Section IV

NGL Masks

Rizvi / Handbook of Photomask Manufacturing Technology DK2192_c009 Final Proof page 191 7.3.2005 6:19pm

© 2005 by Taylor & Francis Group.

9

NGL Masks: An Overview

Kurt R. Kimmel and Michael Lercel

CONTENTS

9.1 Introduction
9.2 Electron Projection Lithography
9.3 Extreme Ultraviolet Lithography
9.4 Ion Beam Projection Lithography
9.5 Proximity X-Ray Lithography
9.6 Low-Energy Electron Beam Proximity Lithography
9.7 Nanoimprint Lithography
9.8 Summary

9.1

Introduction

‘‘NGL’’ or ‘‘next generation lithography’’ entered the semiconductor industry vocabulary
decades ago, as a category to encompass the various emerging and contemplated lithog-
raphy techniques being developed as replacement for the long-standing optical projection
technology that defines microlithography today. The deployment of any of the NGL
technologies has been far later than most expert prognosticators in the field imagined.
The continued ability of optical lithography technologies to provide imaging needs at
justifiable cost has surprised many and crushed the hopes of several technology ventures.
A long-standing joke in the industry applies the parody marketing tag line, ‘‘The Tech-
nology of the Future’’ to any NGL technology a speaker wishes to bash. It reflects the
industry’s fascination, for decades with the perception of a perpetually forecasted end of
the optical lithography era and pursuit of a still unreached dawning of the NGL era.

Eventually, the universal laws of science do prevail, and, clearly, optical lithography

will have a finite limit of application when looking purely at the technology capability.
The pertinent industry question will be whether another business case emerges such that
advances in nonlithography-related areas (design, materials, processing methods, etc.)
will sufficiently promote further evolution of microelectronics and obviate the need to
deploy one or more of the NGL technologies, thus further extending the usefulness of
optical lithography. Or, perhaps the nonlithography-related microelectronics manufac-
turing factors will reach their own fundamental limits before lithography does, and
lithography will no longer be the limiting technology factor. Of course all these assess-
ments of where and when optical lithography will find its demise are made in the context
of a balanced technology (capability) versus business (financial) decision.

Rizvi / Handbook of Photomask Manufacturing Technology DK2192_c009 Final Proof page 193 7.3.2005 6:19pm

© 2005 by Taylor & Francis Group.

Regardless of the timing or business case, the generic challenges facing NGL technolo-

gies are staggering. Nevertheless, the next four chapters assume certain failure of optical
lithography and present the main NGL technologies in development today and recently.
‘‘NGLs’’ addressed in this book are electron beam projection lithography (EPL), extreme
ultraviolet lithography (EUV 13.5 nm wavelength), ion projection lithography (IPL),
and proximity x-ray lithography (PXL 1 nm wavelength). Low-energy electron beam
projection lithography (LEEPL) and imprint lithography (a.k.a. step-and-flash), are intro-
duced in this chapter but are not addressed later in this book.

While talking of NGLs it would also be appropriate to mention another important

lithography technology in development: direct patterning of the device substrate using
electrons or photons. This is collectively know as ‘‘mask-less lithography’’ (or ML2) and is
also not addressed in the book for obvious reasons.

The differences in the overall imaging methods for NGL versus optical technologies lie

in the radically different mask materials and architectures. The ever-shrinking imaging
requirements (at any demagnification) and the lethality of some gases or monolayers as
new classes of defects pose problems that are driving sophisticated — and expensive —
solutions. Also, a crucial mask accessory that is not feasible for any of the NGL technolo-
gies is a pellicle to protect the mask from defects postmanufacture. The full impact of this
seemingly small missing element has not been fully assessed as resources continue to be
focused on the more fundamental functional issues such as resolution, CD control
and image placement error. The NGL technologies also create new pattern transfer
challenges for resists due to new contrast, sensitivity, and proximity situations associated
with each of the mask structure/illumination pairs.

Table 9.1

summarizes some of the

primary differences in NGL technology and mask physical parameters. The pertinence
and details of each of these mask types and parameters are discussed in subsequent
chapters.

Although the NGL technologies employ significantly different types of sources for

exposing wafers, they share some basic fundamental traits. The exposure energies
are much higher than those used in optical lithography (approximately few eV) and
greater than the bonding energy of materials, so the radiation exposure energy cannot
pass through a solid material without some attenuation or change in beam profile.
Therefore, all the mask formats require a fundamentally different approach to transmis-
sion. This may include a thin membrane, a reflective surface, or the use of a stencil mask.
This introduces new challenges to the mask for flatness, film stress, mask distortion, and
uniformity. The higher energy exposure, however, puts an opposite demand on the
absorber material. The patterning layer must have sufficient absorption or scattering
(determined by its thickness and the exposure energy) but remain thin enough to avoid
high aspect ratios that are difficult to pattern into the mask. The absorption or scattering
of the radiation can also lead to a heat buildup in the mask that must be dissipated.

