidth is produced.

The vibrator 61 vibrates up and down by driving the drive electrode 66. The resonance frequency f0 of the vibrator 61 is expressed as f0 = (1 / 2π) √ (k / m) using the spring constant k and the mass m of the vibrator 61. In the structure and dimensions shown in FIG. 16, since k = 3Eh ³ b / l ³ and m = ρLwh, f0 = (1 / π) √ (Eh ² b / ρLwl ³ ). Here, E is the Young's modulus of the vibrator, ρ is the density, and in the case of silicon, E = 1.7 × 10 ¹¹ Pa and ρ = 2.33 × 10 ³ kg / m ⁻³ .

  Specifically, f0 = 92 MHz was obtained by setting the dimensions in FIG. 16 to L = 13.1 μm, l = 10.4 μm, w = 6 μm, h = 2 μm, and b = 1 μm.

  On the other hand, a frequency band such as 800 MHz to 5 GHz is used in a portable terminal or the like, and a mechanical filter with a higher frequency is desired for such a use. FIG. 17 shows a configuration example of such a high-frequency receiving unit, which includes a band-pass filter 171, a low-noise amplifier 172, a band-pass filter 173, a mixer 174, and the like. The mixer 174 is controlled by a phase control circuit 175 including a PLL (Phase-Locked Loop) / VCO (Voltage Controlled Oscillator). A mechanical filter is also desired for the band-pass filters 171 and 173 of the receiving unit and the PLL / VCO of the phase control circuit 175.

In order to further increase the frequency with the filter structure of FIG. 16, it is conceivable to increase h, increase b, decrease L and l, etc., but it is not easy in the current semiconductor process to increase the frequency. Also, the structure and process are complicated.
C. Nguyen, et al., 12th International IEEE Micro Mechanical Systems Conference, 1999, pp.453-458

  As described above, the micromechanical switch proposed so far has a complicated manufacturing process, and it is difficult to obtain a low threshold voltage characteristic. In particular, since the gap g between the contacts depends on processes such as the thickness of the sacrificial layer and the etching amount of the sacrificial layer, it has been difficult to reduce the threshold voltage Vth.

  In the case of micromechanical vibrators as well, conventionally proposed structures and manufacturing processes are complicated, and it is difficult to increase the frequency.

  The present invention has been made in view of the above circumstances, and has as its object to provide a vibrator filter capable of obtaining high-performance characteristics with a simple structure and process, and a method of manufacturing the same.

  An oscillator filter according to the present invention includes a substrate, an input terminal electrode and an output terminal electrode formed at a predetermined interval on the substrate, and a portion between the input terminal electrode and the output terminal electrode on the substrate. It is made of the same material, and both side surfaces thereof face the side surfaces of the input terminal electrode and the output terminal electrode with a small gap, and can be displaced in a direction parallel to the substrate by a columnar support beam fixed to the substrate. And a vibrator held in the main body.

  As described above, according to the present invention, by forming a movable part such as a vibrator so as to be displaced in a lateral direction, a vibrator filter having high performance with a simple structure and a simple process and a manufacturing method thereof. You can get the way.

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

[Embodiment 1]
FIG. 1 is a plan view of a micromechanical switch according to an embodiment of the present invention. FIGS. 2A and 2B are cross-sectional views taken along lines II ′ and II-II ′ of FIG. 1, respectively. Also, a III-III 'cross-sectional view is shown.

  This micro mechanical switch is formed on the silicon substrate 10 by a surface micro machine technology. Both ends of the beam 11 are anchor portions (fixed portions) 12 fixed to the substrate 10, and the other portions are formed so as to float from the substrate 10. Here, an example is shown in which three beams 11 are arranged in parallel. However, at least a pair of beams may be provided, or more beams may be repeatedly arranged.

  A movable contact 13 is formed in a pattern at a central portion of the beam 11 in the longitudinal direction. In this example, two drive electrodes (gate electrodes) 14 are arranged between the beams 11 while being fixed to the substrate 10. The beam 11, the anchor portion 12 and the movable contact 13 formed continuously with the beam 11, and the drive electrode 14 formed separately therefrom are all composed of a polycrystalline silicon layer 21 and a metal layer superposed thereon. The laminated structure 22 is patterned.

  The hole 23 formed in the movable contact 13 is formed through the metal layer 22 and the polycrystalline silicon layer 21 so that the sacrificial layer formed as an underlayer of the beam 11 can be efficiently formed as described later. This is for removing by etching. In other words, the area of the movable contact 13 is larger than the main body of the beam 11, and it takes time to remove the sacrificial layer only from the periphery by lateral etching. Therefore, in order to shorten the etching time of the sacrifice layer, the sacrifice layer etching through the hole 23 is used.

