Ghodssi R., Lin P., MEMS Materials and Processes Handbook

Подождите немного. Документ загружается.

734 D.R. Hines et al.

sample with respect to the microscope. For focus and alignment, it is desirable

for the sample holder also to contain a calibration standard sample s uch as

gold nanoparticles and a Faraday cup. The Faraday cup is a hole in the sample

holder into which t he electron beam can be directed for measuring the beam

current.

6. Vent the antechamber, load the sample holder into the SEM, and pump a vac-

uum on the antechamber. Once the desired vacuum level has been established,

move the sample holder into place in the microscope column.

7. Move the sample holder to position away from the sample and near the Faraday

cup. Some SEMs allow different sample holders to be speciﬁed and speciﬁc

spots on a given sample holder to be deﬁned. In such a case, move to the

predeﬁned position of the Faraday cup.

8. Starting with the magniﬁcation at maximum, establish the electron beam in the

SEM column, for example, open the gate valve and turn on the detector. Slowly

decrease the magniﬁcation (and adjust the focus if needed) to image the Faraday

cup. Center the electron beam in the Faraday cup and adjust the beam current

to the desired value.

9. Move the sample holder to the calibration standard sample and focus the SEM.

How well focused the SEM is at this point will have a large impact on the

resolution and quality of the lithography; so, take your time and do a good job

here.

10. Locate the edge of the sample containing the scratch and move along the scratch

unit until you reach the end. Doing this at too low a magniﬁcation can cause

unwanted exposure t o the resist layer. Doing this at too high a magniﬁcation

can take “forever.”

11. Once at the end of the scratch, refocus on a Si shard. Do this ﬁrst by adjust-

ing the height of the sample (z adjustment) and then ﬁne-tune by adjusting

the focus itself. It is best to not have to adjust the stigmators during this ﬁnal

focus.

12. Switch over to NPGS control of the SEM and turn on the beam blanker. Set the

SEM to the desired magniﬁcation. This is typically 1000x. Move the sample to

the desired location or exposure.

13. Select the desired run ﬁle and verify that it contains the desired settings. Run the

run ﬁle (click on the “process run ﬁle” button). After the exposure is ﬁnished,

move to another exposure location and repeat the exposure using the desired

run ﬁle.

14. Once all the desired patterns have been exposed, set the SEM to the maximum

magniﬁcation, turn off the electron beam, turn off the beam blanker, and switch

back to SEM mode.

15. Move the sample holder to the “home” position and remove it from the SEM

column. Vent the antechamber and remove the sample holder. Remember to

close the antechamber and leave it under vacuum.

16. Develop the sample in a 1:3 solution of MIBK:IPA for 90 s. Rinse in DI water

for 30 s. The developing time may be sample-dependent so several test samples

might be needed. Samples should now be ready for further processing.

9 MEMS Lithography and Micromachining Techniques 735

9.7.8.1 Notes on Using the NPGS Software

The NPGS and DesignCAD software can be loaded onto an ofﬁce PC so that ﬁles

can be prepared ofﬂine from the SEM. Make sure that within the DesignCAD

application that the features are drawn using “PolyFill” under the NPGS menu bar

selection. Select a feature and use the info box to assign it to a speciﬁc layer. Choose

a different color for each layer. Within a run ﬁle, all the beam control parameters can

be independently assigned for each separate layer. The desired beam control param-

eters in a given run ﬁle are dependent on the speciﬁc SEM and feature that will be

written. Several test exposures will most likely be needed. A good place to start can

be determined by looking at other run ﬁles on the NPGS computer connected to the

SEM or by talking with other e-beam users.

9.7.9 Case Study 9: Fabrication of PDMS Templates

PDMS is a two-part elastomer and is typically formulated as 10 parts base to 1 part

curing agent. The resulting material is transparent in the UV and visible regions,

ﬂexible, and has a low Young’s modulus. The surface is tacky and has a low surface

energy. The Young’s modulus can be somewhat controlled by the ratio of base to cur-

ing agent and curing temperature, however, other formulations have been developed

for higher Young’s modulus [15]. The typical formulation is referred to as soft-

PDMS (s-PDMS) with a Young’s modulus less than 2 MPa. Formulations designed

for higher Young’s modulus (9 MPa) are referred to as hard-PDMS (h-PMDS). For

reference, Si has a Young’s modulus of 130 GPa. Molds made for materials such

as PDMS are referred to as soft mold and are used in soft lithography applications.

