5&5: sch :
_ , PVRUH‘R.
3.1. A a
a, . . t #w
'1: t!
'1 .
fikvl 2;»...
.thnfiflis ,. I. ”dun: .
unfiuflalwuu‘ a «1 .
g _
w" .50t.‘ .
ch.)

.8.
5.4::

‘-
5f

.

I
13
t.

. . .-
.25..un2 Jug.“ .. . i. .
.rtciinnaicr .1. ‘ i .. . .Nhoduu‘ I.
. inks? [Fannin‘zvkfit .1 4 IS in}.
1. 1. . 1.. {Lie m. “3..
2..-, innuexmzuéi; ‘ .
‘vfil—l gqli . . u ,. g .
i}- fit{ ,11. u ‘ Art! .1
‘ v. ‘ ans: 1.3.6 .. . , ucx...
. b... . 2. 1 . “firm. .
1. L323.” \ 12.3”.) it, .
, .x.‘ ‘ E!

£21132

.( ‘1‘: 5.4..
«Jilin-Izc .
, Vt

.. 1
3mm?
.

‘. . .vdffi... If?!
:wazzééx s
... .

5.55

. t s 3 .3:
=1. 1.5%.}? 4.0.105... .fir
s 9:... ex a

Q .1 y .
l :8. (I... .
If.uuu.\\§lnz
.I 2:51.311}. ..
gt f‘awzlitfils?!si
«It: . .
l‘éf‘tl‘?
I‘ll‘v‘ll11
I“!!! U. .
43.nuu-i tall? .ffi.’
lurid «ha. .
Lilia}? .r. .
; .16....331!
. “Worn. . 2.2!. 3.1:}
all...) I..AI!....!8..IXJS
.f...s.....? §.!ufl;§. .
kinda“) 5.:
53!

litre...
XIII)- .3,
. 1.5215!!!er JO .
.‘1!S<..-69y\:.utkll..\llltffl
‘ f .Y :v 3. 2:49:91‘ a...)
{strffun {III}: «fill;
I. ‘6‘»: ‘1..’i HI ‘ f I
i i a . .

Ill".
Ibi'bol.

,SV..‘AI

L 9ft?»
151:1»: 1.5.714
l I. ll" r. .59

lovIII .4! \[I’nl
..:.,.n.1.-,.t....: k 49:1. 4 ‘ Judi.

 

8888 Illilllil'illlililflllilllllliiilfiiifillL

3 1293 02058 6396

This is to certify that the

dissertation entitled

IN-SITU AND ON-LINE CHARAC'PERIZATION OF POLYMER SYSTEMS:
CURE MONITORING OF POLYMER COMPOSITES AND
SPECTROSCOPIC STUDIES OF CHEMICALLY AMPLIFIED PHOTORESISTS

presented by

Julie Lynne Pepper Jessop

has been accepted towards fulfillment
of the requirements for

Ph .D . degree in Chemical Engineering

 

 

 

Major professor

Date IZIIq/gq

MSU i: an Aflirmative Action/Equal Opportunity Institution 0-12771

,4 d-—

 

 

LIBRARY

Michigan State

University

 

 

PLACE IN RETURN BOX to remove this checkout from your record.
TO AVOID FINES return on or before date due.
MAY BE RECALLED with earlier due date if requested.

 

DATE DUE

DATE DUE

DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

11/00 chIRCJDabprfi-p.“

IN-SITU AND ON-LINE CHARACTERIZATION OF POLYMER SYSTEMS:
CURE MONITORING OF POLYMER COMPOSITES AND
SPECTROSCOPIC STUDIES OF CHEMICALLY AMPLIFIED PHOTORESISTS
By

Julie Lynne Pepper Jessop

A DISSERTATION
Submitted to Michigan State University
in partial fulfillment of the requirements

for the degree of
DOCTOR OF PHILOSOPHY

Department of Chemical Engineering

1 999

ABSTRACT
IN-SITUAND ON-LINE CHARACTERIZATION OF POLYMER SYSTEMS:
CURE MONITORING OF POLYMER COMPOSITES AND
SPECTROSCOPIC STUDIES OF CHEMICALLY AMPLIFIED PHOTORESISTS
By

Julie Lynne Pepper Jessop

Spectroscopic techniques such as fluorescence and absorption are useful for in—
situ and on-line characterization of a variety of polymer systems. Two examples are
discussed in this dissertation: monitoring of cure during polymerization of composite
parts and characterization of acid mobility in chemically amplified photoresists.

In the field of radiation curing, there is a compelling need for inexpensive on-Iine
sensors for the degree of cure. On-line monitoring techniques, which could be interfaced
with an electronic control scheme, would significantly enhance the reliability of polymer
and composites processing methods while reducing the cost. Fluorescence cure
monitoring strategies based upon trace quantities of solvatochromic probes, such as
pyrene and phenoxazone 660, do not lose sensitivity at the gel point and can be
implemented using optical fibers. Therefore, this method is well suited to in-situ and on-
line monitoring of cure in polymerization systems.

State-of-the-art microlithographic processes used to make features smaller than
0.25 microns are based upon deep-UV lithography and chemically amplified resists. In
these resists, photoacid generated during exposure initiates cascading deprotection
reactions during post exposure bake (PEB) to form a developable image. Reaction may

not be limited to the illuminated areas since the photo-generated protons may diffuse

out
of I
phc
phc
of :
sen
dur
IIICI
redI
mul
tech
Sim

alIet

 

 

outside this region; therefore, it is important to understand the diffusional characteristics
of the photoacid. In order to obtain this information, molecular probes were added to the
photoresist, and measurements were carried out in situ on quartz substrates. The
photoacid generation during exposure was characterized using acid-sensitive fluorescence
of fluorescein, and the acid concentration was quantified during PEB using the acid-
sensitive absorption of crystal violet. To measure the microscopic free-volume changes
during PEB, ground-state recovery measurements of crystal violet, whose excited-state
lifetime increases as the local free volume decreases, were used. The macroscopic
reduction of free volume due to the relaxation of polymer chains was measured using
multi-wavelength interferometry. The further application of these experimental
techniques will provide a systematic study of free-volume evolution and acid mobility in
situ during PEB. For a given resist, a more thorough understanding of the factors that

affect these parameters will allow improved microlithographic pattern generation.

For God, my husband and my parents

iv

di
lis

Iv

ICS

Al:

[08

nick
Pad-r
com
the (

me

 

 

ACKNOWLEDGMENTS

I have many people to thank for their support and assistance in completing this
dissertation, both the research and the manuscript. This is definitely not an all-inclusive
list, but I could not print this document without acknowledging a few names. First of all,
I would like to thank my graduate committee: Drs. Kris Berglund, Gary Blanchard, John
Klier, and Robert Ofoli. I appreciate that you have always been available to me as
resources of guidance and advice. I would especially like to thank my major advisor, Dr.
Alec Scranton, who modeled for me the qualities I wish to possess as a professor (not
only in research, but also in teaching) and who provided me with so many opportunities
to expand my graduate education. I could not ask for a better mentor!

Secondly, I want to thank the members of our research group, affectionately
nicknamed ABSLab: Kiran Baikerikar, Bernhard Drescher, Khanh Nguyen, and Kathryn
Padon. I have enjoyed our group lunches, group meetings (complete with snacks) and
conversations in the office. I value our camaraderie and friendship, and you made life in
the office a lot more interesting and fun! Furthermore, I am grateful to Katy for infecting
me with the quilting bug, which gave me a creative outlet during times of stress. In
addition, I would like to mention Saleha Mohamedulla, my undergraduate research
assistant, who helped me so much the semester I taught by capably working through the
tedium of the Filmetrics experiments. I also want to acknowledge Scott Goldie, my

colleague in chemistry, who was a tremendous driving force in the crystal violet

cxp<

\Nhe

Fou

prox
knox
couh
llafl
rnust
lioxv
and]

great

fahh
CVer
in I

hon,

 

 

experiments. Your enthusiasm and energy were much appreciated, especially on days
when the samples and equipment refused to cooperate!

Next, I would like to acknowledge gratefully support from the National Science
Foundation in the form of a graduate fellowship. I am also thankful for the Amelia
Earhart Fellowship from Zonta International, without which I would still be typing away
on my 386 without Windows.

Of course, I am extremely thankful for the love, encouragement and prayers
provided by my family throughout my college education. It was important for me to
know that they were behind me every step of the way and that they were proud of me. I
could not ask for a better cheering section! Most of all, I am so thankful for my husband
Mark, who suffered through these many years of my education. I know sometimes that it
must have seemed like an eternity to him, especially when it was his turn to cook.
However, he shared fully and willingly in each of my joys and sorrows during this time,
and I could not imagine having accomplished this without him by my side. We make a
great team, and I am looking forward to what the fiIture holds for us together!

Finally, I would like to thank my Lord and Savior, Jesus Christ. His love and
faithfulness have been an anchor in my life, and I have begun to understand that “I can do
everything through him who gives me strength” (Philippians 4:13). I echo Paul’s words
in I Timothy 1:17—“Now to the King eternal, immortal, invisible, the only God, be

honor and glory for ever and ever. Amen.”

vi

 

 

LIST 0

LIST 0

 

.w 3"" c.) '

 

TABLE OF CONTENTS

 

 

 

 

LIST OF TABLES - - x
LIST OF FIGURES xi
1. BACKGROUND AND MOTIVATION l
1.1. Introduction ............................................................................................................ 1
1.2. Spectroscopic background ..................................................................................... 2
1.3. Cure monitoring of polymer composites ............................................................... 3
1.3.1. Advantages of fluorescence ........................................................................ 3
1.3.2. Advantages of optical fibers ........................................................................ 4
1.3.3. Fluorescence-based cure monitoring techniques ......................................... 4
1.4. Characterization of chemically amplified photoresists .......................................... 8
1.4.1. Introduction ................................................................................................. 8
1.4.1.1. Semiconductor manufacturing ........................................................ 9
1.4.1.2. Photoresist systems ....................................................................... 10
1.4.1.3. Advantages of chemically amplified resists .................................. 11
1.4.1.4. Issues for the advancement of chemically amplified resists ......... 13
1.4.1.5. Motivation for research on chemically amplified resists .............. 13
1.4.2. Background ............................................................................................... 14
1.4.2.1. General properties of chemically amplified resists ...................... 14
1.4.2.2. Microlithography for processing of microelectronic devices ....... 15
1.4.2.3. Current models for chemically amplified resists ........................... 17
1.4.2.4. Utility of molecular probes ........................................................... 19
1.5. References ............................................................................................................ 24
2. OBJECTIVES 27
2.1. Cure monitoring of polymer composites ............................................................. 27
2.2. Characterization of chemically amplified photoresists ........................................ 28

3. [MS] T U CURE MONITORING FOR COMPOSITES PROCESSING

 

USING FLUORESCENCE SENSORS 30
3.1. Introduction .......................................................................................................... 30
3.2. Experimental ........................................................................................................ 32
3.2.1. Materials .................................................................................................... 32
3.2.2. Measurements ............................................................................................ 33
3.2.2.1. Fluorescence measurements .......................................................... 33
3.2.2.2. Thermocouple measurements ........................................................ 33

3.3. Results and discussion ......................................................................................... 34
3.3.1. Fluorescent probe molecules ..................................................................... 34
3.3.2. Fluorescence measurements ...................................................................... 35
3.3.3. Thermocouple measurements .................................................................... 37

vii

3.3.4. Sensitivity and interpretation of fluorescence response ............................ 37

3.3.4.1. Effect of temperature on fluorescence response of probes ........... 37

3.3.4.2. Comparison of probe responses during polymerization ................ 38
3.3.4.3. Effect of polarity on fluorescence response of phenoxazone

660 ..................................................................................................... 39

3.3.4.4. Effect of oxygen on fluorescence response of pyrene ...................... 40
3.3.4.5. Efi'ect of temperature and initiator concentration on intensity

ratio of phenoxazone 660 .................................................................. 40

3.3.4.6. Effect of phenoxazone 660 concentration on intensity ratio ......... 41

3.3.5. Equipment design ...................................................................................... 42

3.4. Conclusions .......................................................................................................... 44

3.5. References ............................................................................................................ 6O

. IN-SIT U MONITORING OF ACID CONCENTRATION IN

 

CHEMICALLY AMPLIFIED PHOTORESISTS 61
4.1. Introduction .......................................................................................................... 61
4.1.1. Utility of molecular probes ....................................................................... 62
4.1.2. Studies of acid concentration in CARs ..................................................... 63
4.2. Experimental ........................................................................................................ 63
4.2.1. Materials .................................................................................................... 63
4.2.2. Methods ..................................................................................................... 64
4.2.2.1. Absorption and Fluorescence Measurements ................................ 64
4.2.2.2. Steady-state fluorescence measurements with photoresist
films during exposure .................................................................... 64
4.2.2.3. Steady-state absorption measurements with photoresist films
during processing .......................................................................... 65
4.3. Results and discussion ......................................................................................... 65
4.3.1. Probe molecule selection ........................................................................... 65
4.3.2. Acid-sensitivity of fluorescein .................................................................. 67
4.3.3. Study of photoacid generation during exposure in photoresist films
using steady-state fluorescence ................................................................. 67
4.3.4. Acid-sensitivity of crystal violet ............................................................... 69
4.3.5. Study of photoacid concentration during processing in photoresist
films using steady-state absorption ........................................................... 70
4.3.6. Effect of probe presence on photoacid concentration during
processing in photoresist films using steady-state absorption .................. 72
4.4. Conclusion ........................................................................................................... 74
4.5. References ............................................................................................................ 88
. FREE VOLUME IN CHEMICALLY AMPLIFIED PHOTORESISTS ............. 90
5.1. Introduction .......................................................................................................... 90
5.1.1. Utility of molecular probes ....................................................................... 91
5.1.2. Studies of free volume evolution in CARs ............................................... 92
5.2. Expenmental92
5.2.1. Materials .................................................................................................... 92

viii

A. AP]
Al
A2.
A.3.

 

 

5.2.2. Methods ..................................................................................................... 92
5.2.2.1. Ground-state recovery measurements with photoresist films

 

during processing .......................................................................... 92
5.2.2.2. Multi-wavelength interferometry .................................................. 94
5.3. Results and discussion ......................................................................................... 95
5.3.1. Study of microscopic free volume changes in photoresist films using
ground-state recovery ................................................................................ 95
5.3.2. Multi-wavelength interferometry .............................................................. 97
5.4. Conclusions ........................................................................................................ 100
5.5. References .......................................................................................................... 111
. CONCLUSIONS AND RECOMMENDATIONS 113
6.1. Summary of results ............................................................................................ 113
6.1.1. Cure monitoring of polymer composites ................................................. 113
6.1.2. Characterization of chemically amplified photoresists ........................... 114
6.2. Suggestions for future work ............................................................................... 115
6.2.1. Cure monitoring of polymer composites ................................................. 115
6.2.2. Characterization of chemically amplified photoresists ........................... 118
6.2.2.1. 248-nm resists ............................................................................. 118
6.2.2.2. 193-nm resists ............................................................................. 120
6.3. References .......................................................................................................... 123
. APPENDIX
A. 1 . Determination of equilibrium constants ............................................................ 125
A2. Determination of alpha plots ............................................................................. 126
A3 References .......................................................................................................... 13 1

ix

Table

Table.

