9299.net
大学生考试网 让学习变简单
赞助商链接
当前位置:首页 >> 工程科技 >>

Design of a bidirectional MEMS actuator with high displacement resolution latching

Design of a bidirectional MEMS actuator with high displacement resolution latching


Available online at www.sciencedirect.com

Microelectronic Engineering 85 (2008) 587–598 www.elsevier.com/locate/mee

Design of a bidirectional MEMS actuator with high displacement resolution, large driving force and power-free latching
Weisong Wang a,1, Svetlana Tatic-Lucic b,*, Walter L. Brown a, Richard Vinci a
a b

Materials Science and Engineering Department, Lehigh University, Bethlehem, PA, USA Electrical and Computer Engineering Department, Lehigh University, Bethlehem, PA, USA Received 21 June 2007; accepted 30 October 2007 Available online 12 November 2007

Abstract This paper describes the design and ANSYS modeling of a bidirectional thermal inchworm MEMS actuator featuring high actuation resolution (0.1 lm) combined with large driving force (100 mN) and power-free latching. A promising application of this device is for precision in-package positioning of optical ?bers, but the actuator also has potential for wider use. The inchworm mechanism includes two E-shaped metallic actuators facing each other and a pusher between them that couples to a moveable object such as an optical ?ber. Nickel is chosen as the material of the actuators due to its desirable electrical, thermal, and mechanical properties, and availability in fabrication facilities. Electrothermal actuation is used to move the pusher as well as the optical ?ber in the desired direction by inputting electric currents in a particular sequence. A transient thermal–mechanical analysis shows the feasibility of the input sequence that drives the inchworm actuation. ? 2007 Elsevier B.V. All rights reserved.
Keywords: Inchworm; Electrothermal; Actuator; MEMS; Optical ?ber alignment

1. Introduction Inexpensive, high-resolution optical ?ber positioning is a critical task for optical communications packaging. Generally, an optical ?ber is inserted into one end of a device package and needs to be positioned so that the optical signal is optimized at a receiver at the other end. Traditionally, optical ?ber alignment is carried out with tools external to the ?ber package requiring a large amount of ?oor space and capital investment. In-package MEMS actuators can help reduce the cost and increase the yield. Actuation mechanisms with simple structures, such as electrostatic [1], electrothermal [2], or shape memory devices [3], have been demonstrated for this application. However, in these approaches, the typical adjustable range is limited
Corresponding author. Tel.: +1 610 758 4522; fax: +1 610 758 4561. E-mail address: svt2@lehigh.edu (S. Tatic-Lucic). 1 Present address: Department of Astronomy, University of Texas at Austin, Austin, TX, USA. 0167-9317/$ - see front matter ? 2007 Elsevier B.V. All rights reserved. doi:10.1016/j.mee.2007.10.006
*

by the dimensions of the actuator and latching requires a separate component. In recent years, the inchworm actuator approach has received a great deal of attention as an alternative because it will move a workpiece over a large displacement [4,5], but with high resolution and will be able to provide power-free clamping of the workpiece. Electrostatic comb drives have been used in inchworm motors for the realization of a high-precision and longstroke translation system [6]. Unfortunately, the driving force of electrostatic actuators is generally small. Inchworm motion using piezoelectric actuators has also been studied [7], but these devices are somewhat di?cult to fabricate and to integrate with other micro components. Finally, existing electrothermal inchworm actuators provide large driving force and su?cient driving distance for some applications [8,9]. However, they do not provide power-free latching and their actuation is highly non-linear with decreasing force with increasing driving distance. We were able to better satisfy the requirements above by implementing a new approach.

