Micro-electro-mechanical switch, and methods of making and using it

TECHNICAL FIELD OF THE INVENTION

This invention relates in general to switches and, more particularly, to micro-electro-mechanical switches having flexible membranes.

BACKGROUND OF THE INVENTION

One existing type of switch is a radio frequency (RF) micro-electro-mechanical switch (MEMS). This existing type of switch has a substrate with two spaced and conductive posts thereon. A conductive part is provided on the substrate between the posts, and is covered by a layer of a dielectric material. A flexible and electrically conductive membrane extends between the posts, so that a central portion of the membrane is located above the conductive part on the substrate. An RF signal is applied to one of the conductive part and the membrane.

In the deactuated state of the switch, the membrane is spaced above both the conductive part and the dielectric layer covering it. In order to actuate the switch, a direct current (DC) bias voltage is applied between the membrane and the conductive part. This bias voltage produces charges on the membrane and the conductive part, and the charges cause the membrane and conductive part to be electrostatically attracted to each other. This attraction causes the membrane to flex, so that a central portion thereof moves downwardly until it contacts the top of the dielectric layer on the conductive part. This is the actuated position of the membrane.

In this actuated state of the switch, the spacing between the membrane and the conductive part is less than in the deactuated state. Therefore, in the actuated state, the capacitive coupling between the membrane and the conductive part is significantly larger than in the deactuated state. Consequently, in the actuated state, the RF signal traveling through one of the membrane and conductive part is capacitively coupled substantially in its entirety to the other thereof.

In order to deactuate the switch, the DC bias voltage is turned off. The inherent resilience of the membrane then returns the membrane to its original position, which represents the deactuated state of the switch. Because the capacitive coupling between the membrane and conductive part is much lower in the deactuated state, the RF signal traveling through one of the membrane and capacitive part experiences little or no capacitive coupling to the other thereof.

Although existing switches of this type have been generally adequate for their intended purposes, they have not been satisfactory in all respects. One problem is that, when the membrane is contacting the dielectric layer in the actuated state of the switch, electric charge from the membrane can tunnel into and become trapped in the dielectric layer. As a result, and due to long recombination times in the dielectric, the amount of this trapped charge in the dielectric increases progressively over time.

The progressively increasing amount of trapped charge exerts a progressively increasing attractive force on the membrane. When the membrane is in its actuated position, this attractive force tends to resist movement of the membrane away from its actuated position toward its deactuated position. The amount of trapped charge can eventually increase to the point where the attractive force exerted on the membrane by the trapped charge is in excess of the inherent resilient force of the membrane which is urging the membrane to return to its deactuated position. As a result, the membrane becomes trapped in its actuated position, and the switch is no longer capable of carrying out a switching function. This is considered a failure of the switch, and is associated with an undesirably short operational lifetime for the switch. In this regard, an RF MEMS switch of this type should be capable of trillions of switching cycles before a failure occurs due to fatigue in the metal of the membrane, but trapped charge in the dielectric usually results in failure after only millions of switching cycles.

There are many applications in which a switching function can be implemented using either a field effect transistor (FET) switch or an RF MEMS switch. However, due in significant part to the dielectric charging problem discussed above, the operational lifetime of existing MEMS switches is significantly shorter than the operational lifetime of commercially available FET switches. Consequently, FET switches are currently favored over MEMS switches for these applications.

Prior attempts have been made to solve the dielectric charging problem. One approach was to change the properties of the dielectric material so as to modify the extent to which the dielectric material is “leaky”. For example, by adding more silicon to silicon nitride used for the dielectric material, the conductivity of the dielectric material increases, and then it becomes easier for the trapped charges to recombine in a manner which neutralizes them. However, this approach also increases the power consumption of the MEMS switch, and has not been shown to provide a significant increase in its operational lifetime.

Another prior approach to the dielectric charging problem is to alter the waveform used for the DC bias voltage. For example, lowering the actuation voltage reduces the amount of charge which tunnels into the dielectric material, and thus reduces the rate at which the amount of trapped charge within the dielectric material can increase. Further, the slope of the release waveform can be decreased, so as to give the trapped charges more time to recombine. These types of changes to the actuation waveform can produce a significant increase in the operational lifetime of a MEMS switch. However, they also significantly increase the switching time of the switch, for example by a factor of approximately 20, which in turn renders such a MEMS switch highly undesirable for many applications that involve high switching speeds.