Molds made from materials such as Si are referred to as hard molds and are used in

imprint lithography applications. Molds made from materials with a Young’s mod-

ulus between that of PDMS and Si are referred to as rigiﬂex molds [15]. A review

of mold materials can be found in [15].

1. By weight, mix 10 parts PDMS base with 1 part PDMS curing agent. Use a

container that is clean of particulates and other possible contaminants. Make sure

that the container is large enough to contain all the PDMS as it is degassed in

a vacuum chamber and that the container is small enough to ﬁt into the vacuum

chamber.

2. After thorough mixing, place the PDMS into a vacuum chamber and pump a

vacuum into it until no gas bubbles are visible in the PDMS. Depending on the

vacuum level and amount of PDMS, this could take around 20 min. To avoid

contamination of the PDMS it is recommended to use an oil-free pump such as

a scroll pump or diaphragm pump.

3. Once degassed, the PDMS can be applied either by pouring on or by spin coating

onto a substrate. PDMS has a viscosity of 362 cP (about the same consistency as

honey). The viscosity can be decreased by mixing with a solvent. For example

736 D.R. Hines et al.

a dilution of 69% by weight with toluene will lower the viscosity to 50 cP [15].

The speed at which the PDMS ﬂows will depend on the viscosity of the material.

4. After pouring, let the PDMS sit at room temperature long enough to completely

ﬂow over the entire area and reach equilibrium. This could take some time

depending on how much PDMS is used and the size of the area to be covered.

Tilting the substrate around to different angles will allow gravity to assist in get-

ting the PDMS to cover all areas of the substrate. It is best to hold the substrate

from the backside with vacuum tweezers so that the PDMS does not ﬂow off the

edge of the substrate due to wetting the tip of the mechanical tweezers.

5. For spin coating, the thickness of the ﬁlm can be controlled by adding toluene

to lower the viscosity of the PDMS prior to spin coating. After spinning, let the

PDMS ﬁlm stand at room temperature (no baking on a hotplate) until all the

solvent has had a chance to evaporate and the remaining PDMS has reached a

uniform thickness over the entire substrate area.

6. The PDMS can be cured by letting it sit at room temperature for 24 h. Faster

curing is possible such as at 70

◦

Cfor2h.

7. Once cured the PDMS can be left in place or peeled off as a free-standing ﬁlm.

8. For microﬂuidics, the PDMS would be cast onto a mold containing the negative

topology of the desired ﬂow channels. Once removed from the mold, the PDMS

can be placed onto the desired substrate to create the ﬂow channels. Where it

touches the substrate, the PDMS ﬁlm will wet a substrate surface to form a con-

formal contact. With no prior surface treatment, adhesion between the PDMS and

substrate is generally nonpermanent and the PDMS ﬁlm can be easily removed

and repositioned.

9. The adhesion between PDMS and a Si substrate (or a second PDMS ﬁlm) can be

made permanent in the following ways.

a. Place a clean Si substrate (free of particulates or contamination on the sur-

face) into a UVO cleaner with the substrate surface about 2 mm below the

light source and expose for 15 min.

b. Remove the Si substrate and similarly place the PDMS ﬁlm in the UVO

cleaner and expose for 150 s.

c. Remove the PDMS ﬁlm and contact it to the Si substrate. While in contact,

place the assembly into a 70

◦

C oven and heat for at least 20 min. Once

removed from the oven, the PDMS will be permanently bonded to the Si

Substrate. Note that the position of the materials i n the UVO cleaner, duration

of exposure, and amount of “aging” after exposure(s) are all critical parame-

ters affecting the interfacial adhesion strength between the PDMS ﬁlm and Si

substrate (Childs also contains useful information about PDMS processing).