Table I

Table 4

 

Table 3.1.

Table 3.2.

Table 3.3.

Table 4.1.

LIST OF TABLES

Chapter 3

Comparison of solvent polarity and phenoxazone 660 emission
maximum ................................................................................................... 39

Effects of temperature and initiator concentration upon the time to
polymerization and final intensity ratio plateau for phenoxazone 660
in styrene with AIBN ................................................................................ 41

Dependence of fluorescence peak emission wavelength and final
cure monitoring ratio of phenoxazone 660 in Derakane® 470-45 with
0.75 wt.% AIBN upon probe concentration .............................................. 42

Chapter 4

Comparison of the absorbance ratio at 278 nm for the neat Apex-E
2408 DUV photoresist film to that of the crystal violet doped films
during various stages of processing .......................................................... 74

Fig

F igt

Figu
Figu

Figul
Figu:
Figur

Figur

Figu r1
Figu re

Fl'gure

Figure

Figure ;

Figure 1.1.

Figure 1.2.

Figure 3.1.
Figure 3.2.

Figure 3.3.

Figure 3.4.

Figure 3.5.

Figure 3.6.

Figure 3.7.

Figure 3.8.

Figure 3.9.

Figure 3.10.

Figure 3.11.

LIST OF FIGURES

Chapter 1

Jablonski diagram for a molecule showing the pathways for
activation and deactivation: absorption (a), vibrational relaxation (b),
fluorescence (c), external conversion (d), intersystem crossing (e),

and phosphorescence (t) ............................................................................ 22
Deprotection of PTBOC in DUV photoresist ........................................... 23
Chapter 3
General chemical structure of the Derakane” vinyl ester resins ............... 45
Chemical structures of phenoxazone 660 (a) and pyrene (b) .................... 46
Spectral shift of phenoxazone 660 associated with curing Derakane"
411-C50 (with AIBN as initiator) ................................................................. 47
Blue-shift in phenoxazone 660 peak emission as the cure of
Derakane‘” 411-C50 (with AIBN as initiator) progresses ......................... 48
Change in phenoxazone 660 fluorescence intensity ratio in curing
styrene with AIBN as initiator .................................................................. 49
Fluorescence response of pyrene in MMA with AIBN as initiator ........... 50

Change in pyrene fluorescence intensity ratio in curing MMA with
AIBN as initiator ....................................................................................... 51

Change in pyrene fluorescence rolling time ratio at 378 nm in curing
MMA with AIBN as initiator .................................................................... 52

Comparison between thermocouple (...) and phenoxazone 660
intensity ratio (I) data in curing Derakane® 411-C50 with AIBN as
initiator ...................................................................................................... 53

Change in pyrene fluorescence intensity ratio in tri(ethylene glycol)
divinyl ether with increasing (U) and decreasing (0) temperature .......... 54

Change in pyrene fluorescence intensity ratio (—) and phenoxazone
660 fluorescence intensity ratio (...) in curing Derakane® 470-45
with AIBN as initiator ............................................................................... 55

xi

 

_____._-————~.‘.___.-

 

 

 

Fi

Fig

Fig

Figu
Figui

Figur

Figurt

Figure

Figure
FlgUre.
I”guru
Figu” 4
Figure 4.

Figure 4.

 

Figure 3.12.

Figure 3.13.

Figure 3.14.
Figure 3.15.

Figure 4.1.
Figure 4.2.

Figure 4.3.

Figure 4.4.

Figure 4.5.

Figure 4.6.
Figure 4.7.
Figure 4.8.
Figure 4.9.
Figure 4.10.

Figure 4.11.

Change in pyrene fluorescence intensity ratio in curing MMA with
V-70 as initiator: without oxygen .............................................................. 56

Change in pyrene fluorescence intensity (top) and rolling time ratio
(bottom) at 378 nm in curing MMA with V-70 as initiator: without
oxygen (El) and with oxygen (0) .............................................................. 57

Diagram of in-situ monitoring instrument ................................................ 58

Change in intensity ratio of phenoxazone 660 emission in curing
Derakane® 41 1-C50 with AIBN as initiator monitored by in-situ

apparatus .................................................................................................... 59
Chapter 4

Experimental setup for in-situ fluorescence experiments to determine
photoacid generation during exposure ...................................................... 76

Absorption (abs), excitation (ex) and emission (em) characteristics
of Apex-E 2408 DUV photoresist ............................................................. 77

Molecular structure of fluorescein ............................................................ 78

Absorption and emission characteristics of fluorescein in Apex-E
2408 DUV photoresist ............................................................................... 78

Change in fluorescein fluorescence emission in Apex-E 2408 DUV
photoresist upon addition of sulfuric acid: before acid addition (—)
and after acid addition (. . .) ....................................................................... 79

Fluorescence spectra of fluorescein in Apex-E 2408 DUV
photoresist film as a function of increased exposure time ........................ 80

Intensity ratio at 525 nm and 560 nm of fluorescein in Apex-E 2408

DUV photoresist film as a function of increased exposure time ............... 81
Protonation/deprotonation equilibria for crystal violet. The
corresponding spectra are shown in Figure 4.1 1 ....................................... 82

Absorption and emission characteristics of crystal violet in Apex-E
2408 DUV photoresist ............................................................................... 83

Absorption and emission characteristics of crystal violet in bulk
acidic (top) and basic (bottom) solutions .................................................. 84

Absorption spectra of crystal violet in aqueous solution. [l-F] for the
solution spectra are indicated on the plot .................................................. 85

xii

 

 

Fi

Fl;

Fig
Fig

Fig

F igt

F igu

Figu

Figur

FlgUrt

I‘Nigure

Figure

Figurefi

Figure 6.

Figure 4.12.

Figure 4.13.

Figure 5.1.

Figure 5.2.

Figure 5.3.

Figure 5.4.

Figure 5.5.

Figure 5.6.

Figure 5.7.

Figure 5.8.

Figure 5.9.

Figure 5.10.

Figure 6.1.

Figure 6.2.

Absorption spectra of crystal violet in Apex-E 2408 DUV

photoresist film during and afier processing ............................................. 86

Effect of 0.3 wt.% crystal violet in Apex-E 2408 DUV photoresist

film upon the absorbance spectrum before and after processing .............. 87
Chapter 5

Experimental setup for ground state recovery experiments to
determine microscopic fiee volume in the photoresist films .................. 102

Experimental setup for dynamic film thickness measurements to
observe macroscopic free volume evolution ........................................... 103

Molecular structure of crystal violet ....................................................... 104

Schematic representation of the transient change in ground state
absorption of crystal violet in response to excitation by a pump laser
pulse ........................................................................................................ 104

Instrumental response function (....) and experimental ground state
recovery data (—) for crystal violet in the Apex-E 2408 DUV
photoresist film ........................................................................................ 105

Linear calibration curve relating the short time constant. n, from the
ground state recovery experiments of viscous liquids doped with 0.1
wt.% crystal violet to the matrix viscosity .............................................. 106

Instrumental response function (....) and experimental ground state
recovery data (—) for crystal violet in glycerol ...................................... 107

Change in Apex-E 2408 DUV photoresist film thickness as a
function of bake time at 100°C for unexposed (top) and exposed
(bottom) films .......................................................................................... 108

Change in thickness of Apex-E 2408 DUV photoresist film as a
function of exposure dose and post exposure bake time ......................... 109

Change in thickness of UV5-0.6 positive DUV photoresist film as a
function of exposure dose and post exposure bake time ......................... 110

Chapter 6

Differential scanning calorimetry plots for polymerization of
Derakane® 411-C50 at 70°C: heat of reaction (top) and conversion
(bottom) ................................................................................................... 121

Chemical structure of the latest 193-nm resist formulation .................... 122

xiii

 

 

 

Figun

F igun

F igu I’l

 

Figure A.l.

Figure A.2.

Figure A.3.

Appendix

Change in absorption spectrum of crystal violet with increasing acid
concentration in buffered solution series ................................................ 128

Extrapolation of extinction coefiicients for CV (violet species, top)
and HzCV3+ (yellow species, bottom) based upon reported literature

values ....................................................................................................... 129
Change in relative amounts of CV+ (a2), HCV2+ (at) and Hsz3+
(do) with decreasing acid concentration .................................................. 130

xiv

 

 

 

 

 

 

 

 

 

 

 

often li
situ an
process
comple
in-situ ;
be dism
pans ar
amplifit
probe r
Pertiner
moniIOr
Sensitiv
method
InfOrma

the “11d

 

Chapter 1

BACKGROUND AND MOTIVATION

1.1. INTRODUCTION

The ability to make improvements and advancements in chemical processes is
often limited by the unavailability of in-situ information on the state of the system. In-
situ and on-line information is ofien difficult to obtain, and many quality control or
process optimization efforts are based upon off-line measurements after the process is
complete. Spectroscopic techniques such as fluorescence and absorption are useful for
in-situ and on-line characterization of a variety of polymer systems. Two examples will
be discussed in this dissertation: monitoring of cure during polymerization of composite
parts and characterization of the deprotection reaction afier the exposure of chemically
amplified resists. In the first instance, by adding small amounts of a solvatochromic
probe molecule to a system, it is possible to use spectroscopic techniques to gather
pertinent processing information in a quick and timely manner. Fluorescence cure
monitoring strategies based upon trace quantities of solvatochromic probes do not lose
sensitivity at the gel point and can be implemented using optical fibers. Therefore, this
method is well suited to in-situ and on-line monitoring of cure in polymerization systems.
Information about the concentration and mobility of photo-generated acids is critical for
the understanding and optimization of deep-UV chemically amplified resists. The
addition of carefully selected probe molecules to the resist matrix enables the detection of

the photoacid and its diffusion during the post exposure bake cycle using lifetime,

 

absorP"
change

picture

and difi
scales v
diagram
molecu]
energy 5
This act
then dis
such as 1
the term
of 1041
Ph0ton

State. M
011 a tim.

the first

 

Which I
SCalc 0f

using a

 

absorption and fluorescence measurements. In addition, correlation of the free volume
changes within the photoresist with diffusion of the photoacid present a more complete
picture of the changes occurring during resist processing.
1.2. SPECTROSCOPIC BACKGROUND

Spectrosc0pic techniques are especially amenable to the investigation of reaction
and diffusion mechanisms due to the variety of methods available and to the short time
scales which allow resolution of these polymer processes. Figure 1.1 shows the Jablonski
diagram enumerating possible pathways of activation and deactivation for a molecule. A
molecule may be promoted fi'om a low energy state (i. e, the ground state So) to a higher
energy state (e. g., the first excited singlet state 8.) through the absorption of a photon.
This activation process typically occurs on the order of 10"5 seconds. The molecule may
then dissipate this energy by one of several means. There are non-radiative pathways,
such as rotational motions of the molecule or collisions with other molecules, denoted by
the term external conversion. These deactivation processes typically occur on the order
of 10'II seconds. There are also radiative pathways, which involve the emission of a
photon of light. If this process occurs between a singlet excited state and the ground
state, wherein both electronic states are spin-paired, it is termed fluorescence and occurs
on a time scale of 109 seconds. If this process occurs between a triplet excited state (e.g.,
the first excited triplet state T.), in which the spin is unpaired, and the ground state, in
which the spin is paired, it is termed phosphorescence and occurs on a much slower time
scale of 10‘4 to 104 seconds. Thus, by tracking the electronic state of an excited molecule

using absorption or fluorescence spectroscopy, whether steady-state or transient,

 

 

inform

photo-1

develol
and otl
lack of
control
deterrni
system

respons.

minimiz

1.3.1. .4

 

information concerning the status of a system may be extracted depending upon the
photo-responsive nature of the individual components.
1.3. CURE MONITORING OF POLYMER COMPOSITES

In the field of composites processing, it has been widely recognized that the
development of convenient, compact methods for on-line monitoring of degree of cure
and other process variables would make an important impact on polymer systems. The
lack of such techniques has led to off-line measurements forming the basis of quality
control efi‘orts. Unfortunately, time, materials and energy are wasted when it is
determined that the product is unacceptable via this method. An on-line monitoring
system would allow a feedback control system that could manipulate process variables in
response to changes in the system. In this way, an on-line monitoring method would
minimize defect-related losses.
1.3.1. Advantages of fluorescence

The use of fluorescence to create an in-situ or on-line monitoring system has
numerous advantages. First of all, excitation and emission lights can be easily routed to
and from the system using optical fibers. This is especially attractive in systems like
those for free radical polymerization which must be blanketed in a nitrogen atmosphere.
Secondly, the short intrinsic time scale of fluorescence (10'9 seconds) provides extremely
rapid response and makes it suitable to follow the changing variables in polymer systems.
Since numerous probe molecules are available with large absorptivities and fluorescence
quantum yields, it is easy to obtain measurements with high signal-to-noise ratios without
signal averaging. Finally, fluorescence monitoring can be applied to the on-line control

of continuous processes using feedback control systems.