588

W. Wang et al. / Microelectronic Engineering 85 (2008) 587–598

In this work, we describe a new kind of inchworm actuator with a simple structure that uses electrothermal actuation for high-resolution movement, and has linear motion, constant force, and inherent power-free latching (Fig. 1) [10]. Our particular application, in-package positioning of optical ?bers [11–13], requires the actuation mechanism to be capable of moving a ?ber over a distance of 100 lm, with a resolution of 0.1 lm, at speeds on the order of 1 lm/s. The large distance allows for coarse initial alignment of the ?ber in the package, thereby lowering cost. The high resolution is needed because of the small diameter of the optical ?ber core (as small as 8 lm [14]). Alignment speed is also a cost issue. Power-free latching allows the device to hold the alignment position in the ‘‘o?” state which helps limit power consumption. In order to realize this, a large driving force is desired. Bidirectional movements can be achieved by altering the actuation sequence. The structures of our device have been designed and evaluated by ?nite element analysis (FEA) to meet these goals. Fig. 1 illustrates a simpli?ed top view of the inchworm optical ?ber aligner. The ?ber is embedded in the pusher so that it will move as the pusher moves. The inchworm works by alternately extending the main beams of the two actuators on opposite sides of the pusher then releasing and reclamping the clamp arms in an appropriate sequence. The actual sequence will be discussed in detail below, along with the design rationale and modeling results. 2. Device design and operation principle Fig. 2 is a schematic drawing of the complete chip for Xaxis actuation. The inchworm motor consists of three critical components: two identical ‘‘E”-shaped actuators placed symmetrically – one on each side of the X-axis – and a pusher with the embedded optical ?ber. It is an advanced design compared to that in our preliminary report [10]. The primary improvements have been: reducing the number of electrical connections in each actuator from three to two so that none have to be made on the unsupported metal; increasing the ?exibility of the hinge joint

Pusher Optical fiber Detector
X Y

Clamp arm Actuator - left Release arm Actuator - right Main beam Inchworm motor
Fig. 1. Schematic of inchworm optical ?ber aligner.

to allow greater lift of the clamp arm and greatly reduce stress in the metal at the hinge; increasing the thickness of the actuator to 30 lm from 7 lm to increase sti?ness; incorporating a truss design for the release arm to allow actuation at lower current for the thicker metal. The actuators are made of nickel due to its relatively high electrical resistivity, low thermal conductivity, high thermal expansion coe?cient, and high elastic modulus. It is also available in most MEMS fabrication facilities. Most of the metal structure is released from the base during fabrication. Only two thermal anchors provide connection of each actuator to the base and only two electrical contacts are required for each. These contacts are over the thermal anchors but are electrically isolated from the base with oxide. The release arm is optimized as a truss structures in order to increase its electrical resistance and thus provide large displacement with an easily accessible current. The pusher can be made of nearly any mechanically stable electrical and thermal insulator, such as a ceramic, since it is inserted into the device after the remainder of the fabrication is complete. The clamping of the pusher by the arms is provided by designing the pusher to be slightly wider than the default space between the two clamp arms. Once the pusher is inserted, the actuators’ arms will have a built-in stress at room temperature. The di?erence between the stress-free arm spacing and the pusher width must be small enough to be overcome by actuation of the release arms during device operation. The small stationary slider blocks are important in aligning the pusher during actuation. When the pusher is centered between the sliders, the gap on each side is designed to be 0.25 lm in our device. Fig. 3 illustrates the other major dimensions of the motor. In order to obtain inchworm movement, two types of actuation are needed: (1) extension and retraction of the length of the main beams and (2) clamping and release of each clamp arm on the pusher. The two basic classes of movement are both produced by electrically generated heat. In our design, the selection of one actuation type over the other depends entirely on actuation time and current. The clamped pusher moves forward a small distance in X when an electrical current induces an increase in the length of one of the main beams. Then, using a higher current ?owing for a very short time, the release arm on that actuator transiently extends in Y and swings the clamp arm up in Y to momentarily release and then reclamp it on the pusher. This step needs to be quick in order to avoid extra X-extension motion produced in the process. Retraction in X is achieved by turning o? the current in the actuator. By controlling the time and the magnitude of the electrical current inputs to the two actuators, inchworm-type motions can be achieved in either the plus- or minus-X direction. As discussed in the following sections, the materials, design details, and operation sequence have been optimized based on computer modeling of the device’s thermal and mechanical behavior.

W. Wang et al. / Microelectronic Engineering 85 (2008) 587–598

589

Fig. 2. Schematic drawing of the complete chip.

