Anda di halaman 1dari 15

This paper was presented at the 2006 SPIE Microlithography Conference, #6151-125

Using Phase Mask Algorithms to Direct Self Assembly


F.M. Schellenberg* a, J.A.R. Torres
a
Mentor Graphics, 1001 Ridder Park Dr., San Jose, CA, 95131
b
Mentor Graphics, 8005 SW Boeckman Rd., Wilsonville, OR, 97070

ABSTRACT

Recent descriptions of novel directed self assembly (DSA) materials have suggested that exceptional linewidth control
may be achieved when imbalanced block co-polymers assemble in a lithographic setting. By laying down initial patterns
that serve as a guide for polymer film formation, relatively isolated periodic lines 20-40 nm can be formed. Because the
linewidth is dictated by the polymer molecular weight, such a process should be immune to some sources of noise, such as
shot noise, that plague other pattern formation methods at these dimensions.

This technique can work well for the formation of fine features, but larger patterns need to be formed by other means. This
is quite similar to phase-shifting mask (PSM) double exposures, in which periodic patterns are also preferred. We discuss
here the applicability of already developed phase mask data parsing algorithms to fine line formation with DSA materials,
and propose process flows in which these algorithms can be used directly with little or no modification.

Keywords: Directed Self-Assembly, DSA, Block copolymers, BCP, phase-shifting lithography, LER.

1. INTRODUCTION

Conventional lithographic imaging processes, using liquid immersion to increase the effective numerical aperture (NA)1
,
and making use of extensive resolution enhancement techniques (RET),2 3 appear adequate to meet the lithographic needs
of integrated circuits through the 45nm generation. However, roughness and variability issues seem to repeatedly occur
when patterning smaller features, and there seems to be an empirical limit at around 30 nm, beyond which line edge
roughness (LER) impairs the ability to form robust, reliable features. The origin for some roughness and resolution
limitation when using chemically amplified resist can be attributed to the diffusion limits of the photoacid catalyst,4 but
limitations around these dimensions seem to appears even when used with other non-chemically amplified processes in
optical, E-beam, X-ray and EUV lithography. Some have speculated that these may be shot noise effects, others that these
are due to excitation dynamics within the polymers that are not yet well understood.5

Directed self-assembly (DSA) has recently emerged as a possible candidate for lithographic patterning for dimensions
smaller than 50 nm.6 In self-assembly, the formation of features of fine geometric dimensions actually occurs not through
external patterning, but through the spontaneous phase behavior found among polymers on the molecular level. Of
particular interest are diblock copolymers formed by chemically connecting normally incompatible species, such as
poly(styrene) (PS) and poly(methyl-methacrylate) (PMMA). By creating linked chains of these materials and controlling
the relative molecular weight, various structures can spontaneously form. This is illustrated in Figure 1.

Schellenberg, 6151-125, Feb. 23, 2006 1


a) b) c)

Figure 1: Morphology for PS-b-PMMA copolymers of dimension A (PS) and B (PMMA) if


a) A>>B (forming spheres), b) A>B (forming cylinders) and c) A ≈B (forming lamella).

Schellenberg, 6151-125, Feb. 23, 2006 2


The appeal of these structures, easily formed between 20 and 40 nm in size, is that the boundaries between the two
disparate polymers can be quite uniform, with the uniformity dictated not by the noise properties of a patterning process,
but by the relative molecular weight of the polymers. This is a quantity that can be very precisely controlled. There is
therefore potential for lithographic patterning with very low roughness using block co-polymers.7

One problem with these polymers is that, although local order and roughness can be quite good, small variations in the
polymer chain can lead to kinks that reset the self-assembly process. Long range patterns therefore appear somewhat
chaotic.8 Clearly, this is unacceptable for patterning large structures for integrated circuits.