 

1.3.2

detec
easily
sigma;

thed

the as
instrur

1.3.3.

in an e
and co
molecu
drawba.

v1'SCositj

forming
fonns e
gromd
and the

of the

1.3.2. Advantages of optical fibers

Optical fibers provide a versatile method of transmitting radiation, either to the
sample surface or its interior, retrieving the modulated light and transmitting it to the
detection system via total internal reflection. Because optical fibers are flexible, they are
easily incorporated in and around many system geometries. They also transmit light
signals over large distances, which means the sample need not be positioned right next to
the detection unit.1 In addition, commercial optical fibers designed for optical
communications are attractive for monitoring studies because the fibers themselves and
the associated fiber couplers, manipulation devices, optical components, and related
instrumentation are readily available.
1.3.3. F luorescence-based cure monitoring techniques

In the past several years, many research groups have devoted their time and energy
in an effort to use fluorescence and fluorescent probes in the cure monitoring of polymers
and composites. The majority of these methods are based on the sensitivity of probe
molecules to the change in viscosity due to the polymerization. However, the major
drawback to these methods is the loss of this sensitivity at the gel point where the
viscosity diverges. Several of these methods are described in the following paragraphs.

F.W. Wang et al. have used 1,3-bis-(l-pyrene) propane (BPP) as an excimer-
forrning probe in polymerizing methyl methacrylate2 and epoxy resin mixtures.3 BPP
forms excimers when an excited state molecule associates with another molecule in its
ground state. The probability of this excimer formation is related to molecular mobility
and the microviscosity of the environment around the probes. By taking an intensity ratio

of the BF? monomer fluorescence (used as the internal standard) with respect to BPP

excimer fluorescence, it is possible to monitor the crosslinking reaction of the epoxy (i.e..
as the viscosity of the system increases, so does the intensity ratio). 0. Valdes-Aguilera
et al. have similarly used the excimer-fanning property of pyrene to monitor the cure of
methyl methacrylate.4 By monitoring the ratio of excimer to monomer fluorescence, they
are able to determine the rate of polymerization at low conversions.

A second cure monitoring method developed by F .W. Wang et a1. uses l-(4-
dimethylaminophenyl).6-phenyl-1,3,5-hexatriene (DMA-DPI-I) as a viscosity-sensitive
probe in epoxy resin mixtures.3 DMA-DPI-I exhibits an increase in fluorescence intensity
as the viscosity of the local environment increases. By taking an intensity ratio of DMA-
DPH fluorescence with respect to 9,10-diphenylanthracene (used as the viscosity-
insensitive internal standard), the crosslinking reaction of the epoxy may be monitored by
observing the increasing intensity ratio.

F.W. Wang and ES. Wu also experimented briefly with using fluorescence
recovery after photobleaching (FRAP) to monitor the cure of epoxy resins.5 Using 1,1’-
dihexyl-3,3,3’,3’-tetramethylindocarbocyanine perchlorate [DiIC6(3)], the rate of
recovery of fluorescence intensity was measured to determine the diffusion coefficient.
This diffusion coefficient decreases as the viscosity increases, enabling the study of
crosslinking progression. However, this method is too slow to use practically in an
industrial process setting.

S.F. Scarlata and J.A. Ors have used derivatives of 4-aminonaphthalimide to
monitor the cure of bisphenol-A-epoxy-acrylated resin mixtures.6 The fluorescence
polarization of the non-reactive probe in the curing epoxy mixture is measured, since the

probe molecule depolarizes light as it rotates. Therefore, an increase in the polarization

 

due
incrt

impl

optro
obser
The I
have ;
increa
decay
fluore

metho-

such a
TWO f]
a Spect
reCentl)
Probe

SOIValoc

3130 ad

mOlCCU

hal'e

due to a decrease in the rotational freedom of the dye is observed as the viscosity
increases during the cure. This method is not particularly amenable to on-Iine
implementation either due to the expense of the monitoring equipment.

R.L. Levy and SD. Schwab have specially built and patented a fluorescence
optrode cure sensor to monitor the cure of epoxy resins.7 With this instrument, they
observed changes in the resin fluorescence and the wavelength of its maximum emission.
The fluorescence emanating from the resin has been attributed to resin impurities that
have yet to be isolated. However, it is postulated that the fluorescence of these molecules
increases with increasing viscosity because molecular motion leading to fast non-radiative
decay is inhibited by the crosslink structure of the epoxy. Because the exact origin of the
fluorescence response is not known and is thought to be intrinsic to the system, this
method is not a good candidate for widespread industrial use.

J.C. Song et al. have been working with intramolecular charge transfer molecules
such as dansyl amide and its derivatives to monitor the polymerization of acrylates."9
Two fluorescence peak intensities characteristic of the dye molecule used are ratioed, and
a spectral blue shift of the ratio is observed as the polymerization proceeds. Until just
recently, this shift has been attributed to increases in the microviscosity surrounding the
probe molecule as the polymer cures; however, D.C. Neckers has cited the
solvatochromic nature of these probes as the reason for spectral changes.'0 They have
also adapted a fiber optic attachment for in-situ monitoring of the polymerizations.

In an effort to circumvent the problem of sensitivity loss with viscosity-sensitive
molecular probes, other methods that use self-quenching probes or solvatochromic probes

have been developed. For example, K.E. Miller et al. have used ethyl (2-fluoreny1)

 

metha
polym
by the
poly'm
quencl
the mt
synthe

proces

diamir
epoxy
involve
cure.l3
fluores

during

allows

WOrks \

 

methacrylate (FLEMA) as a self-quenching comonomer/molecular probe in the radical

” The fluorescence due to fluorene is quenched

polymerization of methyl methacrylate.
by the C=C in acrylate monomers; however, after this bond is broken during the
polymerization for incorporation into the polymer backbone, the fluorescence ceases to be
quenched. Therefore, the emission intensity from the fluorene is directly proportional to
the monomer conversion. Again, this method is system specific, and it entails an extra
synthesis to create a polymerizable probe molecule, which may limit its use in industrial
processes.

HJ. Paik and N.H. Sung have exploited the intrinsic fluorescence of aromatic
diamine curing agents such as diaminodiphenyl sulfone (DDS) to monitor the cure of
epoxy resins.'2 Prior to this research, Sung et a1. had used an extrinsic method that
involved adding a reactive label and an internal standard to the reaction system to monitor
cure.l3 However, in this system, the progressive red shift of the amine hardener’s
fluorescence peak is simply monitored as it is changed from a primary to a tertiary amine
during the crosslinking reaction. They have also developed a fiber optic attachment that
allows them to perform in-situ measurements of the epoxy cure cycle. This method
works very well for systems containing the diamine curing agents; however, it does not
apply to the broader spectrum of polymer composites.

D.L. Woerdeman and RS. Pamas have used 4-(N,Nodimethylamino)-4’-
nitrostilbene (DMANS) as a polarity-sensitive probe to monitor the cure of an epoxy

'4 A fluorescent wavelength shift to the blue is observed as the resin cures due to

resin.
the sensitivity of the probe to the changing polarity of the reaction mixture. They have

also developed an evanescent wave sensor to monitor the cure in-situ using a standing

 

“'8

ex;

ma
Ma
imp
that
ShO\
need
syste

proce

1.4.1.

the inf

and on:

People

 

wave at the fiber/resin interface that propagates into the medium and decays
exponentially.

As can be seen, there has been an assortment of methods developed in an effort to
make use of fluorescent probe molecules to monitor the cure of polymers and composites.
Many groups have also realized the utility of in-situ techniques in commercial
implementation of the monitoring methods. Of all the methods discussed above, those
that do not rely upon some viscosity-dependent characteristic of the probe molecules
show the most promise in revealing the entire cure profile. However, there is a definite
need for the development of a method that can be used in a wide variety of polymer
systems and yet can be simply and cost-effectively incorporated in an industrial
processing scheme.

1.4. CHARACTERIZATION OF CHEMICALLY AMPLIFIED RESISTS
1.4.1. Introduction

Semiconductors provide the enabling technology for the Information Age. Today,
the information technology sector accounts for 11 percent of our gross domestic product
and one-fourth of our manufacturing output. This sector employs more than 4.2 million
people with wages nearly double those of the average manufacturing worker. The
productivity gains made possible by information technology are the most important
reason why America has the world's most competitive economy. The US semiconductor
manufacturers supply some 40 percent of the world's output of microchips, and US
chipmakers derive more than half of their sales from international markets.'5 The chip
industry reached $132 billion in annual sales in 1996 and is projected to exceed $232

billion worldwide in the year 2000."5

 

1.4.1..

cleanli
CDVII'OI-
with t}
These

damagt
deposit

insulati

photore
spin co;
placed)
Next, t
PhOIore
and har
exPosed

PhOIOm,

   
  
  

pallet-n
reaetio

the Ca.

1.4.1.1. Semiconductor manufacturing

Semiconductor manufacturing is a complex process marked by its attention to
cleanliness and purity. Great lengths are taken to provide dust- and lint-free
environments in which the microchips are constructed. The manufacture itself begins
with the creation of a single silicon crystal ingot that is sliced into thin silicon wafers.
These wafers are lapped to remove mechanical defects and etched to remove surface
damage. They are then. polished and cleaned to provide a smooth, planar surface for
deposition. Finally, a thin layer of silicon dioxide is formed over the wafer as an
insulation against electrical charge.'7

Once the silicon wafer has been thus prepared, it is coated with a liquid
photoresist formulation. Typically, the deposition of the resist is accomplished using a
spin coater which rotates the wafer (in the middle of which a puddle of the resist has been
placed) at approximately 3,000 rpm to produce an even film about one micrometer thick.
Next, the wafer undergoes a prebaking process which eliminates solvent from the
photoresist film, increases the adhesion between the film and the silicon dioxide layer,
and hardens the film to promote damage resistance. After the prebake, the resist is
exposed to ultraviolet (UV) light which penetrates through the quartz areas of a
photomask onto the photoresist but is blocked by the chromium areas, and a circuit
pattern is created. The photons impinging upon the photoresist initiate a chemical
reaction that results in the degradation (in the case of positive resists) or crosslinking (in
the case of negative resists) of the exposed photoresist. The photoresist is subsequently
developed by rinsing off the degraded or unreacted portions with base or acid to reveal

the desired pattern. The wafer with the remaining photoresist is then postbaked to

increase pinhole resistance and surface adhesion, as well as to eliminate any volatiles
remaining and to increase the chemical resistance of the resist. '8

When the desired pattern has been formed in the photoresist, a bufi'ered oxide etch
is used to remove the exposed areas of silicon dioxide. Metals are then deposited onto
the silicon surface to provide the conductive pathways of the microchip. The photoresist
is selectively removed from the substrate after the completion of the pattern using a
process called ashing where plasma is used to strip off the photoresist without damaging
the other layers on the chip. This process involving the photoresist is repeated numerous
times using various patterns to form the complex circuitry layers. Once the circuit pattern
is complete, it is tested with a “needle-like” probe to ensure the connections are operating
properly. Since there are many individual components created on one silicon wafer, it is
now sliced to separate each chip. Each functional chip is placed on a lead fiarne, and
ultra-thin wires are attached from the lead flame to the connection points of the circuit.
Finally, the device is encapsulated in a polymer to protect the circuit and its connections
from physical and chemical damage.l7
1.4. 1.2. Photoresist systems

One of the key elements in the formation of a desired image on the semiconductor
chip is the photoresist. Photoresist technology grew out of the printing industry where
patterns were first etched on copper in the 18005 using gelatin which had been sensitized
with a dichromate salt. Improvements were subsequently made in the process in the
19305 by replacing the natural colloids with synthetic polymers. By the 19505, the circuit
board industry was experiencing rapid growth, and the resist formulations of the time did

not offer the processing ease and accuracy needed for “advanced” circuit integration. At

10

this time, Kodak introduced a photoresist based on cinnarnic acid derivatives which
revolutionized circuit board processing and led to the formulation of more sophisticated
photoresists as used in the microelectronics industry today."

Although the photoresists contain the same elements (photosensitizer, resin, and
additives in a solvent), there are basically two types of photoresists used to etched the
circuit patterns in chips: negative and positive. In a negative resist, the area exposed to
the light undergoes crosslinking or cyclization to render it insoluble in the developing
solution. Thus, when the resist is washed with solvent, the exposed regions remain to
form the pattern while the unexposed region is removed by the developing solution.l9
One example is the polyvinyl cinnamate photoresist developed by Kodak that undergoes
extensive crosslinking when exposed to light. Another class is the isoprenoid resins that
form an insoluble cyclized rubber when illuminated.l8 In a positive resist, the irradiated
area undergoes degradation or solubilization. Here, when the resist is washed with
solvent. the exposed regions are removed to form the image while the rest remains. An
example of this is the degradation of poly(methyl methacrylate), specifically about the
acrylic portion of the molecule, upon light exposure.'9
I. 4. 1.3. Advantages of chemically amplified resists

Although these photoresist systems have allowed the microelectronics industry to
make great strides in computing technology, they now pose a barrier to state-of-the-art
semiconductor processes. In order for smaller feature sizes to be etched onto chips,
thereby allowing more circuitry to be placed onto one chip at a time, shorter wavelengths

of light must be used. However, the sensitizers used in the photoresist mixtures are

usually activated by visible or near-UV light,'8 not the deep-UV light necessary for

11

advancing to the next generation of microchips. Moreover, since one photon yields only
one photo-event in these systems, it is difficult to provide enough light to initiate the
reactions which will result in pattern formation because the photon count from standard
light sources is generally quite low in the deep-UV area of the spectrum.