3. Actuation analysis 3.1. Material selection and actuator mechanics Each material is characterized by thermal properties such as thermal expansion coe?cient (a) and thermal conductivity (k). Linear expansion can be expressed as

DL ? aDT ; ?1? L where DL is the length change, L is the unheated material length, and DT is the average temperature change. If heat is supplied uniformly along the length by Joule heating of the actuator itself and all the heat ?ows out of one end,

590

W. Wang et al. / Microelectronic Engineering 85 (2008) 587–598

2000 ?m

1000 ?m

Main beam 1

50 ?m 350 ?m 5 ?m 10 ?m 900 ?m 1300 ?m Release arm 1 Clamp arm 1 Contact 1a Contact 1b

500 ?m Contact 2a Contact 2b Clamp arm 2 Release arm 2

Main beam 2

Fig. 3. Major dimensions of the actuator.

the steady-state heat balance of heat in and heat out is expressed according to the equation A I 2R ; k DT ? L 3 ?2?

where A is the cross-section of the material, I is the input electrical current, and R is the resistance of the current carrying structure. Consequently, the steady-state relationship between the length change and the input current can be derived from Eqs. (1) and (2):   DL I 2 q L2 ?a ; ?3? L 3 k A2 where q is the resistivity. Eq. (3) can guide the choice of the material and the geometrical design of the structure. The important material parameters are a, q, and k combined

in the ?gure of merit aq/k, for which a large value is desirable. Table 1 shows this ?gure of merit for several common metals that are typically available for electroplating in the major step of the fabrication process. Among these materials nickel is the clear choice, with a ?gure of merit an order of magnitude larger than for gold or copper.

Table 1 Physical properties of metals Material Thermal expansion coe?cient, a (lm/lm °C) 14.2 ? 10?6 16.5 ? 10?6 12.7 ? 10?6 Electrical resistivity, q (X lm) 0.022 0.017 0.0693 Thermal conductivity, k (W/lm °C) 32 ? 10?5 40 ? 10?5 9.05 ? 10?5
aq k

(X lm2/W) 0.98 ? 10?3 0.7 ? 10?3 9.7 ? 10?3

Au Cu Ni

W. Wang et al. / Microelectronic Engineering 85 (2008) 587–598

591

Heat input by electrical current through a small metallic structure equilibrates quite quickly. The typical thermal time constant of the nickel structures we have modeled is 0.5 s or less. 3.2. Mechanical forces and structural compliance of the inchworm Mechanical compliance is de?ned as the ability of a mechanical structure to comply with deformation forces. It is the reciprocal of the mechanical sti?ness of the structure. The inchworm has two types of essential motion: extension and retraction (in the X direction) and release and reclamping (in the Y direction). The elastic deformation of the clamp arm in both the X and Y directions is important in the actuator sequence that involves these two motions. The dimensions of the actuator determine these compliances. As shown in Fig. 3, the main beam of each ‘‘E”-shaped actuator is 2000 lm long and 1000 lm wide. Each clamp arm is 350 lm wide and 1300 lm long. The compliance of the clamp arm in Y is critical because it is a major factor in determining the frictional force between the clamp arm and the pusher. This friction is essential for driving the pusher along the X-axis, for de?ecting the optical ?ber (or other load) as the pusher moves, and for holding the position of the ?ber without power after ?ber alignment is complete. This frictional force is the product of the coe?cient of friction between the clamp arm and the pusher and the force in the Y direction exerted by the clamp arm on the pusher. This force, in turn, depends on the compliance of the clamp arm in Y and the displacement of the clamp arm that results from the width of the pusher. The Y compliance has been determined by FEA by applying a force in the Y direction on the end of the clamp arm and calculating the Y de?ection. For the structure as designed, the compliance is 0.0385 lm/ mN (or the sti?ness is 25.96 mN/lm). Thus, for every lm of Y motion the clamp arm has from its release position without the pusher in place to its position when it is clamping the pusher, the clamping force will be roughly 26 mN based on the arm sti?ness. In the design, this Y motion of the end of the clamp arm is 4 lm, so the clamp force has a value of about 100 mN. The compliance of the clamp arm in the X direction is also important in the inchworm operation. In the steps of the actuation cycle in which only one clamp arm is operated, there is bending of the clamp arms as illustrated, for example, in the steps in Fig. 4. The X compliance that is important for that bending has been determined by FEA by applying a force in the X direction on the end of the clamp arm and calculating the X de?ection. It has the value 0.0804 lm/mN (or a sti?ness of 12.44 mN/lm). If this compliance is too small, the clamp arm will not bend, but will slip on the pusher and fail to provide the desired function of steps in the sequence. If it is too large, the position of the pusher (and the optical ?ber) is not well de?ned.