However, the problem of long range order can be addressed by using a conventional patterning process to guide and direct
the spontaneous formation of the block co-polymer structures. This “directed self-assembly” (DSA) can take the form of
having the self-assembly occur in grooves or other geometrically confined regions or by chemically patterning a surface to
,
create local affinities to the various portions of the block co-polymers.9 10

Using a very creative process of creating a patterned hydrophilic polystyrene “brush” on a flat surface, regular periodic
structures that follow straight lines, curves, and even a range of angles, including 90° bends, have been demonstrated.10
This is possible by providing a blend of co-polymers with an excess of monomer. The monomer has been found to fill the
‘gaps’ that would normally be left if only the regular co-polymers were used.

DSA is clearly not yet ready for insertion in a manufacturing process. In the above mentioned example of forming 90°
bends, the assembly time was on the order of a week (150 hrs), not the few minutes required for most semiconductor
processing steps. However, these initial results have generated great interest, and investigations actively under way by
, , ,
several organizations.11 12 13 14 Polymer science is a well researched area, and so there is a good chance that rapid
improvement in both the materials and processing technology can occur.

One as yet unexamined aspect of the adoption of DSA is the consequence for the design of ICs, and the techniques to
incorporate the “design rules” inherent with DSA structures into the IC design flow. Although the work to date on DSA
has been extremely impressive, the forte of the technique is the creation of regular periodic patterns. However, it will take
more than regular patterns of gratings to fabricate the structures required in ICs in the 20-40 nm scale. The ability to design
IC layouts that anticipate the fabrication restrictions inherent in DSA will be critical to the adoption of the technology.

There is, however, a previous technology also highly suited for the creation of regular periodic gratings that has seen some
adoption into semiconductor production: phase-shifting mask (PSM) technology. Since the original invention of phase-
shifting, several layout processing algorithms have been demonstrated and some used for the production of ICs. In this
paper, we examine the possible applicability of these phase mask algorithms to the design problems of DSA.

2. MULTI-EXPOSURE AND PHASE-SHIFTING

Phase shifting lithography improves resolution and image contrast by


adding additional phase-shifts on a reticle, typically phase-shifts of 180°.
0° 180° 0° 180°
These phase shifts are typically formed by etching away the substrate
material in certain regions of the photomask, creating a phase shift
dependent on the relative refractive index of air and the quartz
substrate.15,16 The interference patterns formed on the wafer when the
images of phase-shifted regions overlap create dark fringes that have
enhanced contrast and large process windows.

The most straightforward phase mask structure is the alternating aperture


phase mask, in which every other aperture has been phase shifted. This is Figure 2: Alternating aperture phase mask,
illustrated in Figure 2. This has been shown to be in which alternating clear apertures
have a relative phase-shift of 180°.

Schellenberg, 6151-125, Feb. 23, 2006 3


quite successful at doubling the effective resolution of a stepper lens.15,16 Design problems arise however, when these
features must end.17 Unless a design comprises only isolated bright apertures (which can be alternately shifted), a phase
shifter will end in a region that should normally print as bright, but will be dark due to the phase transition.

There are several solutions to the phase termination problem. One is to create an even more complex layout comprising
partial shifters (60° & 120°, or ±90° ) that make transition zones instead of sharp 0° to 180° transitions.18 Although not
being as bright as a clear field, the reduction in intensity at these partial shifter transitions is not dark enough to be below
the photoresist exposure threshold, and the resist is still exposed in these regions.

These masks function, but are also very complex to design and fabricate. Furthermore, the additional resolution possible
using phase shifting is tempered by the need to create these buffer zones of partial shifters around certain regions,
compromising the improvements in pattern density that can be achieved. For that reason, they have not become a popular
solution to the phase mask problem.

Instead, techniques which break the exposure problem into two masks, with one phase shifted and the other a regular
binary mask have become adopted. This allows the phase mask to be used at the full density possible using a phase
shifting mask, while the other exposure takes care of the various problems that using a phase mask entails.

There are 3 distinct approaches to the phase mask double exposure. Each has certain advantages and disadvantages. We
will first describe these approaches briefly, illustrating the technique with the layout shown in Figure 3, and then
contemplate whether these are suitable for adoption to the problem of directed self-assembly.