To address these shortcomings of normal photoresists, a specialized group of
photoresists called chemically amplified resists (CARs) was developed in the early
19805.20 These positive-resist systems have risen to the forefront of semiconductor
manufacttuing and are most promising for the further advancement of semiconductor
technology. In a CAR system, one photon induces the dissociation of the photoinitiator
that generates an acid. This acid reacts with a functional group on a nearby polymer
chain, cleaving off the long pendant chain and regenerating the acid. Thus, one photon
initiates many deprotection events, and it is this multiplicative process by which the
system is characterized as amplified.2|

With this new photoresist technology, it is then possible to utilize the deep-UV
light to decrease the sizes of patterns etched into the photoresist. First of all, the
photoinitiators used in CAR systems undergo photolysis with deep-UV light (i.e.,
wavelengths on the order of 250 nm). Secondly, the reaction-diffusion process of the
CARs provides high sensitivity to the impinging deep-UV photons. Due to the
amplification nature of the resist, it only takes a small number of photons to begin a chain
of deprotection reactions that will propagate to form a pattern as the photoacid diffuses

throughout the medium.

12

1.4.1.4. Issues for the advancement of chemically amplified resim

Although chemically amplified resists offer many advantages to the
semiconductor manufacturing process, there are several issues that have yet to be
addressed in order to apply this technology more successfully. The first and foremost
issue deals with resolution and line-width control when etching patterns in the
photoresist.”22 Due to the regenerative nature of the photoacid, diffusion of a photoacid
molecule into an unexposed region during the post exposure bake may lead to a pattern in
the resist which is different than that which is on the mask by beginning a cascade of
deprotection reactions in an unintended area. Therefore, it is imperative to understand
and control the diffusion and mobility of the photoacid in the resist. However, this leads
into a second issue concerning CAR technology: the accepted models are geared toward
normal resists in which there is only one deprotection reaction for every photo-event. In
such resist systems, optical principles may be applied to develop a model that will predict
the appearance of the end product given a certain photomask. F urtherrnore, attempts to
develop new models specifically designed for the CAR systems have been unsuccessful
because the chosen parameters do not accurately predict the diffusion and reaction of the
photoacid. Therefore, there is a compelling need to fit realistic parameters to the CAR
systems so that the technology may be used with even shorter wavelengths.
1. 4. 1.5. Motivation for research on chemically amplified resists

The key to utilizing chemically amplified resists to their fullest potential lies in
tracing the lifetime and activity of the photoacid. Without a clear understanding of
photoacid diffusion and mobility during illumination and postbake of the photoresist, it

will be difficult to extend semiconductor processing technology into the next generation

13

using chemically amplified resists. Furthermore, current photoresist models must be
improved to encompass these concepts such that suitable parameters are obtained, thereby
allowing accurate prediction of the end result from a given pattern upon a CAR.

Once an understanding of the diffusion and mobility of the photoacid in
chemically amplified resists is obtained, then strides may be taken to improve
semiconductor technology. The direct repercussions will include smaller feature sizes
and increased circuit density. This would result in more compact and faster processing
units for microelectronics users everywhere.

1.4.2. Background
1.4.2.1. General properties of chemically amplified resists

Positive chemically amplified resists were first developed by Ito and coworkers to
meet the technological initiative to produce microelectronic devices with increased circuit
density.” Their system incorporates onium salts (e.g., diphenyliodonium
hexafluoroarsenate) which are widely used in the photocuring of coatings. These cationic
photoinitiators dissociate upon absorption of deep-UV light to yield a cation radical and a
Brbndstedt acid.23 The polymer used to form the basis of the photoresist is poly(p-t-
butoxycarbonyloxystyrene), also known as PTBOC, which consists of long chains with
bulky pendant groups. The acid formed during the photo-event cleaves off the pendant
chain, thereby deprotecting the polymer chain and regenerating the acid. Figure 1.2
shows an example of the deprotection reaction that takes place. A small number of
deprotection events occur during the illumination of the photoresist; however, the
majority of the deprotection reactions occur during poste xposure bake at elevated

temperatures. The deprotected chains are then rendered soluble in base and may be

14

 

 

 

washed away when the photoresist is developed, while the chains that are still protected
remain to form the pattern.
1.4.2.2. Microlithograph y for processing of microelectronic devices

The most critical semiconductor technology is the microlithographic process used
to define the features on an integrated circuit chip. There is a continual drive to produce
devices with smaller feature sizes, since smaller sizes allow for more transistors to be
packed onto a single chip while simultaneously increasing performance and reliability.
Reduced feature sizes are obtained by using higher energy or shorter wavelength light to
pattern the resist, and, at the leading edge, the industry is undergoing a transition from
365-nm to 248-nm technology. At these higher energies the photon flux is greatly
reduced, and in order to increase throughput, deep-UV resists are “chemically amplified”
where the effect of a photOrgenerated acid is multiplied by a chemical chain reaction.
The state-of-the-art lithographic processes used to make features smaller than 0.25
microns are based upon deep-UV lithography and chemically amplified resists. The
ability of CARs to produce nearly vertical sidewalls from a relatively low intensity light
source has greatly extended the working resolution of optical lithography. Deep-UV
chemically amplified resists rely on a photoinitiated reaction followed by a post exposure
bake during which diffusion and chemical reaction (deprotection) form a developable
image. In these resists, the photo-active compound reacts with light to produce acid,
which, during the post exposure bake (PEB) process, diffuses and reacts in a chain
reaction that degrades the polymer matrix (for a positive resist) and regenerates the acid.
As the polymer matrix degrades, the viscosity in the vicinity of this reaction site (i.e., the

local or microviscosity) is reduced. This reduced viscosity (or increased free volume)

15

results in an increase in the localized diffusion coefficient of the acid in the polymer. In
its turn, the larger diffusion coefficient results in a higher acid concentration and
accelerates the degradation of the exposed resist, thus forming a front. It is this non-
linear phenomenon that produces sharp sidewalls, even in the presence of severe standing
waves. While this interplay of diffusion and reaction forms sharp sidewalls, it also results
in a steady increase in line width of about 30 to 50 nm/min. For the typical 90 second
PEB, this would translate to a total increase in line width of 90 to 150 nm, a very
significant fraction of critical dimensions (CD5) at small feature sizes.

Processing factors which have a dramatic effect on line width are post apply bake
(PAB) temperature and time, exposure intensity, post exposure bake duration and
temperature, as well as developer conditions such as the concentration of developer,
temperature and time. Physically, the PAB temperature determines the viscosity of the
resist prior to the PEB. Increasing the duration or temperature of the PAB will reduce the
free volume and result in an increased viscosity of the resist. Higher PEB temperatures
increase the sensitivity of the resist but also increase the mobility of the acid, making the
process more susceptible to process variations. Both reaction and diffusion are very
sensitive to temperature and time, and even slight non-uniformity on hot plates or a delay
in reaching a cool plate (and therefore delay in quenching the reaction) can cause
significant changes in CD5. These non-linear interactions between exposure light
intensity, chemical reaction and a non-constant diffusion make optimizing and running a

stable process a challenge.

16

1. 4.2.3. Current models for chemically amplified resists

To date, there are no viable computer models for deepoUV chemically amplified
resists. Current approaches to identifying model parameters rely on one of two methods.
By far, the most common method is by performing a design of experiments (DOE) to
sample CD variation as a function of PAB and PEB conditions. A second, less common
method is FTIR spectroscopy that has been used to monitor acid formation during the
PEB process. While the DOE approach allows for efficient sampling of a limited phase
space, its results cannot be easily extended to regions outside this space. Therefore, this
approach limits the utility of the model parameters and does not allow for efficient and
reliable extrapolation. The other major problems with this approach arise from the fact
that the results obtained are very dependent on the stability and sensitivity of the process.
While the use of FTIR spectroscopy provides direct insight into the deprotection reaction,
it does not provide any information on physical changes (such as viscosity, free volume,
diffusivity, etc.). In addition, F TIR studies are complicated by the background noise from
the resist and cannot be used as a tool to study changes occurring during resist
development due to the very high absorbance of infrared by water.

Classical modeling approaches for non-chemically amplified resist are based on
the use of an empirical correlation between the energy absorbed by the resist volume and
the rate at which a resist dissolves into a given developer (dissolution rate curve). The
strong dependency upon experimental conditions makes this modeling approach ill-suited
for predictive studies where new process conditions need to be explored. The dissolution
rate curve approach has severe limitations in the modeling of CARs, where no direct

relation can be postulated between the exposure conditions and the final developed resist

17

profiles. In the case of CARs, the dissolution rate depends not only on the development
process but also on the coupled effect of photoacid diffusion and thermally controlled
catalytic reactions during PEB.

Current PEB models described in literature can be summarized in the following
two coupled partial differential equations that describe the kinetics of the species

involved in the chemical reactions and diffusion.24

32%]- opIMIPIqu “'1’

% = ’kAIHlm -VlD(M)V[H]] (1 1)

The first equation describes the reaction leading to the deprotection of the polymer sites.
[M] and [H] refer to the concentrations of protected sites and photoacid respectively. The
model parameter kDp is the rate constant for the sites’ deprotection, while p and q are the
reaction orders for each species. The second equation describes the reaction and
diffusion of photoacid (in time and space) during the PEB. [H] represents the photoacid
concentration available for deprotection, m is an effective reaction order, kn the photoacid
loss reaction rate and D the photoacid diffusion coefficient (not necessarily constant). It
is interesting to note that different reaction and diffusion mechanisms (and consequently
different values for model parameters) have been used to describe the kinetics of deep-
UV CARs. While some experimental work indicates that the diffusion coefficient varies
exponentially with the photoacid concentration,” other papers assume that the process is
characterized by a constant diffusion coefficient.2°'27

Based on the above discussion, it is clear that the applicability of these models is

currently limited by the experimental resist process characterization, which translates into

18

a lack of reliable parameter values. Typical experimental parameter extraction has been
based on two techniques, namely FTIR spectroscopy and CD metrology. More powerful
techniques that are capable of simultaneously monitoring the dynamics and evolution of
several species over wide process ranges are badly needed, and developing such
techniques is the focus of this research.
1.4.2. 4. Utility of molecular probes

Luminescence (fluorescence and phosphorescence) techniques offer a unique
method to study the kinetics of a number of reactions, especially diffusion-controlled
reactions. For example, luminescence quenching and viscosity-sensitive fluorescence
have been applied to various polymer systems to characterize initiation, polymerization
and free volume on short time scales. Such approaches provide a basis for extracting
similar parameters in chemically amplified resist systems.
1. 4. 2. 4. 1. Luminescence quenching

A very powerful and well-established technique for studying mobility and
transport in polymer systems is luminescence quenching. The process of quenching
results in a deactivation of the excited singlet or triplet state (S. or T.) by interaction with
another molecule called a quencher, resulting in a radiationless transfer to the ground
state. Dynamic quenching relies on a direct collision between an excited-state
luminescent molecule and the quenching molecule.l Therefore, reactions between the
quencher and the luminescent molecule manifest themselves by a reduction in
fluorescence and phosphorescence lifetimes. Since the rate of this type of reaction is
diffusion-controlled, this process is very useful in characterizing other diffusion-

controlled reactions in polymer systems. Phosphorescence quenching is typically more

19

useful in polymer systems where the diffusion times are relatively long. These techniques
have been used to study diffusion-controlled reactions occurring due to polymer-small
molecule interactions,28 as well as due to interactions between small molecules in
polymeric systems.29

Scranton and coworkers have investigated the rate of reaction between anthracene
and iodonium salt in methanol by observing the change in emission lifetime of a

30'3 ' The phosphorescence lifetimes of anthracene

fluorescent probe upon “quenching.
were quenched from 9.55x10'6 seconds in the absence of initiator to 3.85x10"I seconds in
the presence of 2.0 wt% initiator. These studies demonstrate the utility of this technique
for characterizing rapidly changing systems.

1. 4. 2. 4. 2. Viscosity-sensitive fluorescence

In addition to measuring the rate of diffusion-controlled reactions, fluorescence
can also be used to probe the local environment of the system. The addition of a small
amount of a viscosity-sensitive probe can be used to measure the microviscosity or free-
volume of the system. Many researchers have exploited this technique to gather such
information in polymer systems.

For example, Moorjani and coworkers demonstrated the utility of fluorescence
monitoring when dealing with diffusion-controlled reactions in media of different
viscosity.32 In their studies of a polymer system, they used the fluorescence of the
photosensitizer anthracene to determine how the reaction rate changes as the polymer
cures. As the viscosity increased, the rate of diffusion-controlled reaction dramatically
decreased. Victor and Torkelson have also used molecular probes to measure the free

3

volume available in a polymer structure.3 The molecules chosen for their studies

20

(stilbene, azobenzene, m-azotolutene, etc.) undergo photoisomerization and require
different free volumes in order to carry out this conformational change. By observing the
fluorescence change of the probe molecules, it was possible to quantify the free volume
distribution in the polymers. In addition, several methods developed to monitor cure in
polymer systems and based upon viscosity-sensitive fluorescence were discussed
previously in this chapter.”9 These methods also involve the addition of specially chosen

probe molecules into the systems.

21

 

 

 

 

 

‘
Ix

 

 

 

 

a—r Y“
>1“

1.
1

 

 

 

 

 

H
‘

 

l

‘I-bd--—-—-—--

<

So‘

 

 

 

 

Figure 1.1. Jablonski diagram for a molecule showing the pathways for activation and
deactivation: absorption (a), vibrational relaxation (b). fluorescence (c),
external conversion (d), intersystem crossing (e), and phosphorescence (f).

22

 

 

i r
tNWWtNVVWC—-—CMNVVV\NV\N‘ | |
.0 .0 ——» O O
0H 9 H‘ on a.
— .1” 9:0
(I) H30\
H3C-(l3—CH3 C02 + 1.!" + /C—CH2

Figure 1.2. Deprotection of PTBOC in DUV photoresist.

23

10.

ll.

12.

1.5. REFERENCES

. J.D. Ingle, Jr. and SR. Crouch, Spectrochemical Analysis, Prentice Hall, New Jersey,

1988.

F .W. Wang, R.E. Lowry and W.H. Grant, “Novel excimer fluorescence method for
monitoring polymerization: 1. Polymerization of methyl methacrylate,” Polymer,
25(5), 690 (1984).

F.W. Wang, R.E. Lowry and BM. Faconi, “Novel fluorescence method for cure
monitoring of epoxy resins,” Polymer, 27, 1529 (1986).