In addition, the higher exposure energies and lower beam numerical apertures pre-

clude the use of an organic defect protective layer (the pellicle) out of the focal plane. As
such, reticle defect protection schemes are a challenge for all the NGL technologies. New
approaches of keeping defects off of the reticle and ensuring that they do not return are
required.

Nanoimprint lithography is the one exception to the higher exposure energy concerns.

However, the same concerns about reticle protection apply, and the near-perfect image
transfer fidelity leads to near-perfect defect printing.

The following sections give brief introductions of six NGL technologies, the first four of

which are described in detail in the subsequent chapters of this book.

© 2005 by Taylor & Francis Group.

9.2

Electron Projection Lithography

Electron projection lithography (EPL) has a fundamental attraction of resolution derived
from the extremely short effective wavelength afforded by using electrons as an illumin-
ation source. Using charged species also provides the opportunity to do global scale
pattern distortion correction by deflecting the beam appropriately. This is a crucial
advantage for EPL considering that the delicate, tensile membrane substrate and perfor-
ated mask structure create a mask inherently prone to distortion. Recent work in Japan
and Europe has created a continuous membrane mask structure, where the transmissive
areas of the mask are extremely thin (20 nm) membranes, which help reduce distortions
imposed by a perforated structure and resolve the so-called ‘‘donut problem.’’ The donut
problem is present for all stencil technologies and refers to a pattern where an island of
opaque material exists. A completely disconnected island structure is physically impos-
sible to build in a stencil mask structure. The solution is to split the troublesome pattern
into two complementary patterns that additively create the desired island. The severe
penalty for this is that two masks are needed for every one layer that contains an island or
other similar feature.

EPL development is centered in Japan with Nikon leading the commercialization effort.

EPL mask technology benefits from the extensive membrane mask development work in
Japan and elsewhere to support PXL.

Primary issues for this technology are: in-plane pattern distortion control, mask inspec-

tion, field-to-field stitching, and defect protection postmask manufacture.

TABLE 9.1

Summary of Primary NGL Mask Attributes

EPL

EUV

IPL

PXL

LEEPL

Imprint

Wavelength

100 keV

Electrons

13.5 nm

75 keV Heþ

Ions

1 nm

2 keV

Electrons

Not

applicable

Type

Transmission

Reflective

Transmission

Transmission

Transmission

Mold

Substrate

Silicon

Quartz

Silicon

Silicon

Silicon

Quartz

Blank
Structure

Membrane
With struts

Reflector
Buffer

absorber

Membrane
Single field

Membrane
Single field

Membrane
With struts

Quartz
Substrate

only

Absorber

Material

Si membrane

TaN

Si membrane

TaSi alloy

Si absorber

Polymer

filler

Transmission

Material

Hole or

ultra-thin
SiN þ C
membrane

MoSi reflector

stack

Hole in

membrane

SiC or C

(diamond)
membrane

Hole in

membrane

Quartz

substrate

Magnification

4

4

4

1

1

1

Challenge 1

Distortion

Blank

defects

Distortion

Resolution

for 1

Resolution

for 1

Resolution

for 1

Challenge 2

Field stitching Flatness

Ion damage

Distortion

Distortion

Mask defects

Challenge 3

Defect

protection

Defect

protection

Defect

protection

Defect

protection

Defect

protection

Defect

protection

© 2005 by Taylor & Francis Group.

9.3

Extreme Ultraviolet Lithography

Extreme ultraviolet lithography (EUVL) was introduced originally as a 4 demagnifica-
tion alternative to 1 projection x-ray lithography and was referred to as ‘‘soft-x-ray
reflective lithography.’’ Later, it changed names to the less accurate but more easily
marketed ‘‘extreme ultraviolet’’ to distinguish itself from PXL, which had begun its fall
from favor due to the challenges imposed by the 1 mask architecture. EUVL has
ascended in just the last few years to be the primary NGL choice for most technologists
and International SEMATECH even established an annual EUVL lithography symposium
in 2001 to further collect and focus on the reports from the sizable resources working on
this technology.

Primary issues for this technology are: fabrication of the complex, multilayer reflective

mask blanks at reasonable cost, mask defect repair, defect protection postmask manufac-
ture, generation of sufficient photon power for reasonable scanner throughput, and
contamination of the optical path including the mask.

9.4

Ion Beam Projection Lithography

Ion beam projection lithography (IPL) was proposed as an alternative charged-particle
technology affording pattern distortion control capability but imposing additional mask
challenges. The ions, having appreciable mass and energy, physically erode the mask over
time. This was addressed by applying a carbon buffer layer on top of the mask to absorb
the ions, act as a discharge layer, and provide additional thermal dissipation capacity.
Although the mask is 4 magnification, it is a perforated stencil type which is, therefore,
prone to in-plane distortion and to the donut problem described in the EPL section earlier.
Overall, this technology did not accumulate the resources and support necessary to bring
it to a commercial state, and the technology now continues to develop as a backup
technology or a possible alternative for meeting special needs.