In the design of MEMS switches, a traditional design goal has been to try to maximize the capacitance ratio of the switch, which is the ratio of the capacitance between the membrane and conductive part in the actuated state to the corresponding capacitance in the deactuated state. In an effort to maximize the capacitance in the actuated state, pre-existing MEMS switch designs attempt to position the membrane as close as possible to the conductive part in the actuated state of the switch, which in turn means that the dielectric layer separating them needs to be relatively thin. Consequently, the surfaces of the membrane and dielectric layer which engage each other have traditionally been intentionally polished or otherwise fabricated to make them as smooth as possible, so that both surfaces have their entire areas in direct physical contact with each other when the membrane is in its actuated position, thereby positioning as much of the membrane as possible in very close proximity to the conductive part.

SUMMARY OF THE INVENTION

From the foregoing, it may be appreciated that a need has arisen for a method and apparatus for making and operating a switch of the type having a flexible membrane, in a manner so that the switch has a significantly increased operational lifetime. According to the present invention, a method and apparatus are provided to address this need.

More specifically, according to one form of the invention, a switch includes a base having a first section which includes an electrically conductive part, and also includes a membrane having first and second ends supported at spaced locations on the base, and having between the ends a second section which includes an electrically conductive portion. The membrane is capable of resiliently flexing so as to move between first and second positions, the conductive part and the conductive portion being physically closer in the second position than in the first position. One of the first and second sections has a textured surface, and the other thereof has a further surface which faces the textured surface, the textured surface having mutually exclusive first and second portions which are respectively in physical contact with and free of physical contact with the further surface when the membrane is in the second position. The first portion of the textured surface has an area which is substantially less than a total area of the textured surface.

According to a different form of the invention, a switching method uses a switch that includes a base having a first section with an electrically conductive part, and that includes a membrane having first and second ends supported at spaced locations on the base, and having between the ends a second section which includes an electrically conductive portion, where one of the first and second sections has a textured surface and the other thereof has a further surface which faces the textured surface. The method includes: responding to an applied voltage between the conductive part and the conductive portion by resiliently flexing the membrane so that the conductive part moves closer to the conductive portion as the membrane moves from a first position to a second position. This includes causing mutually exclusive first and second portions of the textured surface to respectively be in physical contact with and free of physical contact with the further surface when the membrane is in the second position, the first portion of the textured surface having an area which is substantially less than a total area of the textured surface.

According to still another form of the invention, a method of fabricating a switch includes: forming on a base a first section which includes an electrically conductive part; forming a resiliently flexible membrane having first and second ends engaging spaced portions of the base disposed on opposite sides of the first section, and having between the ends a second section which includes an electrically conductive portion, the membrane being capable of resiliently flexing so as to move between first and second positions so that the conductive part and the conductive portion are physically closer in the second position than in the first position; and forming on one of the first and second sections a textured surface and on the other thereof a further surfaces which faces the textured surface, the textured surface having mutually exclusive first and second portions which are respectively in physical contact with and free of physical contact with the further surface when the membrane is in the second position, the first portion of the textured surface having an area which is substantially less than a total area of the textured surface.

BRIEF DESCRIPTION OF THE DRAWINGS

A better understanding of the present invention will be realized from the detailed description which follows, taken in conjunction with the accompanying drawings, in which:

FIG. 1

is a diagrammatic fragmentary sectional side view of an apparatus which includes a micro-electro-mechanical switch (MEMS) that embodies aspects of the present invention;

FIG. 2

is a diagrammatic fragmentary sectional side view similar to

FIG. 1

, but showing the switch of

FIG. 1

in a different operational state;

FIG. 3

is a diagrammatic fragmentary sectional side view showing a portion of the switch of

FIG. 1

, at an intermediate point during its fabrication;

FIG. 4

is a diagrammatic fragmentary sectional side view similar to

FIG. 3

, but showing part of the switch of

FIG. 1

at a later point during fabrication of the switch;

FIG. 5

is a diagrammatic fragmentary sectional side view similar to

FIG. 1

, but showing a micro-electro-mechanical switch (MEMS) which is an alternative embodiment of the switch of

FIG. 1

;

FIG. 6

is a diagrammatic fragmentary sectional side view similar to

FIG. 5

, but showing the switch of

FIG. 5

in a different operational state;

FIG. 7

is a diagrammatic fragmentary sectional side view showing a portion of the switch of

FIG. 5

, at an intermediate point during its fabrication;

FIG. 8

is a diagrammatic fragmentary sectional side view similar to

FIG. 7

, but showing part of the switch of

FIG. 5

at a later point during fabrication of the switch; and

FIG. 9

is a diagrammatic fragmentary sectional side view similar to

FIG. 8

, but showing the switch of

FIG. 5

at still a later point during its fabrication.