O. Valdes-Aguilera, C.P. Pathak and DC. Neckers, “Pyrene as a Fluorescent Probe
for Monitoring Polymerization Rates,” Macromolecules, 23, 689 (1990).

F.W. Wang and E.-S. Wu, “Cure monitoring of epoxy resins by fluorescence recovery
after photobleaching,” Polym. Commun., 28, 73 (1987).

SF. Scarlata and J.A. Ors, “Fluorescence polarization: A method to monitor polymer
cure,” Polym. Commun., 27, 41 (1986).

R.L. Levy and SD. Schwab, “Monitoring the Composite Curing Process With a
Fluorescence-Based Fiber-Optic Sensor,” Polym. Comp., 12(2), 96 (1991).

J.C. Song and DC Neckers, “A fluorescence probe method for characterization of
photopolymerized networks,” PMSE, 71, 71 (1994).

.l.C. Song, Z.J. Wang, R. Bao, and DC. Neckers. “Cure Montioring of UV Coatings
Using a Fiber-Optic Fluorescence Technique,” PMSE, 72, 591 (1995).

W.F. lager, A.A. Volkers and DC. Neckers, “Solvatochromic Fluorescent Probes for
Monitoring the Photopolymerization of Dimethacrylates,” Macromolecules, 28, 8153
(1995).

K.E. Miller, E.L. Burch, F.D. Lewis, .l.M. Torkelson. “In Situ Monitoring of Free-
Radical Polymerization by Fluorescence Quenching of Fluorene and Reactive
Derivatives,” J. Poly. Sci. : Part B. 32. 2625 (1994).

H.-J. Paik and N.-H. Sung, “Fiberoptic Intrinsic Fluorescence for In-Situ Cure
Monitoring of Amine Cured Epoxy and Composites,” Polymer Engineering and
Science, 34(12), 1025 (I994).

24

l3.

14.

15.

l6.

l7.

l8.

19.

20.

N.-H. Sung, W. Dang, H.-J. Pail, C.S.P. Sung, “In-Situ monitoring of epoxy cure by
fiber-optic molecular sensors,” 36'” International SAMPE Symposium, 36, 1461,
(1991).

D.L. Woerdeman, K.M. Flynn, J.P. Dunkers, R.S. Pamas, “The Use of Evanescent
Wave Fluorescence Spectroscopy for Control of the Liquid Molding Process,”
Journal of Reinforced Plastics and Composites, 15(9), 922 (1996).

W.J. Sanders, III, SIA Washington Conference, Washington, DC, March 11, 1997.

E.M. Kirschner, “Electronic Chemicals,” Chemical & Engineering News, 75(47), 2S
(1997)

F ullrnan Company, “The Semiconductor Manufacturing Process,” on the Worldwide
Web at http://www.fullman.com/semiconductors/, 1997.

W.S. DeForest, Photoresist: Materials and Processes, McGraw-Hill Book Company,
New York, 1975.

G. Baker, CEM 956: Polymer Chemistry, Michigan State University, Fall 1996.

G. Wallraff, J. Hutchinson, W. Hinsberg, F. Houie, P. Seidel, R. Johnson, and W.
Oldham, “Thermal and acid-catalyzed deprotection kinetics in candidate deep
ultraviolet materials,” J Vac. Sci. Technol. B, 12(6), 3857 (1994).

. H. Ito and CG. Willson, “Applications of Photoinitiators to the Design of Resists for

Semiconductor Manufacturing,” Polymers in Electronics, edited by T. Davidson,
American Chemical Society, Washington. DC, 1984.

.J. Nakamura, H. Ban, K. Deguchi, and A. Tanaka, “Effect of Acid Diffusion on

Resolution of a Chemically Amplified Resist in X-Ray Lithography,” Japanese

' Journal of Applied Physics, 30(10), 2619 (1991).

23.

24.

25.

26.

J .-P. Fouassier, Photoinitiation. Photopolymerization. and Photocuring:
Fundamentals and Applications. Hanser Publishers, New York, 1995.

M. Zuniga, G. Walraff and AR. Neureuther, “Reaction Diffusion Kinetics in Deep-
UV Positive Tone Resist Systems,” Proceedings of SPIE. 2438, 113 (1995).

M. Zuniga and AR. Neureuther, “Post Exposure Bake Characterization and
Parameter Extraction for Positive Deep-UV Resists through Broad-Area Exposure
Experiments,” Proceedings of SPIE, 2724, 110 (1996).

A.A. Krasnoperova, M. Khan, S. Rhyner, J.W. Taylor, Y. Zhu, and R. Cerrina,
“Modeling and simulations of a positive chemically amplified photoresist for x-ray
lithography,” J. Vac. Sci. T echnol. B, 12(6), 3900 (1994).

25

27.

28.

29.

30.

31.

32.

33.

L. Capodieci, A. Krasnoperova, F. Cenina, C. Lyons, C. Spence, and K. Early,
“Novel postexposure bake simulator: First results,” J. Vac. Sci. Technol. B, 13(6),
2963 (1995).

MS. Gerbert and J .M. Torkelson, “Synthesis, characterization and photophysical
properties of benzil-labelled polymers for studies of diffusion-limited interactions by
phospherence quenching,” Polymer, 31, 2402 (1990).

MS. Gerbert, D.H.-S. Yu, and J.M. Torkelson, “Diffusion-Limited Polymer-Small
Molecule Interactions: A Study of Phospherence Quenching,” Macromolecules, 25,
4160 (1992).

E.W. Nelson, T.P. Carter and AB. Scranton, “Kinetics of Cationic
Photopolymerizations of Divinyl Ethers Characterized Using In Situ Raman
Spectroscopy,” J. Poly. Sci, A: Polym. Chem., 34, 403 (1996).

E.W. Nelson, T.P. Carter and AB. Scranton, “Fluorescence Monitoring of Cationic
Photopolymerizations: Divinyl Ether Polymerizations Photosensitized by Anthracene
Derivatives,” Macromolecules, 27, 1013 (1994).

S.K. Moorjani, “Polymers From Renewable Resources: Cationic Photopolymerization
of Epoxidized Natural Oils,” Master’s Thesis, Michigan State University, 1996.

J.G. Victor and J.M. Torkelson, “On Measuring the Distribution of Local Free
Volume in Glassy Polymers by Photochromic and Fluorescence Techniques,”
Macromolecules, 20, 2241 (1987).

26

Chapter 2

OBJECTIVES

2.1. CURE MONITORING OF POLYMER COMPOSITES

On-line monitoring of polymer cure is crucial for the prevention of losses due to
defective samples. Fluorescence spectroscopy provides a clear advantage for the
development of a cure monitoring technique in that it allows investigation on the time
scale of the polymerization. However, judicious selection of molecular probes must be
made in order to maintain sensitivity throughout the entire reaction, especially after the
gel point. This, coupled with the necessity of simplicity and cost efficiency in order to
incorporate the technique into industrial processing situations, poses a definite challenge.

The overall objective for this research has been to demonstrate the utility of
solvatochromic fluorescence probes for an inexpensive, non-intrusive, portable means of
cure monitoring for polymer composites using optical fibers. The specific research
objectives by which this was accomplished are as follows: (i) to identify appropriate
solvatochromic probes for the cure monitoring of polymers, (ii) to demonstrate the
fluorescence cure monitoring technique with several model systems, and (iii) to
investigate the underlying physicochemical origins of probe spectral response.

These objectives were fulfilled through the careful selection of two molecular
probes, namely phenoxazone 660 and pyrene, whose fluorescence response is based on
the [polarity of their environments. This response was successfully demonstrated for three

major polymer classes (acrylate, styrene and vinyl ether systems) and studied to

27

 

dett

COIII

the

2.1

of

determine the efi'ects of oxygen, cure temperature, probe concentration, and initiator
concentration. Finally, an optical fiber spectrometer was proposed and tested to confirm
the utility of this in-situ technique.

2.2. CHARACTERIZATION OF CHEMICALLY AMPLIFIED PHOTORESISTS

It is clear from the discussion in the first chapter that a fundamental understanding
of the reaction-diffusion mechanism is needed to optimize and extend the range of
processes using deep-UV chemically amplified resists. Isolating and measuring each of
the kinetic parameters and activation energies for the various steps of the reaction
diffusion during the post exposure bake process will enable the development of an
accurate model with predictive capabilities. However, spectroscopic techniques, which
are powerful and fast, must first be designed to obtain various model parameters. This is
a daunting task to perform in situ because the photoresist films are less than one micron
thick, making it extremely difficult to obtain any optical signal from the sample.
Therefore, careful selection of molecular probes and strategies for signal collection and
analysis are paramount in order to acquire acid production, viscosity, diffusivity, and
reaction kinetics as a function of temperature for a variety photoresist processing
conditions (e. g., bake temperatures and exposure times) in the thin photoresist films.

The overall objective for this research has been to develop highly sensitive, time-
resolved spectroscopic techniques to investigate the reaction-diffusion phenomena during
chemically amplified photoresist processing with the aid of probe molecules. The
specific research objectives by which this was accomplished are as follows: (i) to

monitor acid generation during exposure, (ii) to quantify acid concentration throughout

28

post exposure bake, and (iii) to measure the evolution of free volume during post
exposure bake.

These objectives were fulfilled through the development of several spectroscopic
techniques especially tailored for the thin film photoresist system. First of all, steady-
state fluorescence monitoring using the probe molecule fluorescein was applied to obtain
information about photoacid generation during exposure to 248-nm light. Secondly,
steady-state absorption of a second probe molecule crystal violet was applied to measure
the proton concentration during post exposure bake. Finally, two techniques were used to
measure the evolution of free volume in the resist. The first is a microscopic technique
based on ground-state recovery measurements also using the probe molecule crystal
violet. The second is a macroscopic technique that measures the dynamic film thickness
using multi-wavelength interferometry. Each of these techniques was realized in situ

with the one-micron thick photoresist films.

29

Chapter 3

lN-SITU CURE MONITORING FOR COMPOSITES PROCESSING
USING FLUORESCENCE SENSORS

3.1. INTRODUCTION
In the field of radiation curing, there is a compelling need for the development of
inexpensive on-line sensors for the degree of cure. On-line monitoring of
photopolymerizations is extremely diflicult since these reactions are typically carried out
in high-speed, continuous processing formats. Due to the lack of appropriate on-line
sensors for degree of cure, most quality control efforts rely on ex-situ measurements made
after the polymerization is complete. Therefore, if a polymer has cured improperly, it is
too late to make process adjustments and the entire run must be discarded. A similar
situation exists for traditional polymer and composites processing methods. For example,
the manufacture of polymer matrix composites involves a series of complex chemical and
physical changes that must be adequately controlled to produce products with desirable
properties. Again, due to the lack of appropriate on-line sensors for properties such as
viscosity and degree of cure, most quality control efforts are based on off-line measure-

ments made after the product has completed its cure cycle.
Clearly, on-line monitoring techniques, which could be interfaced with an
electronic control scheme, would significantly enhance the reliability of polymer and
composites processing methods while reducing the cost. The need for such sensors has

been recognized widely, and the measurement of a variety of physical phenomena has

30

been explored, including dielectric constant and loss, acoustic and ultrasonic wave
propagation, and optical techniques.I To date, none of these approaches has resulted in
the development of an inexpensive on-line sensor.

Fluorescence techniques have been proposed for the development of on-line
viscosity sensors due to the fact that the fluorescence behavior of a variety of molecular
probes depends on the local viscosity. For example, many fluorescent probes exhibit
enhanced fluorescence intensity with increasing viscosity due to decreased non-radiative
energy transfer. Levy and Schwab2 and Wang et al.3 demonstrated that such an increase
in fluorescence intensity can be used to monitor the viscosity increase during cure. The
measurement of absolute fluorescence intensity, however, has some significant
limitations. For example, the method has limited utility for monitoring the degree of cure
above the gel point (limited dynamic range) and is subject to uncertainty due to variations
in background fluorescence. Stroeks et al.4 investigated the use of viscosity—sensitive
excimer formation to monitor degree of cure. In addition to the aforementioned
limitations of the previous technique, the investigators concluded that apparent intensity
changes in the excimer peak were actually dominated by changes in the intensity of the
overlapping monomer peak (which was increasing because of reduced non-radiative
transfer). Valdes-Aguilera et al.5 have also utilized an excimer-forming probe to monitor
cure at low conversions, using a ratio of the monomer to excimer fluorescence emission,
which can be controlled by adjusting the probe concentration. Finally, Scarlata and Ors‘5
have monitored an increase in fluorescence polarization that they attribute to a viscosity-

induced decrease in the rotational mobility of the probe molecule. However, due to its

31

complexity, this technique is unlikely to form the basis for an inexpensive on-line sensor
for polymer and composites processing due to its need to measure absolute quantities.

In this chapter, a novel fluorescence monitoring scheme utilizing a solvatochromic
probe that exhibits a shift in fluorescence wavelength with increasing cure is reported.
This research was motivated by the fact that the use of a solvatochromic probe avoids
many of the problems and pitfalls associated with viscosity-sensitive fluorescence. First,
the solvatochromic method is based upon the relative fluorescence intensity ratio of two
interdependent peaks, thereby avoiding the problems (such as broadband background
interference) associated with absolute fluorescence intensity measurements. In addition,
unlike the viscosity-sensitive fluorescence probes, the solvatochromic probe does not lose
sensitivity afier the gel point (where the viscosity has diverged). One monitoring method
based upon an oxazone probe and a second based upon a polycyclic aromatic
hydrocarbon probe will be presented, and the advantages and limitations of these cure-
monitoring schemes will be discussed.