Desired Poly Structure


Cross Section

Figure 3: Representative layout for poly (striped) with diffusion underneath, and desired cross section along the middle line.

Schellenberg, 6151-125, Feb. 23, 2006 4


2.1. Trim Mask PSM

The first approach to phase mask double exposure is sometimes called the Trim Mask approach. This was first proposed in
1990,19 and is illustrated in Figure 4. In this method, the first mask contains both opaque chrome structures and phase-
shifters. The shifters are generated in the original layout for every region that requires the resolution increase (e.g. the
gates for a polysilicon layer), and the additional unwanted shifter line edges that print are noted. Then, a second “trim
mask” is created. This mask is typically a dark field binary mask that has only apertures that expose these unwanted
additional regions. The resist on the wafer must be exposed to both patterns before processing.

The advantage to this mask is that the design is fairly straightforward. Once the shifter apertures are placed, creating an
additional data layer of polygons for the apertures in the trim mask becomes a matter of a few Boolean operations, and the
fabrication costs of the trim mask are relatively low. And although the fabrication costs of the phase mask may still be
high, especially considering that there is still some sensitivity to the overlay of the phase structures and the opaque
structures underneath, the layout problems are generally manageable, and the main additional resign restriction with the
technique is only that the minimum width for the phase shifter be large enough.

Although the design problems may be manageable, and the technique may be effective for double exposure phase shifting,
it cannot be easily applied to the problems of DSA. Nowhere in the layout is there a large region of only periodic patterns
amenable to the block co-polymers as described above.

180°
180°

0° 180° 0° 180°

a) b)

Figure 4: Bright field PSM double exposure mask layouts for the pattern in Fig. 3. a) Phase mask layout, and
b) Trim mask layout. The trim features remove the unwanted artifacts created by the phase shifters.

2.2. GRATEFUL and IDEAL

Dense-only lithography techniques, such as GRATEFUL20 or IDEAL21 have been proposed more recently. The approach
here is quite different. Here, the initial exposure is made using a mask that contains only a regular, high resolution
alternating phase mask structure. This produces an exposure that is sub-threshold in the IDEAL case, or would, if
processed, create an array of periodic lines on the wafer in the GRATEFUL.

These lines are then further patterned into circuit structures using a second exposure with a binary trim mask. Features
from the first exposure that should be retained correspond to opaque regions in the trim mask, while features that should be
removed correspond to the bright regions. The sum of the two exposures is selected such that the general patterns are
defined by the opaque trim mask features, while the critical features are defined by the phase-mask structures. This is
illustrated in Figure 5.

Clearly, the initial step of this process produces something nearly identical to the patterns that are naturally produced by
directed self assembly, especially in the GRATEFUL case. In that regard, this is perfectly amenable to substitution of the
initial step with a DSA processing step to produce the original high resolution grating.

Schellenberg, 6151-125, Feb. 23, 2006 5


180° 0° 180° 0° 180°

a) b)

180°

180°

180°

c) d)

e)

Figure 5: GRATEFUL PSM double exposure mask layouts for the pattern in Fig. 3. a) 1st Phase mask layout,
b) 1st Trim mask layout, c) 2nd phase mask layout, d) 2nd Trim mask layout, and e) combined double exposure result.

Schellenberg, 6151-125, Feb. 23, 2006 6


There are, however, severe design problems with these techniques. To use the technique as illustrated, circuits must have
critical features that are placed on the regular grid, not arbitrary locations, and in some implementations, all critical
features must lie in only one orientation. Currently, although some restricted design rules (RDR) approaches have has
some success,22 designers are not use to these constraints, and design software places gates and high resolution
interconnects without these restrictions. These restrictions carry additional costs to circuit performance that designers
currently are unwilling to accept.