  In such a configuration, when a gate voltage is applied to a predetermined drive electrode 14 with the anchor portion 12 as a reference potential, the beams 11 forming a pair with the drive electrode 14 interposed therebetween are attracted by electrostatic force, and the movable contact 13 Are displaced laterally and come into contact, causing a short circuit. In this case, since the pair of beams 11 is displaced so as to be attracted to each other, even if the gap d0 between the beam 11 and the drive electrode 14 is smaller than the gap g between the adjacent movable contacts 13, the beam 11 and the The movable contact 13 can be short-circuited without the drive electrode 14 being short-circuited.

  In the conventional method in which one beam is displaced in the vertical direction as shown in FIG. 15, in order to obtain a low threshold, the gap g between the contacts is smaller than the gap d0 between the gate electrode and the beam. In the present embodiment, the gap d0 between the beam 11 and the drive electrode 14 is made smaller than the gap g between the contacts in order to utilize the lateral displacement of the beam 11 forming a pair. As a result, a low threshold characteristic can be obtained.

  In this embodiment, three beams 11 are arranged symmetrically with respect to the center beam. Therefore, a double contact relay can be realized using the drive electrodes 14 on both sides of the center beam.

  Furthermore, the contact area of the movable contact 13 can be freely set by the thickness of the metal layer 22, and high reliability can be obtained.

  The manufacturing process of the micro relay switch according to this embodiment will be described with reference to FIGS. 2A and 2B correspond to the cross sections of FIGS. 2A and 2B, respectively.

  As shown in FIG. 4, a sacrificial layer 31 is first deposited on the silicon substrate 10 to a thickness of about 1 μm. Specifically, as the sacrificial layer 31, an insulating layer such as a silicon oxide film or a silicon nitride film, which is a material having a high etching selectivity with respect to the beam constituent material and the substrate 10, is used. Then, the sacrificial layer 31 is selectively etched to form a hole 32 that reaches the substrate 10 at a location that needs to be fixed to the substrate, such as an anchor portion or a gate electrode portion to be formed later.

  Next, as shown in FIG. 5, a polycrystalline silicon layer 21 serving as a base material of the switch member is deposited by about 1 μm. Then, this polycrystalline silicon layer is selectively etched to form a beam 11, a movable contact 13 and an anchor portion 12 (not shown in FIG. 6) continuous with the beam 11, The electrode 14 is patterned. At this time, several holes 23 used for etching the sacrificial layer 31 later are formed in the movable contact 13 at the same time.

  Next, as shown in FIG. 7, a metal layer 22 is deposited on the entire surface by about 1 μm. Then, as shown in FIG. 8, the metal layer 22 is selectively etched with substantially the same pattern as that of the polycrystalline silicon layer 21 so that the beam 11, the anchor portion 12, the movable contact 13 and the drive electrode 14 are formed. As a laminated structure. However, the movable contact 13 is patterned so as to project laterally from the edge of the polycrystalline silicon layer 21 of the base material so as to form a minute gap between the movable contact 13 and the adjacent one. In the movable contact 13, holes communicating with the holes 23 formed in the underlying polycrystalline silicon layer 21 are patterned.

  Finally, the sacrifice layer 31 is removed by etching to form a state in which the beam 11 and the movable contact 13 float from the substrate 10 as shown in FIG. The movable contact 13 is formed wider than the beam 11, but since the etching of the sacrificial layer 31 proceeds simultaneously from the periphery through the hole 23, the sacrificial layer 31 can be removed by etching in a relatively short time. It is possible.

In the case of this embodiment, the lithography process is the following three processes.
(1) Patterning of sacrificial layer 31 (FIG. 4)
(2) Patterning of polycrystalline silicon layer 21 (FIG. 6)
(3) Patterning of the metal layer 22 (FIG. 8)
Therefore, the manufacturing process is simpler than the conventional method using two conductor layers. According to this embodiment, since the gap between the contacts and the gap between the beams are not influenced by the thickness of the sacrificial layer and the etching amount and are determined by the lithography accuracy, a minute gap can be obtained with high accuracy. Thereby, a low threshold voltage relay switch is obtained.

  In the above embodiment, the anchor portion 12 and the drive electrode portion 14 are fixed to the substrate 10 by removing the sacrificial layer 31 in advance. On the other hand, without patterning the sacrifice layer 31, the anchor portion 12 and the drive electrode portion 14 can be fixed to the substrate 10 via the sacrifice layer 31. FIG. 10 shows a cross-sectional view corresponding to FIG. 3 in that case.

