High isolation micro mechanical switch

This application claims the benefit under 35 USC 119(e) of co-pending provisional application no. 60/157,793 filed on Oct. 5, 1999.

This invention relates to microwave devices, and in particular to high isolation micromechanical microwave switches.

Microwave switches are used in many telecommunications applications, in particular satellites. Such switches should have high isolation, low insertion loss, and low return loss. Typically, the performance of a switch degrades linearly with increasing frequency.

An example of a typical microwave switch is described in U.S. Pat. No. 5,578,976, the contents of which are herein incorporated by reference. In this device, a micro mechanical cantilever arm causes a contact to close a gap in the signal line when the switch is in the on position. The cantilever arm is actuated by electrostatic forces. While the device described in U.S. Pat. No. 5,578,976 offers an improvement in isolation, it is still limited to about −50 dB at 4 GHz.

An object of the invention is to provide an improved microwave switch.

According to the present invention there is provided a micro-mechanical microwave switch comprising a signal line formed o a substrate and determining a gap forming an open circuit in the off-state of the switch; a dielectric displaceable support member carrying a contact to bridge said gap and close said switch in the on-state; and at least one shield electrode in the vicinity of said gap to reduce the coupling across said gap in the off-state of the switch. The support member is typically a cantilevered arm, although other arrangements, such as a displaceable membrane, can be employed.

In accordance with the principles of the invention, a shield electrode can be placed on the support member for the contact. The shunt capacitance created by the shield electrode significantly reduces the coupling across the gap. A simple shield electrode of this nature will reduce the coupling by as much as 6 dB.

In addition, or alternatively, a shield electrode can be placed under the gap, preferably as a buried layer in the supporting substrate. This serves to re-distribute the electromagnetic field in a manner that also reduces the coupling across the gap. In a preferred embodiment, a shield electrode is fixed above the movable switch contact. In the raised position of the contact, the shield electrode abuts the support member. The shield electrode in this case can be formed on a second substrate bonded to the first.

The invention also provides a method of improving the isolation of a micro-mechanical microwave switch, wherein a signal line formed on a substrate defines a gap forming an open circuit in the off-state of the switch and a displaceable support member carries a contact to bridge said gap and close said switch in the on-state, comprising the step of providing a shield electrode in the vicinity of said gap to reduce the coupling across the gap.

The invention will now be described in more detail, by way of example only, with reference to the accompanying drawings, in which:

FIG. 1 is a plan view of the contact portion of the microwave switch shown in U.S. Pat. No. 5,578,976;

FIG. 2 is a cross-section of the switch of FIG. 1 taken along the section line 2 — 2 ;

FIG. 3 is a section taken along the signal line of a prior art switch;

FIG. 4 is a plan view of a switch with a shielded contact;

FIG. 5 is a longitudinal section of the switch taken along the section line 4 — 4 in FIG. 4;

FIG. 6 shows the equivalent circuit for the switch shown in FIGS. 4 and 5;

FIG. 7 is a plot showing isolation against frequency for a device with and without a shield electrode;

FIG. 8 is a plan view of a switch design with reduced signal line width in the vicinity of the gap;

FIG. 9 is a longitudinal section of the switch shown in FIG. 8 taken along section line 9 — 9 .

FIG. 10 is a section through a switch showing electromagnetic field distribution;

FIG. 11 is a section through a part of a switch with a buried shield;

FIG. 12 is a plan view of another switch design;

FIG. 13 is a section taken along the line 13 — 13 of FIG. 12;

FIG. 14 is a section taken along the line 14 — 14 of FIG. 12; and

FIG. 15 is a section through a switch design employing a pair of bonded substrates.

The switch described in U.S. Pat. No. 5,578,976 is shown in FIGS. 1 and 2. This comprises a GaAs substrate 10 supporting a signal line 12 with a gap 14 closed by a contact 16 when the switch is in the on position. The substrate 10 acts as a thick insulator. Ground plane 13 extends on either side of the signal line 13 in a conventional manner.

The high dielectric constant, ∈ r , of GaAs mandates a line width of about 20 μm with 10 μm spacing on either side. In the off-statc, the isolation is limited by coupling along the open contact through gap ν and across the substrate through gap ω. The relative importance of the effect of gap ν;ω has been determined by the inventors to be about 9 to 1.

U.S. Pat. No. 5,578,976 ignores entirely the effect of the substrate coupling and considers only the effect of the gap ν. The patent teaches only that the capacity of coupling can be reduced by increasing the gap ν, which of course requires an increase in the electrostatic actuation voltage. In accordance with the principles of the invention, significant improvements in isolation are achieved without increasing ν.

FIG. 3 shows a simple prior art switch. In the off-state of the switch, coupling occurs through the open contact 16 . The capacitance of this structure can be estimated by assuming a parallel plane model and then doubling it to account for fringing fields. The result of two gaps in series is 4.4 fF (fentofarads) which is larger than the 0.6 fF found for a bare gap.

In FIGS. 4 and 5, the switch made in accordance with the principles of the invention includes a shield electrode 20 placed on dielectric 18 supporting switch contact 16 . The effect of adding shield electrode 20 is to add a shunt capacitance 17 , as shown in the equivalent circuit in FIG. 6, that reduces the coupling across the gap. Such an electrode gives an improvement in isolation of about 6 dB as can be seen in FIG. 7 . As shown in FIG. 6, the capacitance 17 to the shield is typically about 67 fF.

