Micro Electro Mechanical System Design - James J. Allen Part 7 pptx

Figure 5.18 shows the layout and cross-section of the low- hub and pin joint fabrication in SUMMiT.. Figure 5.20 shows a cross-section of the SUMMiT fabrication sequence forthe low-clear

FIGURE 5.7 Layout of the seismic mass level of an accelerometer utilizing only one

drawing layer.

FIGURE 5.8 Layout of the seismic mass level of an accelerometer utilizing two layers.

© 2005 by Taylor & Francis Group, LLC

FIGURE 5.9 Use of a bit by bit XOR logical function to combine layers (layer and

layer_CUT) to form a mask definition.

FIGURE 5.10 Alternative approaches to layout of an annular MMPOLY2 feature.

© 2005 by Taylor & Francis Group, LLC

SUMMiT Because surface micromachining is an alternating stack of two types

of materials (i.e., structural and sacrificial), mechanical layers are attached byetching a hole or via in the sacrificial material; this enables the next mechanicallayer material deposited to attach to the mechanical material below at the via

Figure 5.13 shows SEM images of various layers of SUMMiT anchored to eachother The attachment between the layers at the via produced by the SACOX_CUTcan be seen between the adjacent layers The term “SACOX_CUT” refers to ageneric operation of opening a via in a sacrificial oxide layer to allow attachment

of adjacent structural layers SACOX#_CUT is the use of a SACOX_CUT on aspecific SACOX# layer

A particular layer cannot be directly anchored to ground in the SUMMiTtechnology because deep SACOX_CUTs are not allowed The SACOX_CUT

FIGURE 5.11 Alternative approaches to the layout of an MMPOLY2 island inside an

MMPOLY2 annular feature.

a MMPOLY2 island within a MMPOLY2 annular feature.

enables the mechanical layers immediately above and below to be attached Forexample, a SACOX3_CUT will enable MMPOLY3 and MMPOLY2 to be

that extends from MMPOLY0 to MMPOLY4 Notice that the SACOX_CUTs areall on top of each other and the size of the SACOX_CUT becomes bigger at eachhigher level The size increase enables the mechanical material at the higher level

to attach to the shoulder of the via to the mechanical material on the level

immediately below This is denoted as a nested anchor.

The nested anchor method can produce the smallest size post possible ever, a nested anchor will have encased silicon dioxide trapped inside the post

How-as shown in the cross-section of Figure 5.15 The encHow-ased silicon dioxide is due

to the inability of the etching processes to remove material completely at locations

that have significant vertical topography This artifact is known as a stringer and

silicon oxide in the post can cause slight deflections [7]; this may be a designconsideration, depending upon the application

A staggered anchor is a method of reducing the amount of encased silicon

approach

5.2.2 ROTATIONAL HUBS

axle The ability to implement this structure at the microscale with no assembly

FIGURE 5.12 Cross-hatch patterns for the SUMMiT V masks.

DIMPLE3_CUT

MMPOLY3

DIMPLE4_CUT

MMPOLY4 SACOX4_CUT

© 2005 by Taylor & Francis Group, LLC

MMPOLY0 (a) One layer anchored to ground

FIGURE 5.14 Example of the five mechanical levels of SUMMiT anchored to each other

and to ground The anchors utilized nested SACOX_CUTs except as noted in the figure where staggered SACOX_CUTs are used (Courtesy of Sandia National Laboratories.)

FIGURE 5.15 Masks and cross-section of a post composed of anchored layers utilizing

necessary is an enabling feature for MEMS devices that require mechanisms.Two methods can be used to produce a rotational hub in the SUMMiT technology:

a cap and post hub and a low-clearance hub, which are discussed next.

A cap and post hub can be implemented in any three-level surface

the SUMMiT technology The central feature of this type of hub is a central postwith a cap of sufficient diameter so that a rotating wheel will be constrainedvertically The figure’s cross-section shows the rotating wheel composed ofMMPOLY1 and MMPOLY2, which are laminated together Functionally, therotating wheel could be only one layer instead of two An MMPOLY3 cap issupported by a post of MMPOLY1 and MMPOLY2

The implementation of the cap and post structure is similar to the anchorsdiscussed in the previous section The clearance for the rotating wheel is defined

by the ability of the lithography process to etch layers MMPOLY1 andMMPOLY2 at the rotating interface The vertical clearance is defined by thethickness of the sacrificial oxide layer or the ability to produce structures such

as dimples to constrain the vertical motion Dimples are small “bumps” neath surface micromachined layers that prevent broad area surface contact whenthe layers contact the substrate or each other Dimples can also be used tominimize clearances

under-The low-clearance hub is a feature that SUMMiT was especially designed to

films of sacrificial material accurately (i.e., silicon dioxide) to control the ance in the hub Figure 5.18 shows the layout and cross-section of the low-

hub and pin joint fabrication in SUMMiT A pin joint is very similar to a hub,but is not attached to ground A pin joint enables linkages between rotatingmembers, as shown in Figure 5.19