9.7.10 Case Study 10: Photomask Fabrication [226, 227]

Photomasks are typically designed using a CAD (computer-aided design or

computer-aided drafting) software application. GDS (graphic data system) is the

9 MEMS Lithography and Micromachining Techniques 737

standard ﬁle format accepted by mask manufacturers and mask fabrication tools.

Alternatively the DXF (drawing exchange format) ﬁle format commonly associ-

ated with the popular AutoCAD drawing application is widely accepted but can

contain problematic conversion errors. Within the CAD software application, the

desired photomask pattern is best laid out using square or rectangle shapes having

◦

corners. This “Manhattan geometry” avoids circles and slanted lines that can

add complexity and cost to mask fabrication. Features are best laid out on a grid

with 0.125 µmor0.1µm grid size. Smaller grid sizes such as 0.05 µm or 0.025 µm

are possible but can dramatically increase fabrication complexity and cost.

Features should be drawn so that they snap to grid lines rather than fall in

between grid lines. Examples of good and bad features are shown in Fig. 9.49.

Figure 9.49a illustrates a good feature. Figure 9.49b–d illustrate rounded corners,

features not aligned to the grid, and features aligned to the grid but incorporating

non-90

◦

angles, respectively. Features should also be aligned so that they are easily

cleaved or separated after ﬁnal fabrication as illustrated at the bottom of Fig. 9.49.

(a)

(b)

(c)

(d)

(e) (f)

Fig. 9.49 Optimized (a)and

(e) and unoptimized (b)–(d)

and (f) grid layouts for

photomask patterns (Modiﬁed

from [26]; used with

permission)

As in a mechanical drawing submitted to a machine shop, critical dimensions and

tolerances of a mask design need to be considered. Typically, minimum feature sizes

of 5–2.5 µm with tolerances of ±0.5 to ±0.25 µm are cost effective. Device fabri-

cation commonly uses several levels of photolithography which require multiple

masks. Each mask has to be aligned to the previously fabricated pattern; there-

fore, layer-to-layer registration must be accounted for during mask design. Overlay

registrations of 0.5 mm to 0.25 µm over 4 in. are typically achievable.

738 D.R. Hines et al.

Additional issues to keep in mind during mask design [26] include the following.

1. Alignment to crystal axis of substrate (wafer ﬂat is ∼2

◦

off axis)

2. Avoid patterns within 2–5 mm of wafer edge.

3. Don’t mix large and small features in the same mask level.

4. If performing a wet etch, account for lateral undercut of substrate in the mask

design.

5. If performing DRIE etch, account for etch rate as a function of distance from

substrate center. Etch rate is s lower in the center and faster at the edges of the

substrate.

6. Include alignment marks: one big enough to see with the naked eye and one

the size of the smallest mask feature. Strategically place marks so that they are

accessible to the stepper of contact aligner to be used.

9.7.10.1 Photomask Defects

Fabrication errors and damage due to handling can cause defects in pattern fea-

tures on a photomask. Figure 9.50 shows several types of both opaque and clear

defects. After it has been fabricated, careful inspection is usually required to verify

the integrity of a photomask. A visual inspection under an optical microscope can

pick out defects on the order of 3 µm and should always be performed prior to using

a photomask. A visual inspection can identify defect, dirt, and dust. If dirt and dust

are found the photomask should be carefully cleaned.

A more careful inspection for defects can be performed using a KLA die-to-die

inspection system. This is an automated tool that optically compares two simi-

lar areas on a photomask. The machine pixelates optical images and compares

Fig. 9.50 Illustration of

feature defect on a photomask

(Modiﬁed from [226])

9 MEMS Lithography and Micromachining Techniques 739

transmitted light intensities. Discrepancies are registered as possible defects and

ﬂagged for further human inspection and classiﬁcation. A similar die-to-database,

KLA Reticle Inspection System (KLARIS) inspection can be performed where the

photomask is compared to a computer-generated database. Defects that can affect

the performance of fabricated devices must be repaired or a new mask must be fabri-

cated. Defects are typically repaired with a focused ion beam system that can either

remove opaque defects of add material to clear defects [228]. These repair steps

are similar to FIB and gas-assisted ion beam process methods discussed in Sections

9.5.2 and 9.5.3, respectively.