3.3. Actuation cycle design Time-dependent FEA has been carried out using ANSYS to determine the transient behavior of the device and to optimize its performance. The analysis is based on the dimensions of the inchworm as shown in Fig. 3. The structure is designed to be 30 lm thick nickel to allow the use of practical heating currents. The locations labeled contact1a and contact1b are mechanical and thermal anchors ?xed on the substrate for the left side actuator. Contact2a and contact2b have the same functions for the right side actuator. These contacts are the places for inputting current for electro-thermal actuation. During the actuation analysis, we studied the thermal coupling of the movements of extension/retraction and release/reclamping. Ideally, one would add a thermal insulator between the main arm and the release arm, but this would be very dif?cult to fabricate. Without this, it is impossible to have the extension/retraction and release/reclamping motions completely independent of each other. Reducing the interaction of these motions is a critical feature of the design. The thermal actuation steps are controlled by two electrical inputs on the chip (one pair of contacts to each of the actuators) as shown in Fig. 2. Since the thermal actuation of all steps is coupled, the actuation cycle to achieve +X and ?X has been carefully designed and examined by computer modeling. The inchworm motor can actively move the optical ?ber in the +X direction by moving the pusher in the +X direction; it can move the optical ?ber in the ?X direction by reversing the motion of the pusher. Fig. 4 shows one actuation cycle in +X and one in ?X as proposed in this work. The red dashed lines indicate the initial position. The two ‘‘E”-shaped actuators are marked as left and right. At step 0, the structure is at rest at the starting position at room temperature. The actuation starts by inputting current (and hence generating heat) into the structure on one side, e.g. the right side (Fig. 4(a), step r). As a result of the expansion of the main beam, the clamp arm pushes the pusher in the +X direction by a displacement of Dx. Since the other side (left side) clamp arm is still trying to hold the initial position of the pusher, the movement of Dx is only half of the displacement the right side would have if it moved without restriction from the left. The two clamp arms are mechanically bent at this point. Since the current causes the whole structure to heat, the metal structure will expand in all directions, especially the release arm in the Y direction. This reduces the clamping force from the right side and results in the pusher being moved toward the right set of sliders to regain force balance. Current is then input into the left side release arm to extend that arm, followed by a release of the left side clamp arm while the right side arm is still holding the pusher at the extended Dx position (Fig. 4(a), step s). As soon as the left side is completely released from the pusher, the pusher will rapidly move to the full displacement caused by the thermal extension in the right side actuator (Fig. 4(a), step t). At this time, due to the heat

592

W. Wang et al. / Microelectronic Engineering 85 (2008) 587–598

Left Starting position

YL X YR
6
Cooling from Right

0
Right

Extension from right

1

Starting release from Left

2

7

Starting release from Right

Fully release from Left

3

8

Fully release from Right

Starting reclamp from Left

4

9

Starting reclamp from Right

Fully reclamp from Left

5

10

Fully reclamp from Right

Left Starting position

YL X YR
6
Cooling from Right

0
Right

Extension from Right

1

Starting release from Right

2

7

Starting release from Left

Fully release from Right

3

8

Fully release from Left

Starting reclamp from Right

4

9

Starting reclamp from Left

Fully reclamp from Right

5

10

Fully reclamp from Left

Fig. 4. Actuation cycles of motions in +X and in ?X. (a) Forward and (b) backward.