Figure 5.20 shows a cross-section of the SUMMiT fabrication sequence forthe low-clearance hub at several key points in the process:

been deposited and patterned to produce dimples and anchorMMPOLY1 MMPOLY1 has been deposited and patterned with thePIN_JOINT_CUT mask A combination of anisotropic and wet etchinghas been performed to form the features beneath MMPOLY1

deposited, patterned, and etched and the MMPOLY2 layer deposited

At this stage, SACOX2 can be seen to define the clearances in theinternals of the low-clearance hub The low-clearance hub lateral and

been performed This etch can etch the laminated MMPOLY1 andMMPOLY2 layers, thus providing an even outside surface for the

© 2005 by Taylor & Francis Group, LLC

rotating wheel and etch release holes through the rotating wheel disc.Note that the MMPOLY2 etch stops on the SACOX2 layer in theinternal hub features.

• Figure 5.20d shows the cross-section of the released low-clearance huband pin joint structure

5.2.3 POLY1 BEAM WITH SUBSTRATE CONNECTION

The MMPOLY1 beam with a substrate connection is a simple structure illustratingthe application of two features useful in design of a number of devices in theSUMMiT technology The MMPOLY1 layer can be patterned in either of twoways in SUMMiT:

MMPOLY1_cut mask and etch

SACOX2 as a “hard” mask and etching with the MMPOLY2 etch

In the previous section, Figure 5.20c showed that the MMPOLY2 etch wouldetch the MMPOLY1 and MMPOLY2 layers except when the MMPOLY1 layer

is protected by SACOX2 In this case, the SACOX2 layer was used as a “hard”

5.21, the SACOX2 mask is used to define the MMPOLY1 beam via theMMPOLY2 etch The MMPOLY1 beam is attached to the substrate using a

to the substrate for electrical grounding purposes, the NITRIDE_CUT mask isused to define the etch of the NITRIDE layer

5.2.4 DISCRETE HINGES

The concept of discrete hinges for MEMS applications was initially proposed byPister [8] Since that time, a number of different variations and types of hinges

FIGURE 5.19 A focused ion beam (FIB) cross-section of a rotational hub and pin joint.

(Courtesy of Sandia National Laboratories.)

© 2005 by Taylor & Francis Group, LLC

etch laminated MMPOLY1 and MMPOLY2

have been developed and applied [9] Figure 5.22 shows SEM images of a up” mirror that utilizes two types of discrete hinges:

the layout and cross-section for a staple and pin hinge implemented inSUMMiT The cross-section shows two posts that go up to the

MMPOLY3 level, which bridges between the posts to form the staple.

The pin is a narrow piece of laminate MMPOLY1 and MMPOLY2 thatconnects out of the plane of the cross-section to the movable plate As

in a cap and post hub, the staple and pin hinge clearances are defined

by the width of the MMPOLY2 etch

about an axis parallel to the plane of the plates and the hinge to deflectoff the substrate, as shown in Figure 5.22 The design of the plate-to-

layout and two cross-sections of the device Figure 5.24 shows theindividual masks required to fabricate the hinge in SUMMiT technol-ogy, as well as the composite mask, “stacked” together To fabricatethis device as well as any other device in SUMMiT requires the masks

be aligned precisely to each other The A-A cross-section shown inFigure 5.25c reveals the major parts of this hinge design AnMMPOLY2 pin is connected out of the plane of the cross-section toplate 2 The pin is trapped by the staple and floor and rotates withinthese objects Figure 5.25b shows that the SACOX2 layer separatesMMPOLY1 and MMPOLY2 and stops the MMPOLY2 etch so that anMMPOLY2 pin is formed The staple is formed by utilizingSACOX3_CUTS to attach MMPOLY3 in two places to MMPOLY2

A subtle feature is the cutting of MMPOLY1 as defined by theMMPOLY1_CUT mask to separate the pin and the floor structures asshown in Figure 5.25d and Figure 5.25e Without the separation of thefloor and pin structures via the MMPOLY1_CUT, this hinge wouldrigidly attach plate 1 and plate 2 (i.e., not functional)

FIGURE 5.21 An MMPOLY1 beam with a substrate connection.