DETAILED DESCRIPTION OF THE INVENTION

FIG. 1

is a diagrammatic fragmentary sectional side view of an apparatus which includes a micro-electro-mechanical switch (MEMS)

10

, the switch

10

embodying aspects of the present invention. The drawings, including

FIG. 1

, are diagrammatic and not to scale, in order to present the switch

10

in a manner which facilitates a clear understanding of the present invention.

With reference to

FIG. 1

, the switch

10

includes a silicon semiconductor substrate

13

having on an upper side thereof an oxide layer

14

. Although the substrate

13

is a made of silicon in this disclosed embodiment, it could alternatively be made of some other suitable material, such as gallium arsenide (GaAs), or a suitable alumina. Similarly, the oxide layer

14

is silicon dioxide in this disclosed embodiment, but could alternatively be some other suitable material.

Two posts

17

and

18

are provided at spaced locations on the oxide layer

14

, and are each made of a conductive material. In this embodiment the posts are made of gold, but they could alternatively be made of some other suitable conductive material. A plurality of diagrammatically depicted nodules

21

are provided on the upper surface of the oxide layer

14

, at a location intermediate the posts

17

and

18

, in order to create a degree of roughness or texture on this part of the top surface of the oxide layer

14

. In the embodiment of

FIG. 1

, the nodules

21

are made of silicon titanium (SiTi), and have a silicon to titanium ratio of about 5:1. However, the nodules

21

could alternatively be made from some other suitable material. In the embodiment of

FIG. 1

, the nodules

21

have a vertical height of approximately 100 to 500 nm, but could alternatively have some other suitable height.

Instead of using the silicon substrate with nodules thereon, it would alternatively be possible to omit the nodules and to use a substrate of some other material, such as alumina, which has a top surface inherently rougher than the top surface of silicon. The electrode and dielectric layer would then conform to the rough surface on top of the alumina in order to create a textured surface on top of the dielectric layer. Although alumina has previously been used in pre-existing MEMS switches, it has customarily been highly polished so as to eliminate any significant roughness, in an effort to maximize the capacitance ratio, as discussed earlier.

An electrically conductive electrode

22

serves as a transmission line, and is elongated in a direction perpendicular to the plane of FIG.

1

. In the embodiment of

FIG. 1

, the electrode

22

is made of gold, but it could alternatively be made from some other suitable material. A central portion of the electrode

22

, which is visible in

FIG. 1

, is approximately 300 to 400 nm thick, and extends over the nodules

21

and the adjacent portions of the top surface of the oxide layer

14

. Consequently, in view of the vertical height of the nodules

21

, the electrode

22

conforms generally to the shape of an upwardly facing surface defined by surface portions on top of the nodules

21

and oxide layer

14

. Thus, the top surface of the central portion of the electrode

22

has a degree of roughness or texture.

This central portion of the electrode

22

is covered by a dielectric layer

23

. In the disclosed embodiment, the dielectric layer

23

is made of silicon nitride, and has a thickness of approximately 100 to 300 nm. The dielectric layer

23

conforms in shape to the top surface of the electrode

22

, and thus the top surface of the electrode

23

has a degree of roughness or texture. The substrate

13

, oxide layer

14

, posts

17

-

18

, nodules

21

, electrode

22

and dielectric layer

23

can be collectively referred to as a base portion of the switch

10

.

A conductive membrane

31

extends between the upper ends of the posts

17

and

18

. In the disclosed embodiment, the membrane

31

is made of a known aluminum alloy, and in fact could be made of any suitable material that is commonly used to fabricate membranes in MEMS switches. The membrane

31

has ends

32

and

33

, which are each fixedly supported on the top portion of a respective one of the posts

17

and

18

. The membrane

31

has, between its ends

32

and

33

, a central portion

36

which is disposed directly above the electrode

22

and the dielectric layer

23

.