W. Wang et al. / Microelectronic Engineering 85 (2008) 587–598

593

introduced into the main beam of the actuator in the release step, the left side is also extended a little beyond its initial starting position. As discussed above, the transient temperature in the main beam, generated by the release current which is higher than the extension current, would be much higher than the temperature reached in the extension step if it were allowed to reach steady state. This would cause signi?cant extra extension to the main beam and lead to inaccurate positioning of the actuator. However, the spreading of heat into the whole body of the actuator takes time. A fast release reduces the spreading of heat into the main beam before release occurs. At this point, the positions of the left side clamp arm (after fast release) and the right side clamp arm (fully extended) are di?erent. As we will discuss later, the di?erence will decide the net Dx. With the left side release, the force on the right side clamp arm will also drive the pusher against the opposite side of the sliders. Once the pusher moves quickly (to full extension), the release current in the left side is turned o? to start reclamping the pusher (Fig. 4(a), step u). The reclamp of the left side clamp arm begins after achieving a su?ciently large frictional force between the clamp arm and the pusher. Before that moment, but after it comes in contact with the pusher, the clamp arm just slides along the surface of the pusher as it retracts in X from the extension produced in the main beam by the pulse of release current. The clamp arm of the left side will then drag the pusher back to half of the remaining extension displacement of the left side caused by the pulse of release current, for reasons similar to those in step 1. After this step, the clamp arms are both in a deformed position (Fig. 4(a), step v). At this time, the pusher will again be driven against the slider on the right side due to greater clamping force on the reclamped left side. Retraction of the extended right side arm and holding the pusher’s position are the next few steps in this cycle. First, the current in the extended right side is turned o? to let the clamp arm drag the pusher back while the left side is still holding the pusher (Fig. 4(a), step w). The result is a new deformed structure, and the pusher is back to the center position between the two sets of sliders. After it has completely cooled, the release current is input in the right side actuator for a quick release (Fig. 4(a), step x). Once the right side completely releases the pusher, the elastic deformation on the left side clamp arm will recover to its neutral state (initial position) and the pusher will be pushed towards +X and against the right side sliders as well by the clamping force from the left side (Fig. 4(a), step y). Even though the release is in a quick mode, heat still extends the main beam. (Fig. 4(a), step z). In this step, the pusher is continually pushed against the right side sliders. When the release current is turned o?, the clamp arm of the right side will reclamp the pusher again and drag the pusher back (Fig. 4(a), step {). The pull-back displacement is only half of the extension caused by the release heat as discussed in connection with the left side reclamp. The

pusher is also being pushed back to the center position between the sliders because of the increased clamping force from the right side. As long as the full extension in step 3 is larger than the extension caused by the release current in step 4, a net positive displacement (Dx) will be achieved. It should be also noted that the compliance of the clamp arm in the X and Y directions is important to keep the pusher clamped from one side or the other, or both, at all times. Backward motion of the pusher is similar to the forward motion, but requires a di?erent sequence of steps. It also starts with extension on one main beam, e.g. the right side, by inputting electrical current and reaching the steady state (Fig. 4(b), step r). The pusher is pressed against the right side sliders due to reduced clamping force from the right side. Instead of releasing the left side clamping, the right side clamp is released ?rst to let the pusher recover to its initial position (Fig. 4(b), step s and t). In these steps, the left side force keeps the pusher against the right sliders. After a quick release, reclamping from the right side will drag the pusher backwards and cause some elastic deformation at the force balanced state (Fig. 4(b), steps u and v) due to the extra extension caused by the fast release. The pusher still stays against the right side sliders in these steps. The cooling of the right side actuator moves the pusher further backward to achieve net negative displacement (?Dx) at the end of the actuation cycle (Fig. 4(b), step w). The clamping force from the right starts increasing and the pusher begins to be driven away from the sliders. After this step, the elastic deformation in the arms is more obvious and the pusher moves back to the center position between the two sets of sliders. The quick release of the left side lets the elastic deformation recover and achieves the full negative movement of the pusher (Fig. 4(b), steps x and y), and the pusher is driven to the left pair of sliders due to the left side release. The reclamp of the left side will force the pusher a little more backwards because of the extra extension caused by release current as shown in Fig. 4(b), steps z and {. The pusher also moves back to the center position because the two clamping forces are equal. In order to achieve movement and locking of the pusher, there are several key considerations as discussed previously in connection with the actuation cycles. There must be enough frictional force between the clamp arm and the pusher for the arm to drive the pusher. This force is determined by the clamping force and the coe?cient of friction between the clamp arm and the pusher. As mentioned above, the clamping force is created by the dimensional difference between the default spacing of the two clamp arms and the width of the pusher, and by the sti?ness of the clamp arms. The dimensional di?erence is limited by the release distance that can be achieved with the release arm. Finally, to create a net Dx in either direction, the extension achieved in the extension step must be larger than the extension caused by fast release.