9.7.10.2 Grayscale Lithography Pixelated Photomasks

As discussed above, partial transmission of UV light during the exposure step pixe-

lates mask patterns below the diffraction limit of the optical projection system. The

pixelation is determined by two basic parameters: the pitch and the pixel size. As an

example, for a standard 5x r eduction stepper the resolution at the substrate level is

generally about 0.6 µm which corresponds to a pixel size of 3 µm on the mask. The

largest pixel size in a grayscale mask for this system must be smaller than this 3 µm

resolvable dimension. A convenient choice for the maximum pixel size is thus 2.0

µm with a maximum spacing between pixels of 2.6 µm. The minimum pixel size is

set by the resolution of the mask fabrication tool which is typically limited to a spot

size of 0.6 µm and a minimum beam step of 0.1 µm. The available grayscale levels

would therefore correspond to 15 pixel sizes from 0.6 up to 2.0 µm with a pixel

size step of 0.1 µm. This is just a basic example of how the particular pixelation

would be performed, which will vary depending on the exact projection lithography

system being used.

Practical guides that include discussions about manufacturing photomasks can

be found at the following.

Cornell University:

http://www.cnf.cornell.edu/cnf5_courses.html

Stanford University:

http://snf.stanford.edu/Process/Masks/Masks.html

University of Alberta:

http://www.nanofab.ualberta.ca/site/?page_id=47

Brigham Young University:

http://www.ee.byu.edu/cleanroom/masks.phtml

University of California, Berkeley:

http://microlab.berkeley.edu/ (click on ‘mask making’)

Much more can be found by exploring links at

Harvard University:

http://people.seas.harvard.edu/~jones/lab_arch/nano_facilities/nano_facilities.

html

740 D.R. Hines et al.

9.7.10.3 Mask Manufacturers

Compugraphics

Advanced Reproductions

Image Technology

Inﬁnite Graphics

Applied Image

Photronics

Delta Mask

Microtronics

ADTEK Photomask

9.7.11 Case Study 11: Multiphoton Absorption Polymerization

(MAP)

Multiphoton absorption polymerization (MAP) exploits the inherent optical non-

linear property of multiphoton absorption to localize the fabrication in the region of

high photon intensity, for example, a laser focal point. Such localization allows com-

plex, arbitrary, three-dimensional fabrication, with feature sizes as small as 40 nm

[229]. MAP can be performed with both negative and positive photoresists. The

exposure laser source can be from near IR to visible light [230, 231]. In this case

study, acrylic negative photoresist is discussed and an 800 nm fs laser is used as the

exposure light source.

1. Start with a 25 mm

glass cover slip substrate that has been cleaned with an

oxygen plasma cleaner for 2 min.

2. Coat the glass surface to promote ﬁnal device adhesion by dipping it in

the solution that has 93 vol% ethanol, 5 vol% DI water, and 2 vol% (3-

acryloxypropyl)trimethoxysilane for 12 h under stirring at room temperature.

3. Rinse the cover slip with ethanol for 1 h and dry it in a 95

◦

C oven for 1 h.

4. Prepare a resist by mixing 48.5 wt% of (6) trimethylolpropane triacrylate

(Sartomer, SR-499), 48.5 wt% tris(2-hydroxy ethyl)isocyanurate triacrylate

(Sartomer, SR-368) and 3 wt% 2,4,6- trimethylbenzoylethoxyphenylphosphine

oxide (BASF, Lucirin TPO-L) on rotator for 12 h.

5. Put a drop of well-mixed resist on the acrylate-modiﬁed glass substrate, use

Scotch tape to ﬁx the substrate on a regular glass slide. The tape also works as

a spacer; it is on the order of 100 µm thick but can be made several millime-

ters thick depending on the sample. Cover the resin with a cover slip having a

thickness smaller than the focal distance of the microscope objective (usually

use #1 cover slip). An example is shown in Fig. 9.51.

