**3. Mitigation of stiction**

#### **3.1 Causes of stiction**

The gap between two metal surfaces or device to substrate is so small that strong capillary forces can be developed during the dehydration which may lead to the adhesion of two surfaces. The same adhesion can occur when device is exposed to high humid conditions which lead to capillary condensation. Microstructures which contaminate the contact surface if stiction occurs, are in fact the synthetic particles of the metals (Alley et al., 1992).

The adhesion may occur due to solid bridging or liquid bridging. In solid bridging, the non volatile impurities present in the drying liquid are deposited on solid surfaces if drying by evaporation is conducted. These impurities may be introduced due to dissolution of the particles or substrate materials by liquid or through dissolution of residues distributed uniformly on the surface of the substrate. The deposition of impurities is pronounced in narrow spaces and between the two metal contacting surfaces upper and lower. This results in adhesion between the metal surfaces. The adhesion strength through solid bridging is

electrodes together by a microscopic surface due to the planner nature of the electrodes. Stiction of MEMS is a common concern. When a sacrificial layer is removed and rinsed in deionized water, the surface tension of rinse water pulls the delicate micro structure to the substrate as the wafer dries. Risk of stiction is caused by the capillary forces originating from the dehydration of meniscuses, van der Waals force or the electrostatic force formed between the suspended beam structures and the substrate following the wet etching (Madou, 2002). These forces keep the structure firmly attached with the substrate. Stiction

Stiction is an inevitable problem we deal with for achieving the working RF MEMS devices. With increase in cantilever length, its flexibility perpendicular to the substrate increases which also increases the susceptibility to stiction. When the structure gets attached with the substrate due to stiction, the mechanical force required to dislodge it from the surface is large enough resulting in damage to MEMS structure (Modou, 2002). The surface morphology has a strong influence on stiction and is a serious problem particularly in metal

In order to achieve a released structure, contact between the structural elements and the substrate should be avoided during processing. Etching can be done by physical damage, chemical damage or combination of both. Release of these suspended beam structures can be done either through wet etching or dry etching. Etching in a plasma environment has several advantages as compared to wet etching. In the wet etching, this may become impossible or very difficult due to large surface tension forces. Moreover, if a MEMS structure is left too long in the etchant, the structure can be over etched and damaged (Harsh et al., 1999). Plasmas are easier to start and stop than simple immersion wet etching (Campbell, 1996). Also sensitivity of plasma etch is less prone to small changes in the temperature of the wafer. Above mentioned factors make plasma etching more repeatable

Different techniques over a period of time have been used to avoid stiction. Method of creating stand-off bumps on the underside of a polysilicon plate was introduced (Abe et al., 1995) which added meniscus shaping microstructures to the perimeter of the microstructure for reducing the chance of stiction. To avoid stiction critical point drying technique using

The gap between two metal surfaces or device to substrate is so small that strong capillary forces can be developed during the dehydration which may lead to the adhesion of two surfaces. The same adhesion can occur when device is exposed to high humid conditions which lead to capillary condensation. Microstructures which contaminate the contact surface if stiction occurs, are in fact the synthetic particles of the metals (Alley et al., 1992). The adhesion may occur due to solid bridging or liquid bridging. In solid bridging, the non volatile impurities present in the drying liquid are deposited on solid surfaces if drying by evaporation is conducted. These impurities may be introduced due to dissolution of the particles or substrate materials by liquid or through dissolution of residues distributed uniformly on the surface of the substrate. The deposition of impurities is pronounced in narrow spaces and between the two metal contacting surfaces upper and lower. This results in adhesion between the metal surfaces. The adhesion strength through solid bridging is

remains a reliability issue due to contact with adjacent surfaces after release.

to metal contact switches (Varadan et al., 2003).

CO2 dryer is used (Chan et al., 2007) to release the structures.

than wet etching.

**3. Mitigation of stiction** 

**3.1 Causes of stiction** 

difficult to estimate because of the variation in deposition process or the density of deposited material. In any case, the adhesion strength tends to be significant.

The liquid bridging occurs due to the surface tension of the trapped capillary liquids. The drying of this trapped liquid is difficult due to the presence of concentrated soluble impurities. These trapped impurities increase surface tension while decreasing the vapor pressure. A third possible adhesion cause can occur if suspended membrane is placed in contact with the lower contact surface due to some external force. This adhesion can occur due to deliberate placement of collapsing forces or can be due to shock effect (Mastrangelo, 2000).