594

W. Wang et al. / Microelectronic Engineering 85 (2008) 587–598

3.4. Transient analysis of actuation motions Con?rming the ability of the structure to follow the sequence just described requires a transient analysis of the device response to current input. Fig. 5(a) shows the transient analysis of the forward actuation motions in the absence of the pusher. Fig. 5(a) is the combined transient results of left and right side. Fig. 5(b) shows details of the release motion of either actuator (in this case, the left side). For instance, the X displacement curve in Fig. 5(b) is the magni?ed equivalent of the X displacement curve on the left side in Fig. 5(a). The Y displacement curve on the left side of Fig. 5(a) is not plotted in Fig. 5(b) because of its large values. A pusher is used which is 8 lm wider than the space between the two clamp arms when the arms

are at room temperature. The pusher is inserted between two clamp arms. This is most easily done if the clamp arms are both provided low steady-state release currents so that their spacing increases by more than 8 lm. In the transient release step in Fig. 5(b), the Y release displacement of the end of a clamp arm must be at least 4 lm. In order to successfully achieve the inchworm motions as described above, a steady-state extension of one main beam is required as the ?rst step of the actuation sequence in this design. When inputting 0.87 A DC current to the right side, a steady-state temperature is reached in about 2 s. The X-axis displacement in this single actuation step (shown in the ?gure as UX) is 1.37 lm and the Y-axis displacement (shown in the ?gure as UY) is 0.98 lm. In this step, UY reaches a bigger value of 1.64 lm because of

1.8

1.4

Extension – Right

UX - Right side UY - Right side UX - Left side

1

Displacement (um)

Retract – Right Release – Left

Release – Right

0.6

0.2

0.00 -0.2

0.50

1.00

1.50

2.00

2.50

3.00

3.50

-0.6

-1

Time (s)
4.5

4.0

UX UY

3.5

Displacement (um)

3.0

2.5

2.0

1.5

1.0

0.5

No slippage zone
0.002 0.004 0.006 0.008 0.010 0.012 0.014 0.016 0.018

0.0 0.000

Time (s)

Fig. 5. Transient analysis.

W. Wang et al. / Microelectronic Engineering 85 (2008) 587–598

595

the expansion of the release arm. Then the heat spreading into the whole body makes the clamp arm longer which causes the UY value to decrease from its maximum. Although the clamping force is reduced at this point

because of the net UY displacement, the clamping force is predicted to be 100 mN and the frictional force it produces will be large enough to ?rmly hold the pusher. The transient left side release involves inputting 1.5 A DC

Fig. 6. Recessed anchor for reduced stress.

596

W. Wang et al. / Microelectronic Engineering 85 (2008) 587–598

current for a time of 9 ms. This value of the current is the maximum allowed for a gold bonding wire of 10 mm length and $50 lm diameter. This current limit can be raised by using di?erent bonding wires. In this period, UY has a displacement of more than 4 lm and at the same time UX reaches 0.94 lm as the heat starts spreading into the whole structure. We note that the clamp arm will lose e?ective clamping force during the 9 ms transient (not just at the end) because the clamping force and hence the frictional force is declining as UY increases. The clamp arm will eventually slide on the pusher, but this has no impact on the movement of the pusher. A no-slippage zone is marked in the ?gure to represent the region of e?ective clamping.

After the pulse of release current, the release arm cools down. Since it has not introduced much heat throughout the structure because of its short duration compared to the steady-state time constant, the cooling is also fast (around 9 ms). When the extension current in the right side is turned o? and the right side cools down, the left side holds the position of the pusher. In the ?gure, UY becomes negative in the absence of the pusher and it indicates the clamping force is increased during the cooling down in the whole structure. After that, the release of the right side clamp arm is carried out. This simulation validates the proposed actuation cycle described in the previous section by con?rming the assump-

Fig. 7. Compliance and stress in the structure. (a) Compliance in Y when applying 90 mN on the clamp arm in Y. (b) Stress in Y when applying 90 mN on the clamp arm in Y.