GRATEFUL addresses these objections by allowing an additional, orthogonal phase mask exposure, to create lines in the
orthogonal orientation. This should in principle allow structures more compatible with existing design paradigms. Yet, in
practice, this is achieved through a second resist coating and a second etch process. Insertion of DSA into this process
sequence would therefore require a similar recoating of the wafer, and subsequent processing for creating the DSA
structures in the orthogonal orientation as well.

2.3. Dark Field PSM

A third approach to phase shifting has also been introduced. This is also a double exposure technique, but parses the data
in a different manner than the previously outlined methods. Here, all critical features are segregated out into a phase
shifting mask layout, and phase-shifters are defined only for those features. The balance of this initial phase shifting mask
remains opaque. The second mask contains all the other structure information, as well as certain additional preserving
structures are created to shield the critical structures to be created by the first mask.23 This is illustrated in Figure 6.

When compared to other PSM processes, this contains some advantages. The area and boundaries of the phase regions can
be derived from, for example, the underlying diffusion layer data, making the creation and assignment of phase shifting
structures relatively simple. The size and scope of the phase regions are also limited, reducing the area of the phase mask
that would be prone to defects. Currently, repair of phase mask structures is difficult, so reducing the risk of phase defects
reduces the potential cost. Finally, the second mask can be made using standard maskmaking techniques, further keeping
costs to a minimum.

With this technique, as long as some simple rules are followed, circuits with arbitrary layout patterns can be fabricated in
which all the critical can achieve the benefits of the resolution enhancement of phase shifting. The main rule to be obeyed
is that the width of each of the phase shift regions be larger than a minimum dimension, so that enough light can pass
through the aperture on both sides to cause the desired interference fringe between them. This is usually dictated by the
Rayleigh resolution of the optical system, and is larger than the width of the chrome left between shifters and certainly
larger than the desired structure on the wafer.

λ
wφ ≥ 0.6 >> Lw
NA
This therefore puts an additional constraint on the pitch at which minimum CD structures can be placed. However, this
minimum spacing is rarely larger than the spacing that would be dictated by other design rules derived from fabrication
tolerances, and is not viewed as an undue burden by designers.

The relative simplicity of this method makes it the most attractive of the various implementations of strong phase shifting,
and this has been carried into production of advanced ICs, usually high value microprocessors.24 The layout conversion
algorithms are well known and established, and phase-ready cell libraries have been generated according to this
methodology.

Schellenberg, 6151-125, Feb. 23, 2006 7


180°

0° 180° 0° 180°


Figure 6: Dark field PSM double exposure mask layouts for the pattern in Fig. 3. a) Phase mask layout, with shifters generally
of width wφ or 2wφ for patterning only the gates; and b) Structure mask layout, which fills in the additional, larger
features and also protects the gates formed using the phase mask.

3. A “DARK FIELD” DSA PROCESS

As can be seen from the illustration in Figure 6, the phase structures themselves are placed in a relatively regular, periodic
array, roughly of period 2wφ, and therefore may be suited for fabrication with the periodically inclined methods of directed
self-assembly. We will now walk through the process steps that would be involved in creating a gate layer for a portion of
the layout shown in Figure 3. The process steps are illustrated in Figures 7a) through 7p). Our intention is to pattern the
polysilicon layer with gates in particular having a small dimension and edge roughness dictated by DSA, not noisy
lithographic processes, while roughness and dimensional control are less critical for the larger features.

Up until now, DSA has generally been carried out using evenly balanced copolymer, to make all dense lamellar structures.
The more recent arbitrary structures (90° bends, curves, acute angles, etc.) have been demonstrated by using a mixture of
copolymer and excess homopolymers.10 The homopolymers migrate to fill the gaps left in the assembled copolymer
formations. We will assume that a suitable mixture of copolymer and homopolymer can be created to fill space with this
layout pattern as well.

3.1. One possible sequence of process steps.

Step 1: Preparation (Fig. 7a)


In this step, the normal underlying diffusion structures are created, and then the polysilicon layer to be patterned is
deposited.