#### **3.2 Stiction due to capillary forces**

The removal of sacrificial layer to achieve a suspended microstructure is the final step in the surface micromachining process. This process mostly requires a wet etching for removal of sacrificial layer. In some cases the removal is also done using plasma etch when sacrificial layer is other than a metal layer like polyimide or photo-resist. After the wet etching the microstructure is rinsed using DI water to remove the residues left during the etching. When the microstructure is pulled out of DI water a strong capillary force develops. A meniscus forms at the interface under the microstructure when the microstructure is pulled out of water. The curved interface creates a pressure called Laplace pressure which is given by (Israelachvili, 1991)

$$P\_L = \mathcal{V}\_l \left(\frac{1}{r\_a} + \frac{1}{r\_b}\right) \tag{1}$$

The liquid surface tension is denoted by γ*l* and two radii of curvature of liquid surfaces are given as *ra* (parallel to surface normal of the substrate) and *rb* (in the plane of the substrate). In most cases, the liquid droplet on the surface of the substrate will not wet it. It will present a definite angle of contact between the liquid and the substrate as shown in figure 1.

Fig. 1. Contact angle at solid liquid interface of (a) non-spreading (b) spreading liquid

In equilibrium condition, the contact angle between liquid and solid is determined by the balance between the surface tension of the three interfaces. The contact angle *θ* at the junction of three interfaces is defined as the angle formed between solid-air, liquid-air and liquid-solid interfacial tensions in equilibrium. The contact is given by the Young's equation (Israelachvili, 1991) as

$$
\gamma\_{SA} = \gamma\_{SL} + \gamma\_{LA} \cos \theta \; \; 0 \le \theta \le \pi \tag{2}
$$

Plasma Based Dry Release of MEMS Devices 273

The stiction between the two metal surfaces due to capillary forces looks quite similar to solid bridging. In the solid bridging, nonvolatile impurities are deposited on the solid surface causing the adhesion during the drying whereas in the capillary forces adhesion of a thin liquid layer works as an adhesion force between the two solid surfaces. If the contact angle *θ* between the solid and liquid is less than 900 (figure 2(a)) then the pressure inside the liquid drop will be less than outside. This results in a net attractive force between the two contacting plates. Figure 4 shows the SEM image of the cantilever beam structure stuck with lower metal surface due to capillary force. The adhesion of the beam was strong. An attempt was made to release the front part of the cantilever using micro probe. This resulted in

Fig. 4. SEM of cantilever beam held with stiction due to capillary forces

capillary force does not come into play at the time of drying.

In this study, fabrication of RF MEMS is done using the CPW structure. When liquid attaches to the long cantilever beam, separation between the beam and CPW is a function of position whereas gap is smaller near the tip than near the anchor. As *ra*→ *∞*, the radius *rb* remains constant as the droplet pressure is constant. As the liquid dries, the decrease in surface area drives the radius *rb* to smaller values. The droplet will form an inside meniscus near the base which forces the droplet to neck down in this region resulting in a negative value of *ra*. The necking continues with lower negative value of *ra* until the two meniscuses on either side of the beam meet and pinch off a separated droplet. This technique for drying of water activates the capillary forces which can lead to adhesion. However, for applications discussed here the drying method has been used while avoiding the meniscus so that

In this drying method, the sacrificial layer was removed by wet etching. During rinsing with DI water when capillary forces could act for stiction during drying, we dipped the device in acetone till the time whole water under the cantilever beam was removed. The acetone dip also removed the diluted impurities which were present in the DI water while removing the

**3.3 Drying method of cantilever beams** 

breaking of the cantilever beam.

If the *SA* surface tension is smaller than the sum of *SL* and *LA* surface tensions, then the contact angle is larger than 00 and liquid will be non spreading as shown in figure 2(a). If the *SA* surface tension is larger, than the sum of *SL* and *LA* surface tensions, it will spread the liquid energetically. Then the contact angle is equal to 00 and liquid will spread thus forming a drop bridging between the two surfaces as shown in figure 2(b). The total surface energy of the area between the metal contacting parts can be calculated by adding the surface tensions of all the three interfaces (Mastrangelo & Hsu, 1993).