W. Wang et al. / Microelectronic Engineering 85 (2008) 587–598

597

tions and requirements. The structure dimensions are designed to satisfy the compliance requirements in both directions in the actuation. A net 0.43 lm Dx movement can be achieved in both forward and backward directions with the conditions described above. Theoretically if the input current in the extension step is adjusted to make the UX 1.1 lm, a net Dx of 0.1 lm can be achieved in one cycle. 4. Mechanical forces and stresses In addition to the compliance issues discussed in the previous sections, stress is an important factor when designing the inchworm. Since the device is built of a metal ?lm, stress induced during actuation might damage the structure. In this design, reducing the stress in high stress regions is critical. In the early stage of our design, we found very high stresses adjacent to the anchor at the base of the main beam. Proper design greatly reduces this stress. As [15] and [16] describe, the yield stress for a nickel thin ?lm is around 700 MPa which means the stress caused in actuation motions should not exceed this number or the structure may fail. Fig. 6 illustrates the observations of the stress distribution in di?erent anchor places. The hinge is designed as a tapered structure in order to give the structure more freedom in Y-axis movement. This same consideration also applies to the connection between the main beam and the release arm. The maximum stress occurs in the release step. If the anchor is ?xed at the edge of the hinge, the maximum stress is 730 MPa on the ?xed side and the stress is non-uniformly distributed. In the release step the temperature rise causes expansion of the metal structures, so the free side of the anchor has more freedom to expand and the stress accumulates on the ?xed side. During release, large stress also exists around the hinge position due to the mechanical movement. In conclusion, the stress distribution is markedly non-uniform. However, if the anchor is ?xed 30 lm away from the edge of the hinge in all dimensions, the stress during the same release conditions can be reduced to a maximum of 350 MPa. Although stress must be minimized in the hinge regions, some stress is necessary elsewhere in the structure. In order to achieve large frictional force in the actuation cycle, large clamping forces are needed. The proposed design will achieve approximately 100 mN of clamping force with 4 lm deformation on the clamp strut end after assembly at the initial position (Fig. 7(a)). Under this clamp force, the maximum stress is around 475 MPa in the truss structure. Fig. 7(b) shows the distribution of the stress along the Y-axis at the location identi?ed in the truss structure. The narrowed connection between release arm and main beam is preferred for several reasons. It provides more freedom in rotation and reduced stress at the connection that is of main concern during release. It also slows the spreading of heat into the main beam during the quick release because of the narrowed cross section. A larger clamp force could be achieved by certain design changes, but the necessary release distance of the actuator does not allow this. In

the current design, the maximum release distance is a little more than 4 lm while the net Dx can be achieved as long as the displacement di?erence in X-axis between extension and release can be realized. Although one might guess that the release arm is designed as a truss for mechanical reasons, the major consideration is actually the high electrical resistance the truss provides. As Eq. (3) states, the higher resistance provides more release arm extension which means more release distance at an attainable current. The cross-members of the truss are 5 lm wide and the sides are 10 lm wide because, as shown in Fig. 7(b), the stress is concentrated in the sides and the width is necessary to prevent structure failure. The ANSYS calculation indicates the resistance gain compared to a solid structure of the same size is a factor of 7.2. 5. Fabrication and assembly The fabrication process involves a subset of standard MEMS processes: optical lithography, dry plasma etching, sputtering, and electroplating. An SOI wafer is selected as the substrate of the inchworm structures. The thickness of

Fig. 8. Fabrication process of the actuator. (a) Optical lithography and DRIE to transfer the pattern to the silicon layer. (b) Sputter Cr/Au as seed layer. (c) Coat thick photoresist and pattern it. (d) Electroplate nickel and polish the top surface with CMP if necessary. (e) Strip photoresist and xenon di?uoride (XeF2) isotropic Si etch to remove silicon sacri?cial layer.