The membrane

31

is approximately planar in the view of

FIG. 1

, but is capable of flexing so that its central portion

36

moves downwardly until it contacts the textured top surface of the dielectric layer

23

. This flexed position is shown in

FIG. 2

, which is a diagrammatic fragmentary sectional side view showing the same structure as

FIG. 1

, but in a different operational state.

During operational use of the switch

10

, a radio frequency (RF) signal having a frequency in the range of approximately 300 MHz to 90 GHz is caused to travel through one of the membrane

31

and the electrode

22

. More specifically, the RF signal may be traveling from the post

17

through the membrane

31

to the post

18

. Alternatively, the RF signal may be traveling through the electrode

22

in a direction perpendicular to the plane of

FIGS. 1 and 2

.

Actuation of the switch

10

is carried out under control of a direct current (DC) bias voltage, which is applied between the membrane

31

and the electrode

22

by a control circuit of a type which is well-known in the art, and which is therefore not illustrated and described. This bias voltage can also be referred to as a pull-in voltage (V

p

). When the bias voltage is not applied to the switch

10

, the membrane

31

is in the position shown in FIG.

1

. As discussed above, an RF signal will be passing through one of the membrane

31

and the electrode

22

. For convenience in the discussion which follows, it will be assumed that the RF signal is passing through the electrode

22

. When the membrane

31

is in the deactuated position of

FIG. 1

, the RF signal traveling through the electrode

22

will pass through the switch

10

and continue traveling through the electrode

22

, with no significant coupling of this RF signal from the electrode

22

over to the membrane

31

.

In order to actuate the switch

10

, a DC bias voltage (pull-in voltage V

p

) is applied between the electrode

22

and the membrane

31

. This bias voltage produces charges on the membrane

31

and on the electrode

22

, which in turn produce an electrostatic attractive force that urges the central portion

36

of the membrane

31

toward the electrode

22

. This attractive force causes the membrane

31

to flex downwardly, so that its central portion

36

moves toward the electrode

22

. The membrane

31

flexes until its central portion

36

engages the textured top surface of the dielectric layer

23

, as shown in FIG.

2

. This is the actuated position of the membrane. In this position, the capactive coupling between the electrode

22

and the central portion

36

of the membrane

31

is approximately 100 times greater than when the membrane

31

is in the deactuated position shown in FIG.

1

. Consequently, the RF signal traveling through the electrode

22

will be coupled substantially in its entirety from the electrode

22

over into the membrane

31

, where it will tend to have two components that travel away from the central portion

36

of the membrane in opposite directions, toward each of the posts

17

and

18

. Alternatively, if the RF signal had been traveling through the membrane

31

from the post

17

to the post

18

, the RF signal would have been coupled substantially in its entirety from the central portion

36

of the membrane over to the electrode

22

, where it would tend to have two components that travel away from the switch

10

in respective opposite directions through the electrode

22

.

Once the membrane

31

has reached the actuated position shown in

FIG. 2

, the not-illustrated control circuit may optionally reduce the DC bias voltage (pull-in voltage V

p

) to a standby or hold value. The standby or hold value is less than the voltage that was needed to initiate downward movement of the membrane

31

from the position shown in

FIG. 1

, but is sufficient to maintain the membrane

31

in the actuated position of

FIG. 2

, once the membrane has reached this actuated position.

While the membrane

31

is in the actuated position of

FIG. 2

, the textured top surface of the dielectric layer

23

causes the actual physical contact between the dielectric layer

23

and electrode

31

to be limited to a number of spaced contact regions that are each relatively small in area. In other words, the total area of physical contact between the dielectric layer

23

and the membrane

31

is substantially less than would be the case if the dielectric

23

had a smooth and flat top surface which, in its entirety, was in physical contact with the smooth and approximately flat underside of the central portion

36

of the membrane

31

. Since the operative coupling between the membrane

31

and electrode

22

involves capactive coupling, rather than direct physical contact, reducing the total amount of direct physical contact between them does not have a significant effect on the operation of the switch

10

.