Fig. 2. The capillary condensation phenomenon of (a) non-spreading (b) spreading liquid showing underneath of a cantilever beam (front view)

Because the lateral dimesions of microstructures like cantilever beams are much larger than the vertical gap spacing (*g*) due to liquid layer thickness, i.e. rb»ra, therefore we may write (1) as,

$$P\_L = \frac{2\gamma\_l \cos\theta}{\mathcal{g}}\tag{3}$$

where *θ* is the contact angle of the liquid at the surface of the substrate and *g* is the gap height between the cantilever and the substrate which is equal to 2*ra cosθ.*

The shape of the meniscus will be concave (rb< 0) under a cantilever structure on a hydrophilic surface (silicon or any metal) which forms a quite shallow contact angle i.e., *cos* 1 as shown in figure 2(a), so the resulting Laplas pressure is negative. This will create sufficient attractive capillary force that will pull the cantilever beam structure down into contact with lower metal surface or substrate as shown in figure 3. Hence the cantilever beam falls into adhesion or stiction with substrate or metal surface following the drying process.

Fig. 3. The process of the microstructure drying that leads to the adhesion of micro cantilever to adjacent surfaces

272 Microelectromechanical Systems and Devices

contact angle is larger than 00 and liquid will be non spreading as shown in figure 2(a). If the

liquid energetically. Then the contact angle is equal to 00 and liquid will spread thus forming a drop bridging between the two surfaces as shown in figure 2(b). The total surface energy of the area between the metal contacting parts can be calculated by adding the

Fig. 2. The capillary condensation phenomenon of (a) non-spreading (b) spreading liquid

Because the lateral dimesions of microstructures like cantilever beams are much larger than the vertical gap spacing (*g*) due to liquid layer thickness, i.e. rb»ra, therefore we may write

*l*

where *θ* is the contact angle of the liquid at the surface of the substrate and *g* is the gap

The shape of the meniscus will be concave (rb< 0) under a cantilever structure on a hydrophilic surface (silicon or any metal) which forms a quite shallow contact angle

 1 as shown in figure 2(a), so the resulting Laplas pressure is negative. This will create sufficient attractive capillary force that will pull the cantilever beam structure down into contact with lower metal surface or substrate as shown in figure 3. Hence the cantilever beam falls into adhesion or stiction with substrate or metal surface following the drying

2 (3)

*L cos <sup>P</sup> g* 

Fig. 3. The process of the microstructure drying that leads to the adhesion of micro

height between the cantilever and the substrate which is equal to 2*ra cosθ.*

 and   and 

*LA* surface tensions, then the

*LA* surface tensions, it will spread the

*SA* surface tension is smaller than the sum of *SL*

surface tensions of all the three interfaces (Mastrangelo & Hsu, 1993).

(a) (b)

showing underneath of a cantilever beam (front view)

*SA* surface tension is larger, than the sum of *SL*

If the 

(1) as,

i.e., *cos*

process.

cantilever to adjacent surfaces

The stiction between the two metal surfaces due to capillary forces looks quite similar to solid bridging. In the solid bridging, nonvolatile impurities are deposited on the solid surface causing the adhesion during the drying whereas in the capillary forces adhesion of a thin liquid layer works as an adhesion force between the two solid surfaces. If the contact angle *θ* between the solid and liquid is less than 900 (figure 2(a)) then the pressure inside the liquid drop will be less than outside. This results in a net attractive force between the two contacting plates. Figure 4 shows the SEM image of the cantilever beam structure stuck with lower metal surface due to capillary force. The adhesion of the beam was strong. An attempt was made to release the front part of the cantilever using micro probe. This resulted in breaking of the cantilever beam.

Fig. 4. SEM of cantilever beam held with stiction due to capillary forces

#### **3.3 Drying method of cantilever beams**

In this study, fabrication of RF MEMS is done using the CPW structure. When liquid attaches to the long cantilever beam, separation between the beam and CPW is a function of position whereas gap is smaller near the tip than near the anchor. As *ra*→ *∞*, the radius *rb* remains constant as the droplet pressure is constant. As the liquid dries, the decrease in surface area drives the radius *rb* to smaller values. The droplet will form an inside meniscus near the base which forces the droplet to neck down in this region resulting in a negative value of *ra*. The necking continues with lower negative value of *ra* until the two meniscuses on either side of the beam meet and pinch off a separated droplet. This technique for drying of water activates the capillary forces which can lead to adhesion. However, for applications discussed here the drying method has been used while avoiding the meniscus so that capillary force does not come into play at the time of drying.