3.2. EXPERIMENTAL
3.2.1. Materials

These studies were performed using commercial vinyl esters (DerakaneQ), methyl
methacrylate (MMA) and styrene. DerakanedD resins 411-C50 and 470-45, the general
structure of which is shown in Figure 3.1, were obtained from Dow Chemical Company
(Midland, MI), MMA and styrene were purchased from Aldrich Chemical Company
Aldrich (Milwaukee, WI) and Polysciences, Inc. (Warrington, PA), respectively. The
radical initiators, 2,2’-azobisisobutyronitrile (AIBN) and 2,2’-azobis(2,4edimethyI-4-

methoxy valeronitrile) (V -70), were purchased from Polysciences, Inc. and Wako Pure

32

Chemical Industries, Ltd. (Richmond, VA), respectively. The fluorescence probes,
phenoxazone 660 and pyrene, were purchased from Exciton and Aldrich Chemical
Company, respectively. All chemicals were used as received except the MMA, which
was vacuum distilled before use to remove hydroquinone inhibitors, and the styrene,
which was washed with Dehibit-IOO (Polysciences, Inc.) before use to remove the
inhibitor 4-tert-butylcatechol.
3.2.2. Methods .
3. 2. 2. 1. Fluorescence measurements

Dcraltane‘ID 411-C50 samples with 0.75 wt.% AIBN and 10“ wt.% phenoxazone
660 were cured in an Aminco-Bowman Series 2 Luminescence Spectrometer at 70°C.
The sample emission spectrum from 550 nm to 800 nm was recorded every five minutes
using an excitation wavelength of 540 nm. Likewise, MMA samples with 0.5 wt.%
AIBN and 10'3 wt.% pyrene were cured in the spectrometer at 70°C, recording the sample
emission spectrum from 350 nm to 600 nm every five minutes with, an excitation
frequency of 334 nm. Styrene samples with 0.78 wt.% AIBN and 10'3 wt.% phenoxazone
660 were cured in the spectrometer at 70°C , recording the sample emission at 570 nm and
590 am every ten seconds with excitation at 510 nm.
3.2.2.2. Thermocouple measurements

For some experiments, fluorescence and temperature measurements were taken
simultaneously in the AB2 spectrometer. A thermocouple was placed in the middle of the
cuvettes, away from the walls and out of the excitation and emission pathways. A second
thermocouple was placed in the recirculating bath reservoir as a control. Temperature

readings were taken every 1.8 seconds.

33

3.3. RESULTS AND DISCUSSION

3.3.1. Fluorescent probe molecules

The use of probe molecules to study chemical and/or physical phenomena hinges
on the sensitivity of the probe molecule to some changing property of its local
environment. Two such changes are the polarity of the local environment and its
structural rigidity. In addition, for the probe molecule to be useful it must also have a
large absorptivity and a moderately high fluorescence quantum yield, so that the
information about its local environment can be measured efficiently. Phenoxazone 660
and pyrene (Figure 3.2) fulfill these requirements for the specific systems under study.

The oxazones are solvatochromic;7 their emission spectra are sensitive to the
polarity of their local environment. The sensitivity of the phenoxazones to the polarity of
their local environment arises from the non-bonding orbital associated primarily with the
heterocyclic nitrogen. Interactions between this lone pair and its immediate surroundings
(several A) determine its energy with respect to the excited state of the molecule, and thus
its coupling to the chromophore. This tunable degree of “delocalization” is highly
selective for interactions with the ring-bound nitrogen. The fluorescence spectrum of
pyrene is also sensitive to solvent polarity. The mechanism by which this polarity-
dependence operates is now well understood and can be predicted from the energy-
dependence of the linear response of these molecules.8 Imbedding these dyes in matrices
undergoing polymerization allows the examination of both the polarity and structural

changes associated with the curing cycle.

34

 

 

 

33

3.3.2. Fluorescence measurements

Monitoring changes in a chemical system in real time and in the presence of
potential background interferences places limits on the ways in which the measurement
can be accomplished. Measuring an absolute quantity is practical only if the background
is known and preferably constant in time. For this reason, absolute emission intensity is
not likely to be a suitable quantity for long-term, on-line or process-control
measurements. The probe molecule fluorescence lifetime is an absolute quantity that can
be resolved fi'om the background since measurement of this quantity depends on time
registration introduced by the instrumentation (background interference will not be
synchronous). Such a measurement is, however, technologically involved9 and not likely
to be useful for practical chemical processing situations. A more practical approach for
probe molecule-based monitoring is to measure the position or shape of the probe
emission band as a function of environment. A particularly simple way to do this is to
monitor two different emission wavelengths, as shown Figure 3.3, and take the ratio of
the intensities at the two wavelengths.l0 Measuring the ratio rather than absolute
intensities eliminates many uncertainties associated with macroscopic inhomogeneities
and broadband background interferences such as scattered white light.

For Derakane® 411-C50, the fluorescence emission maximum of phenoxazone
660 blue-shifts from 617 nm to 597 nm during cure, as shown in Figure 3.4. In Figure
3.5, the time evolution of the intensity ratio 1597/1617 is illustrated for this system during
polymerization. The sample emission at 597 nm and 617 nm were recorded every five
seconds for this data set. It is easy to distinguish between the uncured system with

1597/1617 z 0.8 and the cured system with [597/Ion = 1.3. Only the first 60 minutes of the

35

 

 

 

 

tip

per

the

611‘.

st:

Ch.

the

experiment are shown in Figure 3.5; however, the cured system ratio remained constant
throughout the remaing 90 minues of the data rim. Experiments have also been
performed with up to 50-wt.°/o loading of chopped glass fibers, and the results show that
the intensity ratio method is useful in composite structures as well, although the absolute
emission intensities are diminished due to scattering by the fibers.

Similarly, the fluorescence emission maximum of phenoxazone 660 in the curing
styrene system blue-shifts from 590 nm (monomer) to 570 nm (cured polymer). In this
instance, the ratio 1570/1590 plateaus at a value of 1.30 when the system has attained
complete cure, up from an initial value of 0.98 before cure has begun.

MMA cure characteristics were also examined using pyrene as the probe molecule.
The fluorescence spectrum of pyrene does not shift with polarity, but exhibits a variation in
the peak intensity ratio for two features within the emission band (see Figure 3.6). These
two features, designated bands I (378 nm) and III (389 nm), change with the extent of cure
in the MMA system. The ratio of peak I intensity to peak 111 intensity, Itllm, is 0.93 prior to
cure and subsequent to cure becomes 1.12 (Figure 3.7). However, the most dramatic

indication of complete cure is found in taking a rolling time ratio:

1, (3.1)

ratio:—
t-l

where I. is the intensity of band I or III at time t and I... is the intensity of that band at the
pervious instance in time, as is shown in Figure 3.8 (the ratio is shown for 378 nm;
however, the plot is the same for 389 nm). One explanation for this absolute intensity
change is that the emission spectrum of pyrene is very sensitive to oxygen quenching and

the diffusion rate of oxygen in the polymer system is directly related to viscosity. A more

36

detailed investigation of this effect, which will be discussed in a subsequent section, has
shown that not all of the intensity change on matrix curing can be accounted for by oxygen
availability. Some of this intensity change is the result of cure-dependent changes in the
interactions between the pyrene molecule and its immediate surroundings.
3.3.3. Thermocouple measurements

To confirm that the phenoxazone probe molecules were indeed sensitive to the
degree of cure in the DerakaneO resin, temperature data were collected simultaneously
with the fluorescence data. In these experiments, the heat of polymerization leads to a
significant temperature rise since heat transfer from the 1 cm x 1 cm cuvette is inefficient.
Therefore, a large temperature rise in a short period of time indicates a high
polymerization rate. The experimental results, shown in Figure 3.9, show that the resin
undergoes autoacceleration during the polymerization and that the probe molecules do
indeed follow the curing process accurately. Similar results were obtained in the
MMAN-70 system when the pyrene data and thermocouple data were compared.
3.3.4. Sensitivity and interpretation of fluorescence response
3.3. 4. 1. Eflect of temperature on fluorescence response

To determine the effect of temperature upon the probe molecule fluorescence
emission, the probes were dissolved in non-reactive solvents and the fluorescence spectra
as a function of temperature was monitored in the A82 spectrometer. The samples were
stirred continuously, and their temperatures were monitored with a thermocouple. The
fluorescence emission was recorded for a series of temperatures starting at 30°C, ramping

up to 85°C or 90°C in 5°C increments, and then returning to 30°C.

37

Ethylene glycol with 10'3 wt.% phenoxazone 660 was excited at 550 nm. As the
tempearture was increased from 30°C to 90°C, the wavelength of maximum peak
emission shifted from 653 i 1 nm to 648 i 1 nm. This relatively small effect was
reversible as well. Clearly, the 20-nm permanent shift does not occur from temperature.

Tri(ethylene glycol) divinyl ether with 10'3 wt.% pyrene was excited at 330 nm.
The ratio of peak I (378 nm) intensity to peak 111 (389 nm) intensity, Illlm, is shown as a
function of temperature in Figure 3.10; the noise magnitude due to the spectrometer
detector would introduce a 3% error in the intensity ratio. As the temperature is increased
between 30°C and 50°C, the intensity ratio decreases slightly; however, for temperatures
greater than 50°C, the ratio is fairly constant. (When the sample is placed under a
nitrogen blanket, the temperature effect is even more negligible.) In addition, the effect at
low temperatures is in the opposite direction than that observed during cure, where the
intensity ratio increases dramatically. Here, as with phenoxazone 660, any temperature
effect upon the pyrene fluorescence response is reversible.
3.3.4.2. Comparison of probe responses during polymerization

The cure monitoring response of 10" wt.% pyrene in a solution of Derakane”
470-45 and 0.75 wt.% AIBN was compared to the cure montioring response of 10'4 wt.%
phenoxazone 660 in the same Derakane® system. For both probe molecule systems,
samples were placed in the AB2 spectrometer at 70°C for 75 minutes. The sample with
pyrene was excited at 383 nm, and fluorescence emission data were collected every 30
seconds at 406 nm and 415 nm (corresponding to bands I and 111, respectively). The
sample with phenoxazone 660 was excited at 540 nm, and fluorescence emission data

were collected at 609 nm and 629 nm (corresponding to the peak emission wavelengths

38

for cured and uncured samples, respectviely). As shown in Figure 3.11, the correlation
between the two probes in the Derakane® system confirms that they measure the same
variable. Similar results were obtained with samples containing pyrene and phenoxazone
660 concentrations of 102, 10'3 and 10'5 wt.%.
3.3.4.3. Effect of polarity on fluorescence response ofphenoxazone 660

In order to determine the solvatochromic origin(s) of the observed spectral shift,
the emission spectra of phenoxazone 660 were recorded in a series of solvents spanning a
range of polarity. For each solvent (Table 3.1), the wavelength of the phenoxazone 660
emission maximum is listed along with the values of E1-(30) (one measure of solvent
polarity), so (dielectric constant), 1!. (a second measure of solvent polarity), and u (dipole
moment), which are available in the literature." The data in Table 3.1 shows that the best
correlation is between emission maximum and solvent 1!. value, as reflected by the R2
value of 0.93. These studies suggest that the shift in the phenoxazone 660 emission arises

from alteration of the local polarity-polarizability during cure.

Table 3.1. Comparison of solvent polarity and phenoxazone 660 emission maximum.

 

 

 

 

 

 

Solvent Emission maximum £760) 50 u (Debye) 1t.
THF 595 nm 37.4 7.6 1.63 0.58
Acetone 612 nm 42.2 20.7 2.88 0.68
Acetonitrile 617 nm 46.0 37.5 3.92 0.86
DMF 623 nm 43.8 36.7 3.82 0.88
DMSO 633 nm 45.0 4.7 3.96 1.00

 

 

 

 

 

 

 

 

3. 3. 4. 4. Efl'ect of oxygen on fluorescence response of pyrene
Before implementing this cure monitoring technique in a fiber optic system, it was

desired to observe the response of the fluorescence ratio to changes in process variables.

39

For example, in the case of free radical polymerizations, many process lines are run in
oxygen-free environments to prevent oxygen from scavenging the free radicals that initiate
the polymerization. Thus, it was of interest to investigate the effect of oxygen upon the
ratio.

A sample of MMA and 1 wt.% V-70 with 10'3 wt.% pyrene was degassed by
cycling through three fieeze-pump—thaw stages. The sample was placed in the ABZ
spectrometer and cured at 40°C. The fluorescence emission was collected every five
minutes for 150 minutes with an excitation wavelength of 334 nm. The ratio of peak I
(378 nm) intensity to peak 111 (389 nm) intensity, Mn], is 0.93 prior to cure and subsequent
to cure becomes 1.08 (Figure 3.12), which is typical for the MMA system. However, the
fluorescence emission in the cured sample is approximately two to five times more
intense than that which is observed when oxygen is present in this particular system:
MMA with V-70 initiator (shown for 378 nm in Figure 3.13, top). The increased emission
does not affect the rolling time ratio, which reaches a value between 6.0 and 9.0 upon
cure completion regardless of sample oxygen content (see Figure 3.13, bottom).

3. 3. 4. 5. Eflect of temperature and initiator concentration an intensity ratio of
phenoxazone 660

Two other variables of interest include temperature and initiation concentration. In
order to investigate the effect of these variables upon the probe molecule fluorescence
response, a matrix was constructed. Styrene samples with 0.78, 2, 4, and 5 wt.% AIBN
and 10'3 wt.% phenoxazone 660 were cured in the AB2 spectrometer at 70, 80 and 90°C,
recording the sample emission at 570 nm and 590 nm every ten seconds with excitation at

510 nm. In Table 3.2, the time to cure completion and the final cure ratio, 1570/1590, are

40

shown as a firnction of temperature and initiator concentration. There are no values for
the 90°C and 4 or 5 wt.% AIBN because the bubbles produced by the release of nitrogen
gas as the radical irritator was formed became trapped in the polymer matrix, making it
impossible to obtain an emission spectrum. As can be seen, the final ratio values are
independent of cure temperature. However, the final ratio value does decrease with
increasing initiator concentration, possibly due to the increased presence of chain ends.
As expected, the results with the fluorescent probe molecule indicate that the

polymerization rate increases (i.e., polymerization time decreases) with increasing

temperature and initiator concentration.

 

 

 

 

 

 

 

 

Table 3.2. Effects of temperature and initiator concentration upon the time to
polymerization and final intensity ratio plateau for phenoxazone 660 in
styrene with AIBN.