Step 2: Coating with a PS brush (Fig. 7b)


In this step, the poly is first coated with the polystyrene polymer “brush” – molecules generally aligned perpendicular to
the surface which, if untreated, have exposed hydrophobic ends. These would preferentially be wetted with the polystyrene
portion of the copolymer.

Step 3: Coating with resist (Fig. 7c)


In this step, photoresist is coated on top of the PS-brush. The resist must be chosen to be suitably compatible with the
properties of the PS-brush.

Schellenberg, 6151-125, Feb. 23, 2006 8


Step 4: Gate patterning exposure (Fig. 7d)
The resist is then patterned according to the layout defined by the phase mask data. The intention is to pattern only the fine
features at this point. Although originally intended to be used with a phase mask, this exposure can be done with phase
masks, electron-beam lithography, EUV lithography, or any other technique capable of creating the high resolution
structures.

Step 5: Resist Processing (Fig. 7e)


The resist is then developed. This exposes portions of the PS brush.

Step 6: PS-Brush Processing (Fig. 7f)


The exposed PS-brush coating is now processed to convert the material properties to now preferentially be wetted by the
PMMA portion of the polymer. This can be accomplished in an oxygen plasma, or through various chemical treatments.

Step 7: Clearing Exposure (Fig. 7g)


At this point, an additional exposure may be necessary to clear the remaining resist from the gate portions of the layout.

Step 8: Resist Processing (Fig. 7h)


The resist is then developed, leaving some portions still protected, but a large open region in which treated and non-treated
portions of the PS-brush are exposed.

Step 9: Copolymer Coating (Fig. 7i)


A polymer mixture comprising copolymer (for example, PS-b-PMMA) and suitable amounts of homopolymer (both PS
and PMMA) is coated onto the exposed PS brush surface,

Step 10: Directed Self-Assembly (DSA) (Fig. 7j)


The polymer mixture, in contact with the altered and unaltered portions of the PS brush, will now be annealed. Ideally, the
PS portions will align with the unexposed, hydrophobic portions while the PMMA will align with the exposed regions.
Currently, this anneal step can be very time consuming, taking times up to 150 hrs (almost a week). However, this is
known by the community of chemists to be impossibly long for any commercial process, and mechanisms to speed up the
self-assembly process are under investigation. The wafer now has a coating with resist, PMMA, and PS.

Step 11: Polysilicon Etch (Fig. 7k)


The etch resistance of PMMA is considerably less than that of PS,25 so with a suitable etch treatment, the resist and PS can
act as barriers to etching, while the PMMA erodes and allows the underlying materials to be etched. An alternative
approach may be to chemically process the surface, to selectively dissolve the PMMA, exposing the polysilicon surface.
Clearly, chemists can come up with a variety of techniques to allow selective patterning of the underlying material. After
etch, the remaining polymer will be removed.

Step 12: Resist Coating (Fig. 7l)


To complete the patterning, the surface will be cleaned and again coated with a photoresist.

Step 13: Structure Layout Exposure (Fig. 7m)


The resist is now exposed using the patterning data from the structure mask of Fig. 6. This defines the larger, non-gate
structures, and also leaves resist to protect the already patterned gate structures.

Step 14: Resist Processing (Fig. 7n)


The resist is now developed.

Step 15: Polysilicon Etch (Fig. 7o)


As above, the surface is etched. The exposed polysilicon is etched away, while the protected portions remain.

Step 16: Resist Strip (Fig. 7p)


Finally, the resist is stripped. The polysilicon, now twice etched, has the desired structure as shown in Fig. 3.