SACOX1_CUT

SACOX2

substrate connection

nitride silicon dioxide substrate

NITRIDE_CUT MMPOLY0

© 2005 by Taylor & Francis Group, LLC

FIGURE 5.22 Discrete hinges utilized in a “pop-up” MEMS mirror design implemented

in SUMMiT (Courtesy of Sandia National Laboratories.)

FIGURE 5.23 Staple and pin hinge SUMMiT masks and cross-section.

(a) SUMMiTTMlayout

(b) A-A cross-section

moveable plate

pin staple

encased oxide

© 2005 by Taylor & Francis Group, LLC

(f) Aligned composite masks for the plate-to-plate hinge

(b) cross-section A-A after MMPOLY2 etch (c) cross-section A-A after release etch

5.3 DESIGN RULES

The term design rules originally comes from the microelectronics industry These

rules are a formal communication between the fabrication engineer and the designengineer For the microelectronics industry, design rules are the layout rulesrequired to obtain optimum yield (functional devices vs nonfunctional devices)

in as small an area as possible without compromising circuit reliability

Design rules for MEMS fabrication are also a formal communication betweenthe fabrication and design engineers MEMS fabrication processes are similar tomicroelectronics; however, due to the additional ability of motion MEMS devices(e.g., inertial sensors, mechanisms, pumps, valves, etc.) they are very differentfrom microelectronics (i.e., electronic circuitry) and more varied in function andapplication The varied function and application of MEMS devices make theassessment of yield from the perspective of design rules for a general-use MEMSfabrication process difficult to define Design rules for MEMS fabrication pro-cesses are the layout rules required to produce MEMS devices with minimaldefect with the smallest feature sizes possible MEMS device functionality andreliability are generally very specific to the device design and cannot be totallyencompassed by MEMS layout design rules alone

The layouts for microelectronic and MEMS design are very complex andinvolve a number of mask layers The mechanics of automated design rulechecking for VLSI circuitry layout was established during the rise of the micro-electronics industry [10–12] Yarberry [4] discusses the implementation of auto-mated design rule checking for a MEMS fabrication process, SUMMiT™

the basis for the specification of design rules and the implementation of designrules for a MEMS fabrication process

5.3.1 MANUFACTURING ISSUES

5.3.1.1 Patterning Limits

The definition of MEMS device features is a function of the patterning and etchsteps of the fabrication processes used These processes may be utilized for

used to etch the pattern into the MEMS material The patterning limits will controlthe smallest feature that can be realized in a MEMS device This patterning limit

is frequently called the feature size or CD (critical dimension) The patterning

limits of a fabrication technology are expressed as line width and space design

expression of a line width and space design rule for a layer in the SUMMiT™

lithograph-ically patterning has a “rounding” effect of the sharp-cornered features due tothe patterning limitations

© 2005 by Taylor & Francis Group, LLC

FIGURE 5.26 Example of a line width (W) and space (S) design rule.

FIGURE 5.27 Expected result when line width and space design rule violated.

FIGURE 5.28 “Rounding” of a lithographically patterned angular feature (Courtesy of

Sandia National Laboratories.)

a < minimum feature size

(a) mask (b) resulting patterned

features

© 2005 by Taylor & Francis Group, LLC

5.3.1.2 Etch Pattern Uniformity

Etch pattern uniformity becomes an issue in many fabrication technologies due

to optical effects of patterning large arrays of similar or varied structures, oretching an array of structures of varying size Figure 5.29 illustrates the etching

of a series of trenches in a bulk micromachining process The ability to removethe etch products from the trench can influence the etch rate of the process.This can be an advantage or a disadvantage, depending upon the device to beproduced

An accelerometer produced in a surface micromachining process generallyhas large banks of comb fingers and electrodes Lithographic optical issues ofpatterning a large arrays of repeating structures such as this may cause the

patterning of the electrodes on the edge to become distorted edge effects This

can be accommodated by adding a few extra “dummy” electrodes to ensure thatthe functioning electrodes are fabricated without distortion

5.3.1.3 Registration Errors

MEM fabrication technologies frequently require masks at different stages inthe fabrication process to be aligned for the fabricated device to be produced

as designed For example, the SUMMiT utilizes 14 masks that need to aligned

to each other The alignment is accomplished by the aid of alignment targets(Figure 2.26), which are etched into each layer to enable the alignment Posi-tioning objects on a mask is accomplished to computer precision; however,

aligning masks relative to each other has a finite precision, called registration

Figure 5.30 illustrates an enclosure design rule that ensures that theSACOX1_CUT feature is inside the MMPOLY2 feature One of the things thatdetermines the amount of enclosure is determined by the registration errors ofthe mask alignment

FIGURE 5.29 Pattern uniformity affects the ability to remove etching products and

influ-ences etch rate (Courtesy of Sandia National Laboratories.)