0.78 wt.% 2 wt.% 4 wt.% 5 wt.%

T (°CL t(min) Ratio thin) Ratio t(min) Ratio t(min) Ratio
70 445 1.30 275 1.28 200 1.22 185 1.20
80 225 1.31 125 1.27 80 1.21 85 1.20
90 145 1.30 65 1.26

 

 

 

 

 

 

 

 

3. 3. 4. 6. Efl'ect of phenoxazone 660 concentration an intensity ratio

Finally, it was desired to ascertain the effect of the probe molecule concentration
itself upon the cure monitoring ratio. Derakane® 470-45 samples with 0.75 wt.% AIBN
and 102, 103, 10" and 10'5 wt.% phenoxazone 660 were cured in the AB2 spectrometer
at 70°C, recording the sample emission spectrum every 45 seconds with excitation at 540
nm. In Table 3.3, the results show that probe concentration does not affect the peak
emission wavelengths or the cure monitoring ratio when present in quantities of 10“ wt.%

or less. Furthermore, at these low concentrations, the characteristic pink coloration is not

41

noticeable in the sample, which may make the probe molecule even more appealing in an
industrial setting.
Table 3.3. Dependence of fluorescence peak emission wavelength and final cure

monitoring ratio of phenoxazone 660 in Derakane® 47045 with 0.75 wt.%
AIBN upon probe concentration.

 

 

 

 

 

 

 

Wt.% Probe Peak XL... (nm), Uncured Peak ’1'“. (nm), Cured Cured Ratio
10'2 651 641 1.08:0.05
10‘3 633 614 1.21:0.01
10“ _ 629 604 1.29:0.02
lo" 629 604 1.32:0.04

 

 

 

 

3.3.5. Equipment design

A fiber optic fluorescence sensor has been designed and assembled (Figure 3.14).
In order to design and operate a practical monitoring system, special emphasis needs to be
placed on the simplicity of the measurement scheme and the robust nature of the
equipment used. The wavelength of the light source is necessarily determined by both the
absorption profile of the probe molecule and any photochemistry that the incident light
might initiate. For most practical applications of optical monitoring. the latter constraint
is negligible due to the low light intensities used. A laser is the light source of choice for
such measurements because of its inherent spectroscopic selectivity and the favorable
properties of the output beam, thereby requiring a minimum of conditioning optics.
Routing the excitation light to the process of interest should involve few, if any,
adjustable optical components such as mirrors or lenses. Fiber optics can be used to route
the excitation light to the analyte and then collect emission from the analyte and send it to
the detection system. Optical fibers are available that can transmit UV and visible light

with comparatively low optical loss and thus are ideally suited to this type of application.

42

For phenoxazone 660, inexpensive optical detection components optimized for
pre-detennined wavelengths have been chosen and incorporated into a fiber optic-based
remote sensing system. Optical notch filters are used to separate 625 nm (M) and 605 nm
(1.2) from 540 nm (Am), and sensitive solid state detectors are used for each monitoring
wavelength. Measurement of the ratio of the output from the two detectors is
accomplished using a simple comparator/differential amplifier.

The output of such a system is proportional to the difference in intensity measured
at 625 nm and 605 nm. Calibration of the system is accomplished using two standards:
one where no curing occurs and the other where the system is cured completely.
Construction of the spectrometer is straightforward and has been accomplished using
inexpensive commercial components.

Experimental tests that utilize the in-situ cure monitoring device have successfully
been performed with the Derakane® 411-C50 system. Phenoxazone-doped resin with
AIBN was placed in the bottom of a watch glass on a hot plate. The samples were
illuminated by a 200 W Hg(Xe) lamp (Oriel Corporation, Stratford, CT) with a 540-nm
interference filter for four minutes. A fiber optic was positioned at a 45° to the sample to
collect the fluorescence emission from the excited probe molecules. Fluorescence
information was then routed to an Ocean Optic SD1000 fiber optic spectrometer
(Dunedin, FL), and the spectrum was monitored between 500 nm and 800 nm, with
spectra being taken every six seconds. Figure 3.15 shows an example of the results from
the intensity ratio measurements taken during these experiments. The results show that
the plateau is reached when cure is complete at 1605/1625 = 1.1, thus agreeing nicely with

the experiments performed in the AB2 spectrometer.

43

3.4. CONCLUSIONS

Two probe molecules for fluorescence monitoring of cure in polymer systems
have been examined. The first probe, phenoxazone 660, exhibits a solvatochromic shift
in fluorescence wavelength with increasing cure. For example, during the cure of a
common vinyl ester resin (Derakaneqp 411-C50), this probe exhibits a shift in the
wavelength of maximrun fluorescence intensity from 617 nm in the uncured system to
597 nm in the fully cured system. This shift was found to correlate with the local polarity
as characterized by the empirical quantity 11.. The intensity ratio of these two peaks
provides a measure of the degree of cure that is insensitive to background interferences
and does not lose sensitivity after the gel point. The second probe, pyrene, exhibits a
change in relative intensities for bands I and III (378 nm and 389 nm in MMA), as well as
an overall increase in absolute fluorescence emission. Fluorescence cure monitoring
strategies based upon these probes can be implemented using optical fibers to route both
the excitation light to the sample and the fluorescence emission from the sample;
therefore, these methods are amenable to in-situ and on-line monitoring of cure in

polymerization systems.

 

 

 

 

CH—OH

General chemical structure of the Derakane” vinyl ester resins.

Figure 3.1.

45

 

 

Figure 3.2. Chemical structures of phenoxazone 660 (a) and pyrene (b).

46

 
  
  

 

----- before cure

after cure

 

 

 

emission intensity, a.u.

 

 

 

I V V I 1

l I I ' I I l
550 600 650 700 750 800

wavelength, nm

Figure 3.3. Spectral shift of phenoxazone 660 associated with curing Derakane” 411-
C50 (with AIBN as initiator).

47

618-
616:
614:
612-
610:
608:
606i
604:
602;
6001
598.:
596:
594i

emission maximum, nm

 

592

 

Figure 3.4.

Blue-shifi in phenoxazone 660 peak emission as the cure of Derakane

cure time, minutes

411-C50 (with AIBN as initiator) progresses.

48

.a
In
I

.a.
(a)
l

J

.5
'N
1

l

.5
.a
l

 

4

intensity ratio (597 nm :617 nm)
2
I

 

 

0.9 -

0.8 -

07 T I I I I I ' I
o 15 - 30 45 so

time, minutes

Figure 3.5. Change in phenoxazone 660 fluorescence intensity ratio in curing
' DerakaneQ 41 1-C50 with AIBN as initiator.

49

Bandl Bandl“
. 378 nm 389 nm

\l /

300-

 

200-

100-

 

emission intensity, a.u.

 

 

‘I'

j I ‘ I

r ' 1 I
350 375 400 425 450 475

wavelength, nm

Figure 3.6. Fluorescence response of pyrene in MMA with AIBN as initiator.

50

1.20-

1.15-

1.10- C .

—I d

'o o

o 0:

l l O .
O

O

intensity ratio (I/lll)

 

 

time, minutes

Figure 3.7. Change in pyrene fluorescence intensity ratio in curing MMA with AIBN
as initiator.

51

 

..... 378 nm

 

 

 

rolling time ratio (Ir/'14)

 

 

time, minutes

Figure‘3.8. Change in pyrene fluorescence rolling time ratio at 378 nm in curing
MMA with AIBN as initiator.

52

. 160

 

 

 

1.34 0.
A :3' 140
E l ' ' . I .
C 1.2-
[s
‘- al . P120
(0 .0 p 0‘
-- 1.1- <0
E -1oo 3
c d _ . :
r~ 104 """" ’ -.
3; r80 2’.
Z ' ....................................... T 5
3 091 I I ~60 .
(U I o
h. - I I ,. O
> I
.5 081.. I 1-40
C ' _
9 l
E 0.7 -20

F V ' I I I I I I Y I I
o 10 20 so 40 50 60

time, minutes

Figure 3.9. Comparison between thermocouple (...) and phenoxazone 660 intensity
ratio (I) data in curing Derakane® 41 1-cso with AIBN as initiator.

53

 

 

 

 

 

 

1.30-1
. o D rampingup
125‘ O ramping down
1 D.
A1-20- o
g .
V1.15-1 D
.2
a .
l‘.’ 1.10. .' o
g . B . 1:1 :1
0’ 1.05- . o e
c :1 .
.9 - 1:1
.51.00- ooooooooooo U...D..... oooooooo
I . .
0.95-
” o n
t
0.90 I 1 r I 1 I , I 1 1 1 fi
30 40 50 60 70 80 90

temperature, °C

Figure 3.10. Change in pyrene fluorescence intensity ratio in tri(ethylene glycol) divinyl
ether with increasing (D) and decreasing (0) temperature.

54

 

 

 

pyrene intensity ratio
:3.
l

 

 

066- -03
0.64 - ’

d I"'008
0.62 .. r

‘ . -0.7
0.60

01191 Argsueiug 099 euozexoueqd

 

 

' I r T —r I ' I fi 1 I I I t 0.6

I r
0 500 1000 1500 2000 2500 3000 3500 4000 4500

time, seconds

Figure 3.11. Change in pyrene fluorescence intensity ratio (—) and phenoxazone 660
fluorescence intensity ratio (. . .) in curing Derakane® 470-45 with AIBN as
initiator.

55

intensity ratio (I/Ill)

 

 

0'80 I l I 1 I I r I I l' f I f

r 1
0 15 30 45 60 75 90 105 120 135 150

U I I l j

time, minutes

Figure 3.12. Change in pyrene fluorescence intensity ratio in curing MMA with V-70
- as initiator: without oxygen.

56

 

1 —o-—- without 0,

300 - + with o2 W-O'ofl-D

 

 

 

200 -1

.a
0'0
O
l a

.a
O
O
l

 

\

g

emission intensity at 378 nm, a.u.

 

 

 

   

 

 

 

 

 

 

' I ' I I ‘ I ‘ I r I ' T
0 20 40 60 60 100 120 140
time, minutes
A —l:—without02
.26" +with O,
D
E
C
3
co 4“
on!
10
,9 -1
0.!
m
I-
Q)
s 2"
0‘
G
.5 . n—o-a-o-o-a—o
'5
b
o ' I r I T I ' T T I ' I T I
0 20 40 60 so 100 120 140

time. minutes

Figure 3.13. Change in pyrene fluorescence intensity (top) and rolling time ratio
(bottom) at 378 nm in curing MMA with V-70 as initiator: without oxygen

(Cl) and with oxygen (0).

57

filters for it] 2 deteCtOI'S

 

 

 

-- ‘
light source -- (“1th

 

 

 

xexc comparator
3-way fiber V
coupler

3N1! ‘ WflFgeffW4YIS'fVTJ‘.’ ' 5g’;‘-‘,‘;‘ ‘4‘" '. r
‘ ‘ m ' - ‘0! o t . v o d L _’

 

analyte containing probe molecule

Figure 3.14. Diagram of in-situ monitoring instrument.

58

 

 

1.2-1
5 .
C
1.0
cc .I-db'JIa'H-Wa-
.. . n
E I
=1.o- '
10 ll...
0 d
3 u
209- I.-
E y I‘
3‘
'5 0.8-aI
C
.9
r
0.7 .1 . , . , . ,T r . , .T
50 100 150 200 250 300 350

time, seconds

Figure 3.15. Change in intensity ratio of phenoxazone 660 emission in curing
Derakane” 411-C50 with AIBN as initiator monitored by in-situ apparatus.

59

10.

11.

3.5. REFERENCES

. “Process Monitoring Sensors for Polymer Composites,” US. Department of

Commerce, NISTIR Report #4514, 1991.

R.L. Levy and SD. Schwab, “Monitoring the Composite Curing Process With a
Fluorescence-based Fiber-Optic Sensor,” Polymer Composites, 12(2), 96 (1991).

F.W. Wang, R.E. Lowry, W.J. Purnmer, B.M. Fanconi, and ES. Wu, “Fluorescence
Monitoring of Viscosity and Chemical Changes During Polymerization,”
Photophysics of Polymers, edited by GE. Hoyle and J.M. Torkelson, ACS
Symposium Series 358, American Chemical Society, Washington, DC, 454 (1987).

A. Stroeks, M. Shorhun, A.M. Jamieson, and R. Sirnha, “Cure monitoring of epoxy
resins by excimer fluorescence,” Polymer, 29, 467 (1988).

O. Valdes-Aguilera, C.P. Pathak and DC. Neckers, “Pyrene as a Fluorescent Probe
for Monitoring Polymerization Rates,” Macromolecules, 23, 689 (1990).

SF. Scarlata and J.A. Ors, “Fluorescence polarization: A method to monitor polymer
cure,” Polymer Comm, 27, 41 (1986).

D. Basting, D. Ouw and RP. Schafer, “The phenoxazones: A new class of laser dyes,”
Optics Communications, 18(3), 260 (1976).

D.S. Karpovich and GJ. Blanchard, “Relating the Polarity-Dependent Fluorescence
Response of Pyrene to Vibronic Coupling. Achieving a Fundamental Understanding
of the py Polarity Scale,” J. Phys. Chem, 99, 3951 (1995).

OR. Holtom, ASTM Proceedings, 1204-01, 1 (1990).

W.H. Tan, Z.Y. Shi, S. Smith, D. Bimbaum, and R. Kopelman, “Submicrometer
Intracellular Chemical Optical Fiber Sensors,” Science, 258, 778 (1992).