Schellenberg, 6151-125, Feb. 23, 2006 9


a) Diffusion and Deposition of Polysilicon

PS Brush
Polysilicon

b) Coat with PS-Brush


Photoresist

PS Brush
Polysilicon

c) Coat with Photoresist

180°

0° 180° 0° 180°

d) Expose Resist using PSM layout pattern

e) Develop Resist

Figure 7 (a-e): Proposed initial process steps for creating a polysilicon layout pattern for a directed self-assembly (DSA)
process using layout data derived for a dark field phase-shifting mask.
Note: Layer thicknesses in illustrations not to scale

Schellenberg, 6151-125, Feb. 23, 2006 10


f) Treat PS Brush (Exposed Areas Only)

180°

g) Clearing Exposure (PSM Layout Areas Only)

h) Develop Resist
Lw
wφ wφ
PS-b-PMMA
Mixture = + PMMA
Homopolymer PS

i) Fill Cleared Areas with Copolymer/Homopolymer Mixture

j) Anneal for Directed Self-Assembly (DSA)

Figure 7 (f-j): Proposed intermediate process steps for a DSA process using PSM layout data.

Schellenberg, 6151-125, Feb. 23, 2006 11


k) Etch Polysilicon

l) Resist Coating

m) Expose Resist using Structure Layout Pattern

n) Develop Resist

o) Etch Polysilicon

p) Remove Resist

Figure 7 (k-p): Proposed final process steps a DSA process using PSM layout data. The final polysilicon pattern in p) matches the
desired structure in Figure 3.

Schellenberg, 6151-125, Feb. 23, 2006 12


After all steps are completed, the hope is that the roughness and structure of the final etched poly will reflect the improved
roughness inherent in the copolymer structures. Since some recent studies on etch transfer from resist to etched poly show
that the etched structures have even less roughness than the resist,26 the smooth walls of the resist formed by DSA are
expected to be an excellent candidate for forming very smooth, uniform structures in polysilicon.

3.2. Other Variations.

The process outlined above is not the most efficient for generating fine structures. Three exposure steps, not two, are
required, and there are several coating and cleaning steps that may or may not be required. However, if the
copolymer/homopolymer blend can be suitably prepared such that a single exposure can suitably treat the PS-Brush to
allow copolymer (with just enough homopolymer to fill the excess) to migrate to the gate regions, while the homopolymer
alone migrates to the larger regions, it may be possible to achieve self-assembly with just a single exposure of the PS-
brush. Clearly, other variations may be possible to those who are more familiar with the subtleties of polymer processing

4. CONCLUSIONS

In this presentation, we have considered whether data parsed for phase mask lithography might be suitable patterns to
direct self-assembly processes. The answer, in a limited way, appears to be “Yes”. We have examined 3 possible data
parsing algorithms for phase shifting – bright field, GRATEFUL, and dark field – and investigated their suitability for
DSA. The final summary is illustrated in Table I.

Although Bright Field design solutions exist, there is no generally periodic structure suitable for DSA, and this approach
therefore is unsuitable. Dense only solutions, such as GRATEFUL, present the opposite situation. The periodic structures
of the regular phase mask are perfectly suited for DSA, but these come at a price of great restriction on the designs that can
be created, requiring a high degree of periodicity that is currently unacceptable to the normal design paradigm.

Dark Field PSM layouts, however, may combine the best of Suitable Design
both worlds. There are clearly quasi-periodic structures that for DSA Solutions
can take advantage of the periodic tendencies of the Bright Field PSM ☺
copolymers, especially if a suitable amount of homopolymer
can be used to fill the gaps at corners and other junctions.
GRATEFUL ☺
However, design recipes exist for these ICs, and a large Dark Field PSM ☺
library of patterns is currently used in production of
advanced microprocessors. The limitations on the design Table I: Relative suitability of PSM approaches for
rules are considered light, and even the most critical directing self assembly.
restriction (that the phase-shifting region be wide enough)
can be relaxed if patterning is carried out using, for example, E-beam lithography.

The achievement of these results, however, will depend on chemists being able to derive a suitable blend of molecules to
allow the self-assembly to occur in the manner described, and in a time frame suitable for semiconductor processing.
It does appear, however, based on what is already known about DSA processing, that an off-the-shelf design solution
(dark field PSM) can provide what is needed to enable this once the chemists have addressed the materials problems.