  By increasing the area of the anchor portion 12, the anchor portion 12 can be left even when the sacrificial layer 31 under the beam 11 is etched away. Similarly, the drive electrode 14 is fixed to the substrate 10 via the sacrificial layer 31. Although the area of the movable contact 13 is also large, the sacrificial layer 31 can be completely removed by opening the hole 23 for etching the sacrificial layer.

  In this way, one less lithography step is required and the process becomes simpler.

  As described above, in the case of the above embodiment, the anchor portion 12 and the drive electrode portion 14 are fixed to the substrate by removing the sacrificial layer 31 in advance. Therefore, if the etching time is sufficiently long under the condition that the etching selectivity between the sacrifice layer and the other material portion can be sufficiently obtained without forming the hole 23 in the movable contact 13 portion, the sacrificial layer below the sacrificial layer can be obtained. It is possible to remove the layer only by lateral etching. However, in the case where the anchor portion 12 and the drive electrode portion 14 are fixed while leaving the sacrificial layer 31 thereunder, if the area of the movable contact 13 is substantially the same as that of the anchor portion 13 or the drive electrode 14, the movable contact 13 In order to remove the sacrifice layer 31 below, it is indispensable to open the hole 23.

[Embodiment 2]
In the above embodiment, both ends of the beam are fixed, but a cantilever type in which only one end of the beam is fixed to the anchor portion may be used. A plan view of such an embodiment is shown in FIG. 11 corresponding to FIG. Portions corresponding to those in the previous embodiment are denoted by the same reference numerals. 11A and 11B are the same as FIGS. 2A and 2B. The sectional view taken along the line III-III 'of FIG. 11 is as shown in FIG. 12, and the movable contact 13 is an open end.

  With such a cantilever type, the device area can be reduced as compared with the previous embodiment.

[Embodiment 3]
FIGS. 13A and 13B are a plan view and an II ′ cross-sectional view of an embodiment in which the present invention is applied to a micromechanical resonator filter. The figure shows a unit oscillator filter, which includes a silicon substrate 40, a resonator 41 obtained by forming a polycrystalline silicon layer deposited thereon in a rectangular pattern, an input terminal electrode 42, And an output terminal electrode 43. An actual mechanical filter is configured to have a predetermined pass bandwidth by arranging a plurality of such unit oscillator filters.

  The vibrator 41 is disposed between the input terminal electrode 42 and the output terminal electrode 43, and is fixed to the substrate 40 by a columnar support beam 44 at one or a plurality of locations (four locations in the embodiment). The deflection in the horizontal direction (direction parallel to the substrate) is possible due to the deflection. The input terminal electrode 42 and the output terminal electrode 43 are fixed to the substrate 40 by fixing portions 45 and 46 having large areas.

  Both side surfaces of the vibrator 41 face the side surfaces of the input terminal electrode 42 and the output terminal electrode 43 with a small gap 47, respectively. When a voltage is applied to the input terminal electrode 42, an electrostatic force acts on the vibrator 41, and the vibrator 41 can vibrate in the lateral direction. The vibrator 41 has a natural vibration frequency (resonance frequency) determined by the spring constant of the support beam 44 and the mass of the vibrator body on the support beam 44. Therefore, when an AC voltage is applied to the input terminal electrode 42, the input AC voltage resonates when the input AC voltage is at the natural oscillation frequency of the vibrator 41, and a voltage having the opposite phase appears at the output terminal 43, indicating a filter function.

Specifically, assuming that the sides of the prism of the support beam 44 are a and b and the height is 1, the spring constant is k = 4Ea ³ b / l ³ . When the size of the surface of the vibrator 41 is L × w, the thickness is h, and the density is ρ, the resonance frequency is f0 = (1 / π) √ (Ea ³ b / ρLwhl ³ ). For example, when L = 4 μm, w = 4 μm, h = 1 μm, 1 = 0.75 μm, and a = b = 1 μm, f0 = 1.05 GHz.

One of the reasons that the method of this embodiment is easier to increase the frequency than the conventional method shown in FIG. 16 is that the manner in which the thickness h of the vibrator contributes to the resonance frequency f0 is different. That is, in the conventional method, the resonance frequency f0 is proportional to the thickness h of the vibrator. Considering that the resonance frequency is increased 10 times only by the thickness h of the vibrator, for example, the thickness of 10 μm is increased to 100 μm, which is not easy. On the other hand, in the method of this embodiment, the resonance frequency f0 is inversely proportional to √l ³ (1 is the thickness of the sacrifice layer), and it is easy to reduce the thickness h of the vibrator for higher frequency. is there. In addition, the two-dimensional dimension for determining the resonance frequency can be selected in the processing range of the normal semiconductor process as in the above-described example, and it is easy to increase the frequency.