598

W. Wang et al. / Microelectronic Engineering 85 (2008) 587–598

Actuator
Pusher

B’ – B’

?=125 um

40 um

Optical fiber

Sliders
Fig. 9. Details of coupling to ?ber.

the device layer determines the clearance of the inchworm structures. Fig. 8 shows the fabrication process of the inchworm motor. First, optical lithography and DRIE are used to pattern the device silicon layer for the inchworm anchors (Fig. 8(a)). Then a seed layer of Cr/Au is uniformly deposited along the surface of the substrate by sputtering (Fig. 8(b)). After that, a thick layer of photoresist is coated and patterned with the layout of the inchworm’s main beam and arms by optical lithography (Fig. 8(c). Nickel is then electroplated on the open areas up to the desired thickness. Chemical mechanical polishing (CMP) may be needed to polish the top surface to have uniform thickness of the structures (Fig. 8(d)). After achieving the desired thickness of the structures, gold wire bonding is carried out to connect the anchors with external electrical connection. Stripping the photoresist and using XeF2to isotropically etch Si to remove the silicon sacri?cial layer are the ?nal steps to obtain the inchworm structures (Fig. 8(e)). The pusher is assembled after the inchworm motor is fabricated. It is made of ceramic to avoid thermal or electrical current ?ow through the pusher from the clamp arms. The optical ?ber is coupled with the pusher as shown in Fig. 9. The assembly of the pusher can be done by assembling the pusher in a hot electrical actuation status or by mechanically de?ecting the two sides. 6. Conclusion The design and modeling of a novel thermal inchworm MEMS device are presented in this report. Force analysis helps understand the gripping capability of the inchworm motor for the special case of ?ber optic positioning. Transient FEA analysis of the operating sequence is of central importance in modifying and verifying the feasibility of the design. Compared to previous research e?orts in inchworm motors [7–11], this new design can achieve powerfree latching as well as high resolution, bidirectional movement, and large displacement and force to meet

the needs of other applications in addition to optical ?ber alignment. Acknowledgements Funding support for this research was provided by the State of Pennsylvania Department of Commercial and Economic Development under BFTDA Contract 24-1160001 and Center of Optical Technology, Lehigh University. References
[1] R.R.A. Syms, H. Zhou, J. Yao, D. Uttamchandani, J. Stagg, J. Micromech. Microeng. 14 (2004) 1633–1639. [2] W.C. Tang, Tu-Cuong, H. Nguyen, R.T. Howe, Proc. IEEE MEMS Conf. (1989) 53–59. [3] S.H. Kim, I.H. Hwang, K.W. Jo, E.S. Yoon, J.H. Lee, J. Micromech. Microeng. 15 (2005) 1674–1682. [4] M.P. de Boer, D.L. Luck, W.R. Ashurst, R. Maboudian, A.D. Corwin, J.A. Walraven, J.M. Redmond, J. MEMS 13 (2004) 63–74. [5] T. Akiyama, K. Shono, J. MEMS 2 (1993) 106–110. [6] N. Tas, J. Wissink, L. Sander, T. Lammerink, M. Elwenspoek, Sensors Actuators A 70 (1998) 171–178. [7] X. Chen, B. Cousins, M. McEllistrem, R. Hamers, Rev. Sci. Instrum. 63 (1992) 4308–4313. [8] J. Park, L. Chu, A. Oliver, Y. Gianchandani, J. MEMS 10 (2001) 255–262. [9] H.N. Kwon, S.H. Jeong, S.K. Lee, J.H. Lee, Sensors Actuators A 103 (2003) 143–149. [10] W. Wang, S. Tatic-Lucic, W. Brown, J. Iceman. S. Hyun, R. Vinci, Precision in-package positioning with a thermal inchworm, Sensors Actuators A (accepted). [11] J.M. Haake, M.W. Beranek, IEEE Elctron. Comp. Technol. Conf. (1998) 1446–1449. [12] L. Eldada, Opt. Eng. 40 (2001) 1165–1178. [13] R. Jebens, W. Trimmer, J. Walker, Sensors Actuators 20 (1989) 65– 73. [14] See <http://www.?ber-optics.info/>. [15] V.N. Sekhar, V. Srinivasarao, R. Jayaganthan, K. Mohankumar, A.A.O. Tay, IEEE Electron. Packag. Technol. Conf. (2004) 610–614. [16] N.R. Moody, J.M. Jungk, M.S. Kennedy, S.V. Prasad, D.F. Bahr, W.W. Gerberich, Mechanical Properties of Wear Tested LIGA Nickel, Materials Research Society Symposium Proceedings, vol. 841, R7.8.1, 2005.



推荐相关:
网站首页 | 网站地图
All rights reserved Powered by 大学生考试网 9299.net
文档资料库内容来自网络,如有侵犯请联系客服。zhit325@qq.com