5. ACKNOWLEDGEMENTS

We greatly appreciate helpful discussions with Mark Stoykovich, Prof. Paul Nealey and Prof. Juan de Pablo of the
University of Wisconsin for helpful discussions and Dan Herr of the SRC for his encouragement of directed self-assembly
research. I would also like to thank Mike Fritze and Brian Tyrrell for introducing me to GRATEFUL, and Marc Levenson
for discussions on phase-shifting in general and for getting me into lithography in the first place.

Schellenberg, 6151-125, Feb. 23, 2006 13


Visit our website at www.mentor.com
Copyright ©2006 Made available by Mentor Graphics Corporation. Mentor products and processes are registered trademarks of Mentor Graphics Corporation.
All other trademarks mentioned in this document are trademarks of their respective owners.

Corporate Headquarters Silicon Valley Europe Headquarters Pacific Rim Japan


Mentor Graphics Corporation Mentor Graphics Corporation Mentor Graphics Corporation Mentor Graphics (Taiwan) Mentor Graphics Japan Co., Ltd.
8005 SW Boeckman Road 1001 Ridder Park Drive Arnulfstrasse 201 Room 1603, 16F Gotenyama Hills
Wilsonville, OR 97070-7777 San Jose, California 95131 USA 80634 Munchen International Trade Building 7-35, Kita-Shinagawa 4-chome
Phone: 503.685.7000 Phone: 408.436.1500 Germany No. 333, Section 1, Keelung Road Shinagawa-Ku, Tokyo 140
Fax: 503.685.1204 Sales and Fax: 408.436.1501 North American Phone: 49 (0) 89 57096-0 Taipei, Taiwan, ROC Japan 03/2006
Product Information Support Center Fax: 49 (0) 89 57096-400 Phone: 886.2.87252000 Phone: 81.3.5488.3033
Phone: 800.547.3000 Phone: 800.547.4303 Fax: 886.2.27576027 Fax: 81.3.5488.3021