Primary issues for this technology are: in-plane pattern distortion control for the

membrane structure, erosion from the impinging ions, and defect protection postmask
manufacture.

9.5

Proximity X-Ray Lithography

PXL has one of the longer histories in NGL and reached a peak of development resource
and interest in the mid-1990s being aggressively pursued in both the U.S. and Japan.
Although multiple chip fabrication demonstrations were achieved, the simultaneous
challenges imposed by the mask being 1 and the competitions from ever-advancing
optical extension techniques made x-ray become unjustifiable for the mainstream micro-
electronics industry. Originally, membranes were silicon and the absorber was gold
electroplated into a resist mold patterned on the membrane. This structure evolved
over time to become silicon carbide then diamond membrane with a tantalum–silicide
absorber, patterned with a more traditional subtractive etch process. Inspection is done by

© 2005 by Taylor & Francis Group.

electron beam and repair by focused ion beam. The technology is currently available
commercially from JMAR-SAL using a collimated plasma point source of x-rays and a
vertically oriented mask exposure stage. Masks are not commercially produced but are
available in limited quantities from multiple sources worldwide.

Primary issues for this technology are: resolution requirements for the 1 mask fabri-

cation, in-plane pattern distortion balance and control for the absorber þ membrane, and
defect protection post mask manufacture.

9.6

Low-Energy Electron Beam Proximity Lithography

Low-energy electron beam proximity lithography (LEEPL) is a variant of EPL, which
seeks to take advantage of the much simpler electron optic column design required for a
low energy, 1 magnification projection system. This makes the exposure system much
less costly but imposes some new challenges on the mask. Foremost, the mask at
1 becomes much more difficult to fabricate than the mask at 4 utilized in EPL. Because
the electrons have so little energy, the mask must be a perforated structure to provide
sufficient contrast and, consequently, the continuous membrane structure being devel-
oped for EPL is not feasible. However, the lower energy electrons are easier to deflect so
that pattern distortion correction is more easily achieved. LEEPL development is centered
in Japan and is a natural adjunct to the EPL technology work there.

Primary issues for this technology are: resolution requirements for the 1 mask fabri-

cation, in-plane pattern distortion control for the perforated mask structure, field-to-field
stitching, and defect protection postmask manufacture.

9.7

Nanoimprint Lithography

Nanoimprint lithography involves a direct physical transfer of a pattern from a template
to a transfer material on the wafer surface. For this reason, nanoimprint is entirely
different from the radiative exposure methods. The resolution and pattern fidelity are
determined almost entirely by the mask (or template). The few-nanometer scale reso-
lution is a tremendous advantage but also a nuisance in that any defects or imperfections
on the mask are faithfully transferred to the wafer surface. The mask therefore is a 1
version of the pattern and requires strict control of image size, pattern distortion, and
defects.

Primary issues for this technology are: patterning of the 1 mask, defect control and

repair on the mask, and level-to-level overlay.

9.8

Summary

Clearly, no single NGL technology has overwhelming appeal and each has its own
distinct advantages and challenges. At a fundamental level, the technical requirements
must be balanced against the business constraints of cost to develop, implement, and

© 2005 by Taylor & Francis Group.

maintain a technology. Since lithography has become such a significant component of the
overall microelectronics manufacturing cost structure, fabricators will likely choose im-
aging solutions that are optimized for their business cost sensitivities.

For example, a fundamental difference in business cases is mask utilization, that is, the

number of wafers or chips exposed per mask over the product lifetime. High utilization
fabrications, such as microprocessor and memory, will have stronger justification for
technologies that are more mask cost intensive but offer high resolution and technology
extendability. EUVL may fit the needs best in that case for critical layers while for
foundries, having relatively smaller quantity, shorter-lived products may find that EPL
or even ML2 on appropriate levels to offer the best value.

In any case, users will optimize the cost of their imaging needs commensurate with

their technical needs, and this may mean that more than one imaging solution will find
commercial implementation. The risks are high and very difficult to assess for all cases.
The evidence of this is reflected in the wide array of NGL technologies being pursued and
the substantial investments behind each. Within any particular imaging technology,
generally the mask cost is the primary driver of the cost of ownership. This explains
why the majority of lithography infrastructure development resources are actually direc-
ted to the mask sub-component of the technology.

While the cost of developing a single NGL technology is staggering, developing two or

more to completion may be impossible. A primary goal of microelectronics fabrication
managers and their lithography strategists will be to avoid having to make this dicey
choice of an NGL for as long as optical lithography can viably serve the industry needs.

Rizvi / Handbook of Photomask Manufacturing Technology DK2192_c009 Final Proof page 198 7.3.2005 6:19pm

© 2005 by Taylor & Francis Group.