6. Align the laser beam into the microscope objective. Fix the sandwiched resin

sample on the stage that is mounted on the microscope.

7. Finding the surface of the substrate is critical in MAP. Using the ﬁne-tuning

knob, move the objective up and down until the fabricated line can barely be

9 MEMS Lithography and Micromachining Techniques 741

Fig. 9.51 Illustration of a

sample in MAP [232]

(Reprinted with permission

from [232])

seen on a monitor connected to a CCD camera. Focusing an ultrafast laser into a

prepolymer sample generates a cross-linked polymeric dot. The desired pattern

can be achieved by a moving stage in 3-D with respect to the laser focal point.

A homemade labview program transfers (x, y, z) coordination from a prepared

text ﬁle to stage [233].

8. After determining the location of the surface, run the program to fabricate

desired 3-D microstructures. The laser intensity needs to be adjusted to meet

the requirement of the structure.

9. Upon ﬁnishing fabrication, carefully take off the substrate with the structure.

Develop the resist by rinsing twice in dimethylformamide for 3 min each,

followed by two rinses in ethanol for 3 min each.

10. Dry the sample by either allowing the solvent to evaporate on its own or by

blowing with dry air. Supercritical CO

drying can be performed for ultrafragile

structures.

9.7.12 Case Study 12: Lithography Using Focused Ion Beams

Focused Ion Beam (FIB) has not been widely used for lithography because it is so

powerful a tool for direct micromachining and because it does not give huge advan-

tages over e-beam lithography, which is already well established with its own huge

body of knowledge, software, and tools. We are considering here high-resolution

FIBs (0.005–0.1 micrometer beams) at energies in the range 10–50 keV with heavy

ions such as Ga.

A FIB system can be used for lithography by, for example, adapting software

designed to control an SEM for lithography. This software should be able t o control

beam position and exposure time (ion dosage) via the beam blanker. The major

differences between electron beam lithography and ion beam lithography lie in the

very different sensitivities of resists to ions compared with electrons, and the much

smaller proximity effects because of the secondary ion spectrum of the ion beam at

such low energy. In fact, proximity effects can be ignored.

The process ﬂow for FIB lithography is very similar to that for e-beam lithog-

raphy. The main differences relate to avoiding statistical effects due to the high

sensitivity of the resists, which could tempt one to write too quickly, and the

development of resists. FIB lithography techniques have been developed mostly by

742 D.R. Hines et al.

Melngailis at M.I.T. [234, 235], who used PMMA and novolac-based (negative)

resists (the latter with a unique silylation process), and by Matsui [236]atNEC

Corporation who used PMMA and novolac. The process ﬂows are all generally

similar, including the method below.

We describe here a bilevel spin-on-glass (SOG) process developed by Milgram

at Tektronix and Puretz [237] at the Oregon Graduate Institute to provide high-

resolution negative resist. This resist would be used to provide a precise opening

above a substrate through which implantation or metal deposition could be per-

formed, for example. In the Milgram–Puretz approach a very thin layer of SOG is

spun over a thick resist. The SOG is implanted with Ga ions which all stop in the

SOG. This radically changes the SOG sensitivity in an O

reactive ion etch to make

it equal to that of the underlying thick resist. When both layers are subjected to the

etch the result is an opening where the ions were implanted, with very vertical

sidewalls due to the anisotropic nature of the etch.

1. One begins typically with a Si substrate cleaned as it would be for any lithog-

raphy process. The substrate is then attached to a SEM stub. It is recommended

that a magniﬁcation calibration sample be attached adjacent to the substrate, so

that the FIB magniﬁcation can be precisely known.

2. Two micrometer thick layer of Hunt 206 photoresist (HPR206) is prepared by

spinning on the liquid in the usual manner. This is followed by a 30 min bake at

220

◦

C, in air.

3. A siloxane SOG is applied typically using a pipette, and then spun to produce a

thin (∼125 nm) layer. The glass is then cured at 200

◦

Cinair.