Schellenberg, 6151-125, Feb. 23, 2006 14


REFERENCES
1
Burn J. Lin, “Immersion lithography and its impact on semiconductor manufacturing”, J. Microlithography, Microfabrication, &
Microsystems, 3, pp. 377-500 (2004).
2
Alfred K.K. Wong, Resolution Enhancement Techniques in Optical Lithography (SPIE Press, Bellingham, WA, 2001)
3
F.M. Schellenberg, ed Selected Papers on Resolution Enhancement Techniques in Optical Lithography. (SPIE Press, Bellingham,
WA, 2004).
4
G. Gallatin, “Resist blur and line edge roughness”, Proc. SPIE 5754, pp. 38-52 (2005).
5
Ming L. Yu, Allan Sagle, Benny Buller “Exploring the fundamental limit of CD control: a model for shot noise in lithography” Proc
SPIE 5751, pp.. 687-698, (2005).
6
D. Herr “The Extensibility Of Optical Patterning Via Directed Self-Assembly Of Nano-Engineered Imaging Materials” Future Fab
Intl. 18 (1/12/2005) http://www.future-fab.com/documents.asp?grID=213&d_ID=3022
7
D. Herr, “Update on the Extensibility of Optical Patterning via Directed Self-Assembly” (1/7/2006) Future Fab Intl. 20
http://www.future-fab.com/documents.asp?d_ID=3717
8
S.O. Kim, H. Solak, M. Stoykovich, N. Ferrier, J. de Pablo, & P. Nealey, “Epitaxial Self-Assembly of Block Copolymers on
Lithographically Defined Nanopatterned Substrates”, Nature 424, pp. 411-414 (24 July 2003).
9
E. Edwards, M. Stoykovich, M. Müller, H. Solak, J. de Pablo, & P. Nealey, “Mechanism and Kinetics of Ordering in Diblock
Copolymer Thin Films on Chemically Nanopatterned Substrates” J. Polymer Sci. B: Polymer Physics 43, pp. 3444-3459 (2005).
10
M. Stoykovich, M. Müller, S.O. Kim, H. Solak, E. Edwards, J. de Pablo, & P. Nealey, “Directed Assembly of Block Copolymer
Blends into Nonregular Device-Oriented Structures” Science 308, pp. 1442-1446 (3 June 2005).
11
C.T. Black & O. Bezencenet, “Nanometer-scale pattern registration and alignment by directed diblock copolymer self-assembly”,
IEEE Trans. Nanotechnology 3, pp. 412-415 (2004)
12
C.T. Black, “Self-aligned Self Assembly of Multi-nanowire Silicon Field Effect Transistors” Appl. Phys. Lett. 87, 163116 (2005)
13
L. Chang, P. Wong, “Diblock copolymer directed self assembly for CMOS device fabrication”, Proc. SPIE 6156, paper 6156-66
(2006) in press.
14
J. Cheng, C.A. Ross, A. Mayes, “Nanostructure Engineering by Templated Self-Assembly”, Nature Materials 3 pp. 823-8 (2004).
15
Marc D. Levenson, N.S. Viswanathan, R.A. Simpson, “Improving resolution in photolithography with a phase-shifting mask,” IEEE
Trans. Electron Dev. ED-29, pp. 1828-1836 (1982).
16
Marc D. Levenson, Douglas S. Goodman, Scott Lindsey, Paul W. Bayer, Hugo A.E. Santini, “The phase-shifting mask II: Imaging
simulations and submicrometer resist exposures,” IEEE Trans. Electron Dev. ED-31, pp. 753-763 (1984).
17
L. Liebmann et. al., “Enabling alternating phase shifted mask designs for a full logic gate level”, J. Microlithography,
Microfabrication, & Microsystems, 1, pp. 31-42 (2002)
18
G. Galan, F. Lalanne, P. Schiavone and J.M. Temerson, “Application of Alternating Type Phase Shift mask to polysilicon Level for
Random logic Circuits”, Japan. J. Appl. Phys. 33, Pt. I, pp. 6779-6784 (1994).
19
H. Jinbo and Y. Yamashita, “Improvement of Phase-Shifter Edge Line Mask method”, Japan. J. Appl. Phys. 30, pp. 2998-3003
(1991).
20
B. Tyrrell, M. Fritze, D. Astolfi, R. Mallen, B. Wheeler, P. Rhyins, and P. Martin, “Investigation of the physical and practical limits
of dense-only phase shift lithography for circuit feature definition”, J. Microlithography, Microfabrication, & Microsystems, 1,
pp. 243-252 (2002).
21
T. Ebihara, T. Oga, P. Rhyins, P. Martin and M. Sweis, “150nm dense/isolated contact hole study with Canon IDEAL technique”,
Proc. SPIE 4562, pp. 1068-1076 (2001).
22
Lars Liebmann, Greg A. Northrop, James Culp, Leon Sigal, Arnold Barish, Carlos A. Fonseca, “Layout optimization at the pinnacle
of optical lithography”, Proc SPIE 5042, pp. 1-14 (2003).
23
Hua-Yu Liu, Linard Karklin, Yao-Ting Wang, Yagyensh C. Pati, “Application of alternating phase-shifting masks to 140-nm gate
patterning: II. Mask design and manufacturing tolerances”, Proc. SPIE 3334 pp. 2-14 (1998).
24
C. Spence, M. Plat, E. Sahouria, N. Cobb, and F. Schellenberg, “Integration of Optical proximity Correction Strategies in Strong
Phase Shifter Design for Poly-Gate Layer” Proc SPIE 3873, pp 277-287 (1999).
25
X. Sun, L. Zhuang, W. Zhang and S. Chou, “Multilayer resist methods for nanoimprint lithography on nonflat surfaces”,
J. Vac. Sci. Technol. B 16(6), pp. 3922-3925 (1998).
26
A.R. Pawloski, A. Acheta, S. Bell and H.J. Levinson, “The transfer of photoresist LER through etch”, Proc. SPIE 6153,
paper 6153-45 (in press) (2006).

Schellenberg, 6151-125, Feb. 23, 2006 15

Anda mungkin juga menyukai