In this drying method, the sacrificial layer was removed by wet etching. During rinsing with DI water when capillary forces could act for stiction during drying, we dipped the device in acetone till the time whole water under the cantilever beam was removed. The acetone dip also removed the diluted impurities which were present in the DI water while removing the

Plasma Based Dry Release of MEMS Devices 275

The surface roughness of the upper contact area and lower contact area is rough enough, thus generates a rough interface between the two contact areas. The measured surface roughness of the upper contact area is 18nm and lower contact area is 22nm, respectively.

Texturing of the two solid surfaces was enhanced deliberately by introducing the construction of a small supporting post. In this approach, a dimple was introduced under the front tip of the cantilever. The dimple was constructed by making an extra mask layer in the fabrication process before patterning the cantilever beam. The dimensions of the dimple are 10×20μm with the height 0.9μm. Figure 6 shows the SEM of the dimple. It was taken by

This accomplishes the texturing of contact surfaces.

manually turning the cantilever beam using a lift out microscope.

Fig. 6. SEM of the dimple under the front end of the cantilever beam

**4. Dry etch process optimization using RIE** 

fabrication of RF MEMS switches.

The contact adhesion was also investigated by using a sharp Tungsten probe tip. The radius of curvature of the probe tip was 1μm. The sharp tip of the probe was used to pull the beams down under a high magnification microscope to ensure that cantilever tip has made contact with lower surface. When the tip of the cantilever touched the lower CPW surface, the probe tip was removed and cantilever beam started peeling off the surface. During this experimentation, there were only two options available to verify that either the beam would stick to lower surface or the beam would come off without stiction. A number of samples were tested after the release and no inter-solid adhesion was observed in these samples.

The first batch of wafers was used to optimize the release of the final device. For this purpose whole fabrication process was skipped and only the mask layers which were required for the fabrication of cantilever beams were used. For optimizing the release process Aluminum (Al) metal was used as it was readily available rather than expensive Au metal. Once the release process was carefully optimized, Au layer was used for the final

A 2.5μm thick layer of photo-resist (AZ6612) was deposited and then patterned for anchor. A 1.5μm thick layer of evaporated Al was deposited using e-beam evaporator. A layer of 1.0μm of photo-resist was deposited to pattern the cantilever beam. Two approaches were

sacrificial layer. Drying does not takes place with acetone, but device was dipped into a thin photo-resist (AZ5214E) which took a place underneath the cantilever beam as a supporting layer. The acetone evaporates quickly, therefore at no stage sample was exposed to the air. The photo-resist became a concentrated resist as the acetone evaporates leaving only the supporting layer of photo-resist. Now drying of photo-resist with hard-bake will remove the solvents from the photo-resist and then etch the photo-resist supporting layer with O2 plasma using RIE. This drying method resulted in clean surface without residues and no stiction was observed during the process.

#### **3.4 Stiction by contact adhesion**

Another phenomenon which can produce adhesion between the two surfaces is an intersolid adhesion which can overcome the restoring force of the elastic beam. Figure 5 shows the cross section of a cantilever beam with length *L*, width *W*, height *g*, and thickness *t*. The Young's modulus of the beam is represented by *E* which is 78GPa in terms of Au metal used for the fabrication of the switch. The figure shows that beam is adhering to the substrate at a distance *d = (L-x)* from its tip.

Fig. 5. Schematic of a cantilever beam adhering to the substrate

We can calculate the total energy of the system which is sum of the elastic and surface energies and is given (Mastrangelo, 2000) by

$$
\mathcal{U}\mathcal{U}\_T = \mathcal{U}\_E + \mathcal{U}\_S = \frac{6E\mathcal{l}\mathbf{g}^2}{\mathbf{x}^3} - \gamma\_s \mathcal{V}\mathcal{W}d\tag{4}
$$

where *UE* is the bending energy stored in the beam and *US* is the interfacial energy of the contact area. Shear deformations are particularly important for *x → L,* as *d* = (*L-x*) is very small and tip of cantilever beam changes its elastic energy substantially just before detachment. This causes the beam to detach from substrate (Mastrangelo &Hsu, 1993) when *L* = *x* = (*3Et3h2/8γx*)1/4 where γx is the surface energy which is determined from the detachment length and beam dimensions.