4. The sample can have a ﬁducial mark placed on it using an optical microscope, to

assist in ﬁnding exposed areas. It is then placed in the FIB vacuum chamber and

the system pumped out to normal operating pressure.

5. The beam can be aligned and focused by using the ﬁducial mark, starting at low

magniﬁcation, and proceeding to the magniﬁcation that will ultimately be used

for the resist exposure. Ideally, a Faraday cup would be used to measure the beam

current precisely. Alternatively, if the stage can be biased at ~ +10 V (to suppress

secondary electron emission), beam current can be read directly from the stage.

Magniﬁcation calibration should be performed at the magniﬁcation setting to be

used, prior to resist exposure. In this way ion dosages can be well known.

6. Implants of 10

–10

Ga ions/cm

at a beam energy of 15 keV is sufﬁcient to

expose the desired pattern in the SOG. The FIB needs to be under the control of

the lithography software for this step.

7. After exposure, turn off the ion beam and vent the specimen chamber so the

sample can be removed.

8. The SOG is then etched in an O

reactive ion etch (RIE); a selectivity on the

order of 1:100 has been found for unimplanted versus implanted SOG.

9. The RIE etching of HPR206 shows a selectivity of 100:1 over the unexposed

SOG. The result is an etched layer of material with a highly vertical sidewall (no

trapezoidal shapes) through which implantation into the silicon can be done (see

Fig. 9.52).

9 MEMS Lithography and Micromachining Techniques 743

Fig. 9.52 Example of plasma

processed SOG-HPR206

bilayer lines created with FIB

lithography (Reprinted with

permission from [237])

References

1. G.E. Moore: Cramming more components onto integrated circuits, Electronics 38, 11–117

(1965)

2. W. Brunsvold, W. C onley, W. Montgomery, et al: Acid-sensitive phenol-formaldehyde poly-

meric resists, Polymers for microelectronics – Resists and dielectrics, ACS Symposium

Series 537, 333–347 (1994)

3. Deforest: Photoresist: Materials and Processes (McGraw-Hill, New York, NY, 1975)

4. G.A. Delzenne: In Encyclopedia of Polymer Science and Technology, Plastics, Resins,

Rubbers, Fibers (Interscience, New York, NY, 1976)

5. P.P. Hanson: Unconventional photographic systems – A bibliography of Reviews, Photogr.

Sci. Eng. 14, 438 (1970)

6. J. Kosar: Light Sensitive Systems (Wiley, New York, NY, 1965)

7. W.M. Moreau: Semiconductor Lithography Principles, Practices, and Materials (Plenum,

New York, NY, 1987)

8. D.H. Ziger, C.A. Mack: Generalized-approach toward modeling resist performance, AIChE

J. 37, 1863–1874 (1991)

9. G.K. Ananthasuresh, L.L. Howell: Mechanical design of complicated microsystems – A

perspective & prospects, J. Mech. Des. 127, 736 (2005)

10. S. Kota, G.K. Ananthasuresh, S.B. Crary, et al.: Design & fabrication of microelectrome-

chanical systems, J. Mech. Des. 116, 1081 (1994)

11. Z. Cui: Micro- Nanofabrication Technologies & Applications (Springer, New York, NY,

2005)

12. Z. Cui: Nanofabrication: Principles, Capabilities & Limits, (Springer, New York, NY, 2008)

13. F. Hua, Y. Sun, A. Gaur, M.A. Meitl, L. Bilhaut, L. Rotkina, J. Wang, P. Geil, M. Shim, J.A.

Rogers, A. Shim: Polymer imprint lithography with molecular-scale resolution, Nano Lett.

4, 2467–2471 (2004)

14. S.R. Quake, A. Scherer: From micro to nanofabrication with soft materials, Science 290,

1536 (2000)

15. J.A. Rogers, H.H. Lee: Unconventional Nanopatterning Techniques & Applications (Wiley,

New York, NY, 2009)

16. V.N. Truskett, M.P.C. Watts: Trends in imprint lithography for biological applications,

Trends Biotechnol. 24, 312 (2006)