The substrate 10 in this case is quartz, SiO 2 , which has a dielectric constant of about 3.8, compared to 13.5 for GaAs. The lower dielectric constant of silicon requires a line width of about 60 μ with a 5.7 μ spacing on either side. This leads to lower dispersion and permits operation up to frequencies in the order of 40 GHz.

The addition of the shield capacitor creates a return loss when the switch is closed. However because the series capacitance of the switch is small, the shield capacitor can be made small, less than about 10 fP. Also, for some applications, reactive matching can be used to reduce the return loss. For many other applications, the additional return loss will be acceptable.

As in conventional GaAs MMIC switches, resistive material (not shown) should be added to prevent coupling of RF into the control circuitry. Such resistive material does not form part of the invention and will not be described in detail.

The switch shown in FIGS. 7 and 8, has portions 12 a of the signal line 12 of reduced width facing the gap 14 . The contact 16 also has portion 16 a of reduced width facing the reduced width portions of the signal line 12 . The central portion of the top contact is wider under the support beam at the center of the gap.

The reduced width of the contact improves isolation in the off-state. The reduced line width has the secondary effect of increasing the transmission line impedance in the on-state The introduction of capacitance can create an impedance mismatch at the gap causing unwanted reflections. Reducing the line width has the effect of introducing inductance, which cancels the effect of the additional capacitance, thereby allowing the impedance to be matched and unwanted reflections eliminated.

The reduction in the width of the signal line in the vicinity of the gap also has a convenient manufacturing advantage. A narrower signal line requires a narrower spacing to the ground plane, which is difficult to fabricate except over small areas. In this embodiment, the signal line width can be maintained larger, except in the vicinity of the gap where the line width is reduced. The fine line width lithography only needs to be carried out over a very small area.

In the embodiment shown in FIG. 9, an additional conductive layer serving as an underside shield electrode 22 is added on the underside of the substrate 10 . This redistributed the electromagnetic field as shown in FIG. 10 . The broken lines show the fringing field in the absence of the underside shield electrode 22 . The solid arrows show the direction of the field in the presence of the shield 22 . The redistribution of the field to the ground plane services to reduce coupling across the gap.

The dielectric material 18 can be made thinner (e.g. reduced from a typical value of 2 μ to about 0.1μ to increase the shield capacitance, and also a material with higher dielectric constant, such as Al 2 O 3 , can be used.

Spacers 24 can be introduced on the bottom side of the upper contact 16 to increase the space from the bottom shield 22 to the upper moving contact. Alternatively, these spacers can be provided on the signal line 12 .

The fabrication of the underside shield electrode can also be made by forming a buried subsurface CoSi 2 layer 32 locally in the SiO 2 substrate, as shown in FIG. 11 . This can be made by ion implanting cobalt into a suitably masked silicon semi insulating substrate (SiSI) to form a buried subsurface cobalt layer. The substrate is then heated, for example by rapid thermal heat treatment to form CoSi 2 . The top surface of the substrate is then exposed to oxygen to form an overlaying insulating layer of SiO 2 . Vias are then made in this insulating layer to contact the buried CoSi 2 shield.

An alternative way to redistribute the electromagnetic field is shown in FIG. 9 . In this embodiment, a conductive bar 26 is added in the ground plane across the gap 14 In the open condition, the coupling from electric field lines aligned across the gap will be significantly reduced because the field lines will terminate on the bar and capacitive coupling across the gap will be reduced. However, when the switch is closed, with the upper contact down, there will be additional capacitance from the contact to ground, which will introduce a reactive return loss. The size of this unwanted capacitance can, however, be optimized by making the bottom shield 22 narrow and by shaping the upper contact 16 to increase the vertical air gap.

The preferred embodiment is shown in FIGS. 12 to 14 . In this embodiment a conductive layer 28 is provided above the switch to form a capacitive shield, but unlike the embodiment shown in FIGS. 4 and 5, this shield 28 remains fixed in position when the switch is closed.

This embodiment solves the problem of return loss or shunt capacitance in the on-state. The switch contact 16 is still carried by the dielectric membrane 18 , but does not have the shield layer. When the switch open, the contact is raised until it is in close proximity to the static shield 28 , as shown in FIG. 15 . This arrangement decouples the RF signal on the floating contact and reduces the coupling between the signal lines.

This embodiment can be described as a new switch structure, which behaves more like a strip line than a coplanar waveguide. It can be made from two substrates bonded together as shown in FIG. 15 . The dielectric support 18 in the form of a cantilevered and carrying contact 16 is formed on a first substrate 50 along with signal line 12 and bar 26 in gap 14 . The cantilever arm is fabricated using sacrificial material in a conventional manner.

The fixed upper shield 28 is fabricated on a second substrate 60 so as to extend over cavity 62 . The substrate 60 is inverted and bonded to the substrate 50 so that the upper shield 28 lies over the dielectric support 18 .

This design, which is more difficult to make, is expected to have the best performance.

The described switch designs permit the open state coupling to be reduced at frequencies up to 40 GHz.