  Therefore, according to this embodiment, a high-frequency filter useful for a portable terminal or the like can be compactly configured.

  The manufacturing process of the filter according to this embodiment will be described with reference to FIG. As shown in FIG. 14A, a sacrifice film 48 is formed on the silicon substrate 40 by patterning. The sacrificial film 48 is, for example, a silicon oxide film, and an opening is formed in a pattern in a support beam portion which is a fixing portion of the input / output terminal electrodes 42 and 43 and a fixing portion of the vibrator 41. Then, as shown in FIG. 14B, a polycrystalline silicon layer 49 and an electrode film 71 are deposited. Next, as shown in FIG. 14C, the electrode film 71 is patterned so as to remain only in the input / output terminal portion, and further the polycrystalline silicon layer is patterned to form the input terminal electrode 42, the vibrator 41, and the output terminal electrode. 43 is formed separately. Finally, the sacrificial layer 48 is removed by etching to complete the process.

  In the case of this embodiment, the lithography process is performed three times: patterning the sacrificial layer 48, patterning the electrode film, and patterning the polycrystalline silicon layer 49. Therefore, the process is very simple. The above-mentioned dimensions can be easily realized by the current semiconductor manufacturing process technology.

  Note that the vibrator and the input / output terminal electrodes are not limited to polycrystalline silicon, and may be configured using other appropriate conductor layers.

1 is a plan view of a micro relay according to an embodiment of the present invention. FIG. 2 is a sectional view taken along lines I-I ′ and II-II ′ of FIG. 1. FIG. 3 is a sectional view taken along the line III-III ′ of FIG. 1. It is a figure which shows the sacrificial film formation process and the sacrificial film patterning process in the manufacturing process of the micro relay of the embodiment. FIG. 3 is a diagram showing a polycrystalline silicon layer forming step of the embodiment. FIG. 4 is a diagram showing a polycrystalline silicon layer patterning step of the embodiment. FIG. 3 is a diagram showing a metal layer forming step of the embodiment. FIG. 3 is a diagram showing a metal layer patterning step of the embodiment. It is a figure showing a process of a sacrifice layer removal of the embodiment. It is sectional drawing of the micro relay by another embodiment. It is a top view of a micro relay by other embodiments. FIG. 13 is a sectional view taken along the line III-III ′ of FIG. 11. It is the top view which shows the structure of the resonator filter by other embodiment, and its I-I 'sectional drawing. FIG. 3 is a cross-sectional view showing a manufacturing step of the embodiment. It is the top view which shows the structure of the conventional micro mechanical switch, and its I-I 'sectional drawing. It is the top view which shows the structure of the conventional resonator filter, and its I-I 'sectional drawing. FIG. 3 is a diagram illustrating a configuration example of a high-frequency receiving unit.

Explanation of reference numerals

  DESCRIPTION OF SYMBOLS 10 ... silicon substrate, 11 ... beam, 12 ... anchor part, 13 ... movable contact, 14 ... drive electrode, 21 ... polycrystalline silicon layer, 22 ... metal layer, 23 ... hole, 31 ... sacrificial layer, 40 ... silicon substrate, 41: vibrator, 42: input terminal electrode, 43: output terminal electrode, 44: support beam, 45, 46: fixed portion, 47: gap, 48: sacrificial layer, 49: polycrystalline silicon layer.

Claims (3)

Board and
An input terminal electrode and an output terminal electrode formed at predetermined intervals on the substrate;
The input terminal electrode and the output terminal electrode on the substrate are formed of the same material as those described above, and both side surfaces thereof face the side surfaces of the input terminal electrode and the output terminal electrode with a minute gap, respectively, and A vibrator held displaceably in a direction parallel to the substrate by a columnar support beam fixed to the substrate,
A resonator filter comprising: The input terminal electrode, the output terminal electrode, and the vibrator are formed by forming a polycrystalline silicon layer deposited on the substrate into a rectangular pattern, and the vibrator is held by a support beam at a plurality of locations. The resonator filter according to claim 1, wherein Patterning a sacrificial layer having a predetermined opening in the substrate;
Depositing a conductor layer on the substrate on which the sacrificial layer is formed,
An input terminal electrode and an output terminal electrode which are patterned at a predetermined interval and are fixed to the substrate at openings of the sacrificial layer by patterning the conductor layer, and both side surfaces between the input terminal electrode and the output terminal electrode, And forming a vibrator that is arranged to face the side surface of the output terminal electrode with a small gap and that is held on the substrate by a columnar support beam at the opening of the sacrificial layer;
Removing the sacrificial layer;
A method for manufacturing a vibrator filter, comprising:

Images