When a textured surface of the type shown in

FIG. 1

is used, the capacitance between the membrane

31

and the electrode

22

in the actuated state of

FIG. 2

may be slightly less that it would be if the dielectric layer

23

had a traditional flat top surface. Consequently, the ratio of the capacitance for the actuated state of

FIG. 2

to the capacitance for the deactuated state of

FIG. 1

may be slightly less than when the dielectric layer

23

has a traditional flat surface. However, this small reduction in the capacitance ratio is negligible at higher frequencies, and any minor disadvantage is outweighed by the fact that a significant advantage is obtained from use of the textured surface. In particular, by using the textured surface to reduce the total area of actual physical contact between the membrane

31

and the dielectric layer

23

, there is less total area of physical contact through which electric charge from the membrane

31

can pass, and this in turn reduces the amount of charge that can tunnel into and become trapped in the dielectric layer

23

. This means that the rate at which trapped charge can build up in the dielectric layer

23

is substantially lower for the switch of

FIGS. 1-2

than for pre-existing switches.

As a result, it takes much longer for the switch

10

to reach a state where the amount of trapped charge in the dielectric layer can attract the membrane

31

with a force sufficiently large to prevent the switch

10

from deactuating when the DC bias voltage (pull-in voltage V

p

) is terminated. Therefore, the effective operational lifetime of the switch

10

is substantially longer than for pre-existing switches which do not have the textured surface. In fact, the textured surface extends the operational lifetime of the switch so much that the limiting factor on operational lifetime becomes physical fatigue and/or failure of the membrane

31

, rather than trapping of the membrane

31

due to trapped charges in the dielectric layer

23

. In this regard, switch

10

will have an operational lifetime that can be 1,000 to 1,000,000,000 times longer that the operational lifetime of comparable pre-existing switches that lack the textured surface.

A secondary advantage of the textured surface is that, by reducing the total area of physical contact between the membrane

31

and the dielectric layer

23

, there is a reduction in Van Der Walls forces which tend to cause attraction between the membrane

31

and dielectric layer

23

, and which thus resist movement of the membrane

31

away from the dielectric layer

23

.

In order to deactivate the switch

10

, the not-illustrated control circuit terminates the DC bias voltage (pull-in voltage V

p

) that is being applied between the membrane

21

and the electrode

22

. The inherent resilience of the flexible membrane

31

produces a relatively strong restoring force, which causes the central portion

36

of the membrane to move upwardly away from the dielectric layer

23

and the electrode

22

, until the membrane reaches the position shown in FIG.

1

.

FIG. 3

is a diagrammatic fragmentary sectional side view of part of the switch

10

of

FIG. 1

, showing the switch at an intermediate point during its fabrication. Fabrication of the switch

10

begins with provision of the silicon substrate

13

, and then the silicon oxide layer

14

is grown on the substrate

13

using known techniques.

Next, a layer

71

of photoresist is applied over the oxide layer

14

. The photoresist layer

71

is then patterned and etched using known techniques, so as to define through the layer

71

an opening

72

in the region where the electrode

22

(

FIG. 1

) will eventually be formed. Next, a layer

74

of an aluminum alloy is sputtered over the layer

71

, so that a portion of the layer

74

engages the oxide layer

14

within the opening

72

through the photoresist

71

. In the disclosed embodiment, the aluminum alloy layer

71

is approximately 300 nm thick, and contains approximately 98.8% aluminum (Al), 1% silicon (Si), and 0.2% titanium (Ti). The layer

74

is then wet etched, in order to remove the aluminum through aluminum leach. The aluminum leach that occurs during the wet etch does not remove the silicon and titanium, thereby leaving the SiTi nodules

21

(FIG.

1

), which are approximately 100 to 500 nm in vertical height. Next, the photoresist layer

71

is removed in a known manner. Any SiTi nodules present on the layer

71

itself are removed with the layer

71

, thereby leaving only the SiTi nodules

21

located directly on the oxide layer

14

, as shown diagrammatically in FIG.

1

.

FIG. 4

is a diagrammatic fragmentary sectional side view similar to

FIG. 3

, but showing part of the switch at a later point during its fabrication. With reference to

FIG. 4

, the next step in the fabrication of the switch is to form the electrode

22

over the nodules

21

and the oxide layer

14

, for example by depositing a layer of gold and then carrying out a patterned etch. After that, the dielectric layer

23

is formed, for example by depositing a layer of silicon nitride and then carrying out a patterned etch.