#### **3.4.1 Inter solid adhesion reduction method**

To eliminate the chance of permanent adhesion failure between the two solid surfaces, an inter-solid surface adhesion reduction is required. This can be done using techniques such as use of textured surfaces and posts, low energy molecular coatings and fluorinated coatings. The textured surfaces and posts approach has been used for the method presented here. Contact area between the elastic cantilever beam and the lower metal contact area on substrate was reduced which in turn reduced the adhesion forces.

sacrificial layer. Drying does not takes place with acetone, but device was dipped into a thin photo-resist (AZ5214E) which took a place underneath the cantilever beam as a supporting layer. The acetone evaporates quickly, therefore at no stage sample was exposed to the air. The photo-resist became a concentrated resist as the acetone evaporates leaving only the supporting layer of photo-resist. Now drying of photo-resist with hard-bake will remove the solvents from the photo-resist and then etch the photo-resist supporting layer with O2 plasma using RIE. This drying method resulted in clean surface without residues and no

Another phenomenon which can produce adhesion between the two surfaces is an intersolid adhesion which can overcome the restoring force of the elastic beam. Figure 5 shows the cross section of a cantilever beam with length *L*, width *W*, height *g*, and thickness *t*. The Young's modulus of the beam is represented by *E* which is 78GPa in terms of Au metal used for the fabrication of the switch. The figure shows that beam is adhering to the substrate at a

We can calculate the total energy of the system which is sum of the elastic and surface

*T ES s EIg UUU Wd x*

where *UE* is the bending energy stored in the beam and *US* is the interfacial energy of the contact area. Shear deformations are particularly important for *x → L,* as *d* = (*L-x*) is very small and tip of cantilever beam changes its elastic energy substantially just before detachment. This causes the beam to detach from substrate (Mastrangelo &Hsu, 1993) when *L* = *x* = (*3Et3h2/8γx*)1/4 where γx is the surface energy which is determined from the

To eliminate the chance of permanent adhesion failure between the two solid surfaces, an inter-solid surface adhesion reduction is required. This can be done using techniques such as use of textured surfaces and posts, low energy molecular coatings and fluorinated coatings. The textured surfaces and posts approach has been used for the method presented here. Contact area between the elastic cantilever beam and the lower metal contact area on

6 (4)

2 3

stiction was observed during the process.

Fig. 5. Schematic of a cantilever beam adhering to the substrate

energies and is given (Mastrangelo, 2000) by

detachment length and beam dimensions.

**3.4.1 Inter solid adhesion reduction method** 

substrate was reduced which in turn reduced the adhesion forces.

**3.4 Stiction by contact adhesion** 

distance *d = (L-x)* from its tip.

The surface roughness of the upper contact area and lower contact area is rough enough, thus generates a rough interface between the two contact areas. The measured surface roughness of the upper contact area is 18nm and lower contact area is 22nm, respectively. This accomplishes the texturing of contact surfaces.

Texturing of the two solid surfaces was enhanced deliberately by introducing the construction of a small supporting post. In this approach, a dimple was introduced under the front tip of the cantilever. The dimple was constructed by making an extra mask layer in the fabrication process before patterning the cantilever beam. The dimensions of the dimple are 10×20μm with the height 0.9μm. Figure 6 shows the SEM of the dimple. It was taken by manually turning the cantilever beam using a lift out microscope.

Fig. 6. SEM of the dimple under the front end of the cantilever beam

The contact adhesion was also investigated by using a sharp Tungsten probe tip. The radius of curvature of the probe tip was 1μm. The sharp tip of the probe was used to pull the beams down under a high magnification microscope to ensure that cantilever tip has made contact with lower surface. When the tip of the cantilever touched the lower CPW surface, the probe tip was removed and cantilever beam started peeling off the surface. During this experimentation, there were only two options available to verify that either the beam would stick to lower surface or the beam would come off without stiction. A number of samples were tested after the release and no inter-solid adhesion was observed in these samples.