Next, the posts

17

and

18

are formed, by depositing a layer of gold, and then carrying out a patterned etch so to leave just the posts

17

-

18

. Then, a spacer layer

81

is formed over the oxide layer

14

, dielectric layer

23

and posts

17

-

18

. The spacer layer

76

is a photoresist material of a type known to persons skilled in the art. The spacer layer

76

is patterned, etched, and/or planarized, in order to give it a desired shape and thickness. After that, a layer of a known aluminum alloy is deposited over the spacer layer

76

, the posts

17

-

18

and the oxide layer

14

, and is patterned and etched in order form the membrane

31

. At this point, the structure has the configuration which is shown in FIG.

4

.

Next, an etch procedure referred to as a membrane release etch is carried out, in order to remove the spacer layer

76

in its entirety. The membrane release etch may, for example, be a plasma etch of a known type, or any other suitable etch that will attack the material of the photoresist forming the spacer layer

76

. This etch leaves the membrane

31

suspended on the posts

17

-

18

by its ends

32

and

33

. This is the finished configuration of the switch

10

, which is shown in FIG.

1

.

FIG. 5

is a diagrammatic fragmentary sectional side view of an apparatus that includes a micro-electro-mechanical switch (MEMS)

110

, which is an alternative embodiment of the switch

10

of FIG.

1

. Except for differences which are described below, the switch

110

is generally similar the switch

10

, and identical parts are identified by the same reference numerals.

The switch

110

includes a substrate

13

, oxide layer

14

, and posts

17

and

18

, which are equivalent to their counterparts in the embodiment of FIG.

1

. An electrode

122

is provided on the oxide layer

14

intermediate the posts

17

-

18

, and is covered by a dielectric layer

123

. It will be noted that the SiTi nodules

21

in

FIG. 1

, are omitted from the switch

110

of FIG.

5

. Consequently, the electrode

122

is disposed directly on a flat top surface of the oxide layer

14

, and the electrode

122

thus has an approximately flat upper surface. A portion of the top surface of the dielectric layer

123

which is located directly above the electrode

122

is also flat, rather than textured. Aside from this, the electrode

122

and dielectric layer

123

are generally equivalent to the electrode

22

and dielectric layer

23

in the switch

10

of FIG.

1

.

In

FIG. 5

, an electrically conductive membrane

131

extends between the upper ends of the posts

17

and

18

, and has ends

132

and

133

which are each fixedly supported on top of a respective one of the posts

17

and

18

. The membrane

131

is generally equivalent to the membrane

31

in the switch

10

of

FIG. 1

, except that the membrane

131

has a textured surface

138

on the underside of a central portion

136

thereof. The textured surface

138

includes several projections or bosses that project downwardly toward the electrode

122

and the dielectric layer

123

, and which are spaced from each other.

The membrane

131

can resiliently flex from the deactuated position shown in

FIG. 5

to an actuated position. In this regard,

FIG. 6

is a diagrammatic fragmentary sectional side view showing the switch

110

of

FIG. 5

, but with the membrane

131

in its actuated position. In this actuated position, the dielectric layer

123

engages only spaced portions of the textured surface

138

which are located at the ends of the bosses. Consequently, the total area of actual physical contact between the membrane

131

and the dielectric layer

123

is less than would be the case if the flat surface of the dielectric layer was engaging a flat surface on the membrane.

The switch

110

of

FIGS. 5-6

operates in a manner similar to the operation of the switch

10

of

FIGS. 1-2

. Accordingly, it is believed to be unnecessary to provide a separate detailed explanation of the operation of the switch

110

.

FIG. 7

is a diagrammatic fragmentary sectional side view of part of the switch

110

of

FIG. 5

, at an intermediate stage during fabrication of the switch

110

. With reference to

FIG. 7

, fabrication of the switch

110

begins with provision of the silicon substrate

13

, and then the oxide layer

14

is grown on the silicon substrate

13

. After that, the electrode

142

is formed on the oxide layer

14

, for example by depositing a layer of gold and then carrying out a patterned etch. Next, the dielectric layer

123

is formed, for example by depositing a layer of silicon nitride, and then carrying out a patterned etch.

Next, the posts

17

and

18

are formed, for example by depositing a layer of gold and then carrying out a patterned etch that removes unwanted material, so as to leave just the posts

17

and

18

. Then, a spacer layer

171

is formed over the oxide layer

14

, the dielectric layer

123

, and the posts

17

-

18

. The spacer layer

176

is a photoresist material of a known type, which is patterned and etched in order to give it a desired shape. The resulting structure may be planarized, so that the top surfaces of the posts

17

and

18

are substantially flush with the top surface of the spacer layer

171

.

A mask

176

is then placed over the partially completed device. In

FIG. 7

, the mask

176

is shown resting on the top surfaces of the spacer layer

171

and the posts

17

-

18

, but the mask

176

may alternatively be spaced slightly above these surfaces. The mask

176

includes a glass layer

177

which is transparent to ultraviolet radiation, and a chrome layer

178

which is provided on the underside of the glass layer

177

. The chrome layer

178

is non-transmissive to ultraviolet radiation. The chrome layer

178

has, in a central portion thereof immediately above electrode

122

and the dielectric layer

123

, a cluster of spaced openings, one of which is identified by reference numeral

183

. In the disclosed embodiment, the openings

183

are circular and each have a diameter in the range of approximately 100 nm to 500 nm, but they could alternatively have some other suitable shape or size. Using alignment techniques known to those skilled in the art, the mask

176

is accurately positioned with respect to the structure being fabricated, so that the cluster of openings

183

is accurately centered above the electrode

122

and the dielectric layer

123

.

Next, the structure shown in

FIG. 7

is exposed to ultraviolet radiation for a predetermined time interval, as indicted diagrammatically by arrows

184

. Radiation which impinges on the chrome layer

178

will be either reflected or absorbed. The remaining radiation will pass through the openings

183

, and will “expose” spaced regions of the photoresist material located adjacent the top surface of the spacer layer

171

.

The mask

176

is then removed, and the spacer layer

171

is etched using known techniques, so as to remove material of the spacer layer

171

which has been exposed to light. The result is spaced recesses or dimples in the top surface of the spacer layer

171

. In this regard,

FIG. 8

is a diagrammatic fragmentary sectional side view similar to

FIG. 7

, but showing part of the switch

110

at a later stage in its fabrication. In

FIG. 8

, reference numeral

186

designates one of the recesses that are created in the top surface of the spacer layer

171

by the etch procedure. The strength and duration of the etch procedure are selected to give the recesses

186

a desired depth. In the disclosed embodiment, the recesses

186

have a depth of approximately 100 nm, but it would be possible for the recesses to alternatively have a larger or smaller depth.

Next, the membrane

131

is formed by depositing a layer of a known aluminum alloy over the spacer layer

171

, the posts

17

-

18

, and the oxide layer

14

. This layer of aluminum alloy is then patterned and etched, in order to form the membrane

131

. The central portion

136

of the membrane

131

conforms in shape to the top surface of the spacer layer

171

, including the recesses

186

therein. Thus, the portion of the top surface having the recesses

186

creates the textured surface

138

on the underside of the central portion

136

of the membrane

131

.

Thereafter, a known etch procedure referred to as a membrane release etch is carried out, in order to remove the spacer layer

171

in its entirety. This etch leaves the membrane

131

suspended on the posts

17

-

18

by its ends

132

and

133

. This is the finished configuration of the switch

110

, which is shown in FIG.

5

.

The present invention provides a number of technical advantages. One such technical advantage is that a MEMS switch embodying the present invention has a useful lifetime which is several orders of magnitude better than pre-existing MEMS switches. In this regard, the provision of a textured surface on at least one of the membrane and dielectric layer reduces the total area of physical contact between the membrane and dielectric layer. This in turn reduces the amount of charge from the membrane which can tunnel into and become trapped in the dielectric layer, thereby decreasing the voltage which charge trapped in the dielectric can exert on the membrane in a manner that could eventually latch the membrane in its actuated position.

Due to the textured surface, the operational lifetime of the switch begins to approach the operational lifetime of certain field effect transistor (FET) switches, thereby permitting a MEMS switch which embodies the invention to compete commercially for use in applications that traditionally were restricted to FET switches. A further advantage is that the textured surface tends to reduce the extent to which Van Der Walls forces and/or contamination can resist movement of the membrane away from the dielectric layer when the switch is deactuated.

Although selected embodiments have been illustrated and described in detail, it will be understood that various substitutions and alterations are possible without departing from the spirit and scope of the present invention, as defined by the following claims.