P. Suppan and N. Ghoneim, Solvatochromism, The Royal Society of Chemistry,
United Kingdom, 1997.

60

Chapter 4

IN-SITU MONITORING OF ACID CONCENTRATION IN
CHEMICALLY AMPLIFIED PHOTORESISTS

4.1. INTRODUCTION

With chemically amplified resists, the photoacid is key to circuit pattern
development, and the introductory discussion in this dissertation illustrates that the
deprotection reaction may not be limited to the illuminated areas since the photo-
generated protons may diffuse outside this region. For this reason, it is important to
understand the concentration profile and diffusional characteristics of the photoacid
within the actual photoresist thin film. The two processing steps that have the greatest
impact on the concentration and diffusional mobility of the photoacid are exposure and
post exposure bake. The acid is generally produced by the photolysis of an onium salt
photoacid generator (PAG). For a given PAG concentration, the exposure dose will
dictate the amount of acid that is generated. The temperature and duration of the post
exposure bake impact both the diffusional mobility of the acid and the rate of the
deprotection reaction. The post exposure bake is the time in which the actual acid
diffusion and deprotection reaction take place, and therefore the post exposure bake
conditions have the greatest impact on the resulting linewidth. Since pattern
development depends upon the regeneration and diffusion of each photoacid to effect
hundreds of deprotection events, the loss of any photons during the post exposure bake,
namely by neutralization with airborne bases or evaporation, can adversely affect this

process.

61

4.1.1. Utility of molecular probes

Molecular probes that exhibit a change in an optical property in response to a
change in the local environment can provide valuable in situ information about chemical
and physical processes. For example, a probe molecule may manifest a shift in
wavelength in an absorbance or fluorescence peak, or a change in excited state lifetime,
in response to a change in the local polarity or viscosity. Other probes may exhibit an
optical response that changes with changing acid concentration. Probes that exhibit a
change in optical properties are particularly useful since the excitation light may be easily
generated using a wide variety of sources and the excitation and/or emitted light may be
easily routed using optical fibers. In addition, the inherent timescale for the underlying
photochemical events (absorption or fluorescence) are very short (on the order of
femtoseconds to nanoseconds), providing excellent time resolution. Finally, it is usually
possible to identify probes that possess high quantum yields and large optical responses
such that reliable signals can be obtained using trace quantities of the probe.

There are a number of molecules whose optical response is sensitive to the pH of
the local environment.l Typically, these compounds possess two or more different
protolytic forms, each with a unique absorption and/or fluorescence spectrum. Examples
of such probe molecules include xanthene dyes, aromatic azines and coumarin dyes.
These probes have been used in various biological and medical applications to detect or
quantify acid concentration in a system.2 Recently, acid-sensitive dyes have been used to
investigate photoacid generation in microlithographic systems. For example, Pohlers et
al. have reported the use of such dyes to characterize the efficiency of photoacid

generators.3 These authors observed a shift in the absorbance spectrum of these dyes

62

upon protonation from the neutral state to the singly-charged state. In prior work, Pohler
et al. observed shifts in the absorbance spectra and large changes in fluorescence
intensity upon protonation of aromatic monoazines.4 Dentinger et al. evaluated several
xanthene dyes for monitoring and imaging photoacid production.5 Finally,
Okoroanyanwu et al. reported the use of an aromatic monoazine to monitor photoacid
generation in films and solutions.6
4.1.2. Studies of acid concentration in CARs

In this chapter, spectroscopic techniques are described that enable the
characterization of acid concentration in chemically amplified resists during the exposure
and post exposure bake steps. To measure the photoacid generation during exposure,
steady-state fluorescence of fluorescein is used. Unlike some of the previous studies
reported in the literature, these measurements on the absolute fluorescence intensity, but
rather used the relative intensities of two interdependent acid sensitive peaks. Secondly,
the steady-state absorption of crystal violet is used to characterize the acid concentration
in the photoresist during post exposure bake. Since the absorption spectrum of crystal
violet exhibits three bands, each corresponding to a different protonated form, the acid
concentration can be quantified over a broad range.

4.2. EXPERIMENTAL

4.2.1. Materials

These studies were performed using a corrurrercial photoresist, Apex-E 2408
DUV photoresist (Shipley Company, Marlborough, MA). The acid-sensitive fluorescent
probe fluorescein 548 (Exciton, lnc., Dayton, OH) was used in steady-state fluorescence

studies to determine photoacid generation during exposure. The dye crystal violet

63

(Sigma Chemical Company, St. Louis, MO) exhibits a sensitivity to acid concentration in
its absorption spectrum and was used to determine acid concentration during PEB using
steady-state absorption. All chemicals were used as received.
4.2.2. Methods
4.2.2.1. Absorption and fluorescence measurements

Absorption spectra of the photoresist and probe molecules in the photoresist were
obtained using a Hewlett Packard 8452A diode array spectrophotometer. A special
sample holder was manufactured for the spectrophotometer to hold quartz slides for film
studies. Fluorescence spectra of the materials were obtained using an Aminco-Bowman
Series 2 Luminescence Spectrometer. Fluorescence data were taken using the maximum
peak excitation and the maximum peak emission. Bulk solution samples were placed in
quartz cuvettes; one-micron film samples on quartz slides were obtained using a PWM32
spinner from Headway Research, Inc. (Garland, TX).

4. 2. 2. 2. Steady-state fluorescence measurements with photoresist films during
exposure

Samples for the photoacid generation experiments were prepared by first
dissolving 0.01 wt.% fluorescein in the commerical photoresist. The resist was spin
coated at 1550 rpm onto quartz crystal disks. The fluorescein dye was excited at 457.9
nm with a Coherent Innova 70 argon ion laser. The fluorescence signal was collected
with a Spex 1877 Triplemate monochromator with a subtractive dispersive filter stage
and a second spectrograph stage. A gated EG&G Princeton Applied Research Model
1530 C/CUV CCD, cooled to -120°C to reduce dark current, was used to detect the

signal. Simultaneously, the sample was illuminated with an Oriel 6035 mercury(argon)

pencil style lamp, fitted with a filter to select the UV lines, to induce dissociation of the
photoinitiator into the photoacid. This experimental setup is depicted in Figure 4.1.

4.2.2.3. Steady-state absorption measurements with photoresist films during
processing

These experiments were performed using crystal violet as the molecular probe.
The absorption spectrum of crystal violet exhibits three bands, each corresponding to a
different protonated form.’ To better understand this effect, and how it might relate to
the choice of the pump and probe laser wavelengths, the steady state absorption spectra
of crystal violet were recorded with 1 nm resolution as a function of pH using a Hitachi
U-4001 spectrometer. A constant crystal violet concentration (6x106 M) was maintained
for all absorption measurements by dilution fi‘om a single stock solution. The pH of the
solutions was controlled using a series of 0.2 M potassium chloride/0.2 M hydrochloric
acid buffer systems. Outside the buffering range of this system, concentrated
hydrochloric acid or 0.2 M potassium chloride was used to maintain the pH. A Fisher
Scientific Accumet pH meter and combination electrode with automatic temperature
correction was used for all pH determinations. The pH of the solutions ranged from 0.26
to 5.12, which was measured after the addition of the crystal violet. The ionic strength of
the solutions was not controlled, as it does not give rise to significant error in our system.

4.3. RESULTS AND DISCUSSION

4.3.1. Probe molecule selection

As has been discussed above, fluorescence monitoring of chemical and biological
systems has been widely used to gain a greater insight into the micro-environment about
carefully selected probe molecules. This in-situ technique offers high sensitivity on fast

time scales with minimal changes in the system under investigation since most probe

65

molecules need only be present in minute amounts for adequate signal detection.2
Numerous dyes and chromophores are available for fluorescence monitoring and have
been used successfully to detect such properties as polarity, viscosity, ion concentration,
diffusion coefficients, etc. A literature survey of research accomplished with fluorescent
probe molecules yielded a list of possible candidates that could be used in this study.
From this list, likely candidates were chosen with several requirements in mind. First of
all, the probes, which are typically large aromatic molecules, must be soluble in the resist
mixture, which is composed of long, organic polymer chains. Secondly, the optical
response of the probes should be distinguishable from the optical response of the resist
mixture itself. For example, the Apex-E 2408 DUV photoresist absorbs strongly below
310 nm and possesses a weak fluorescence, as shown in Figure 4.2, so the optical
response of the probe molecules should be in the near-UV or visible regions of the
spectrum. In the case of fluorescence, it is also important that it be possible to induce this
fluorescence emission using available laser lines. Finally, the probes must not
significantly alter the kinetics of the system. If this criterion cannot be strictly met, then
it should be possible to extrapolate what the measurements would have been in the
absence of the probe. For example, probes which trap acids in order to measure the
concentration could adversely affect the photoresist system (if one photoacid molecule is
taken from the system, it would decrease the ntunber of deprotection events drastically).
Finally, the fluorescence probe or series of probes should be sensitive throughout the

entire photoresist process, from prebake to illumination to postbake.

4.3.2. Acid-sensitivity of fluorescein

F luorescein (shown in Figure 4.3) was chosen as a molecular probe to study
photoacid generation in the resist during exposure because it exhibits a large fluorescence
response that is well-characterized and dependent upon acid conc1‘:ntr'ation.8'9 Figtu'e 4.4
shows the absorption and emission spectra of 0.01 wt.% fluorescein in Apex-E 2408
DUV photoresist. From these spectra, it is evident that the optical response of fluorescein
is not hampered by the photoresist itself and excitation of the probe is possible using one
or more lines from an argon laser.

In order to verify the acid-sensitive fluorescence of fluorescein, preliminary
studies were first done in bulk solution. The emission spectrum of 0.01 wt.% fluorescein
in the photoresist was collected using an excitation wavelength of 460 nm. Two
microliters of sulfuric acid were then added to the cuvette, which was stirred in situ. The
emission spectrum of the acidic solutions was collected after waiting ten minutes and
compared to the original formulation.

Figure 4.5 shows that the acid-sensitive response is characterized by a blue-shifi
in the fluorescence emission and by a change in the relative intensities of two
interdependent peaks characterizing that emission.

4.3.3. Study of photoacid generation during exposure in photoresist films using
steady-state fluorescence

Similar studies, in which small amounts of a mineral acid were added to the resist
solution before spinning films, further demonstrated that the acid-sensitive response is
characterized by the relative intensities of these two interdependent peaks. Therefore, we
monitored photoacid generation during exposure by ratioing the intensity of the

fluorescein fluorescence emission at these two wavelengths. By using this ratioing

67

method, rather than absolute intensities, the measurements are no longer dependent upon
fluorescein concentration, detector settings, or scattered and background light.

A representative fluorescence profile obtained via in-situ, real-time experiments
during exposure is shown in Figure 4.6. This plot shows the relative change in peak
intensities at 525 nm and 560 nm as a function of photoacid generation during exposure.
When the intensity ratio of 525 nm to 560 nm is taken for the duration of the monitoring,
Figure 4.7 is obtained. In this experiment the build-up in the photoacid concentration
occurs over a relatively long timescale since a low intensity mercury-argon pen lamp was
used to induce the acid production, rather than the excimer laser common used in
industrial photoresist processing. The photoacid generation in this system can be
characterized as a first order response to a step-change perturbation from equilibrium (in
this case the perturbation corresponds to the sudden illumination and the response is the

dissociation of the photoacid generator):

I

[H*1=1H*1ss<1-e'?1 (4.1)

where [H+] is the acid concentration at time t, [Hijss is the steady-state acid concentration
in the film after exposure, 1: is the time constant associated with a specific lamp intensity
(I will decrease as the intensity is increased), and t is the time of exposure. By
calibrating the fluorescence response of the probe molecule with standard solutions, it
will be possible to measure the concentration of photoacid present in the film as a

function of time or exposure dose.

68

4.3.4. Acid-sensitivity of crystal violet
Crystal violet is a triphenylmethane dye that has been used widely as an indicator

'0 The native form of

in acid/base titrations, especially in nonaqueous solvent systems.
crystal violet (CW is violet in color. Monoprotonation to produce HCV2+ yields a green
solution and further protonation to HzCV3+ produces a yellow solution. The protonation
of the amine groups on the phenyl rings is thought by Tuthill et al. to be responsible for
the changes in the steady-state absorption spectroscopy of this molecule.7 The three
protonated forms of crystal violet and their equilibria are shown in Figure 4.8. The
endpoints and transition points of crystal violet are difficult to differentiate, because of
the similarity of the K. values for the different forms, which are discussed below.

Figure 4.9 shows the absorption spectrum of 1 wt.% crystal violet in Apex-E 2408
DUV photoresist spin coated onto a glass slide. The fluorescence emission of this sample
was not detectable using the Aminco-Bowman spectrometer; however, it was possible to
collect this information using an Ocean Optics SD2000 fiber optic spectrometer
(Dunedin, FL). The sample was excited using all lines from an air-cooled argon ion laser
(Omnichrome, Chino, CA), and the fluorescence emission spectrum was obtained via a
fiber optic positioned at a 45° angle to the sample. This fluorescence information was
then routed to the Ocean Optic spectrometer and is also displayed in Figure 4.9. From
these spectra, it is evident that the optical response of crystal violet is in the spectral
window not occupied by the photoresist itself.

In order to verify the acid-sensitive fluorescence of crystal violet, preliminary

studies were first done in bulk solution. The emission spectra of crystal violet in acidic

and basic water were collected using the Ocean Optic spectrometer. Figure 4.10 shows

69

that the acid-sensitive response is characterized by a blue-shift in the absorption and
fluorescence emission.

4.3.5. Study of photoacid concentration during processing in photoresist films
using steady-state absorption

The various forms of crystal violet in solution were characterized through the use
of buffered pH systems, ranging from pH~5 to pH~0 (corresponding to an acid
concentration range of -_~10’5 M to ~1 M). The changes in the absorption spectra as a
function of acid concentration are presented in Figure 4.11, and these results are
applicable to the resist films as well. The native, deprotonated, form of CV+ absorbs
strongly at 590 nm. As the pH decreases, the onset of a new absorption band centered
near 630 nm, simultaneous with the decrease in absorbance at 590 nm, is observed. This
spectral change is the result of protonation to form HCVZL Additional protonation
produces an absorption band at ~430 nm at the expense of the 630 nm and 590 nm bands.
The equilibria among these three species of crystal violet may be described as:

llzcv3+t—£1—>Hcv2+ +H+ e—EL—scvt +H” (42)
resulting in the following expressions for the two equilibrium constants:

K =[11cv2+][11“]

 

[HZCV3+] (4'3)
K = [CV+][H+] (4.4)
[HCV2+]

From these data, the constants for the protonation/deprotonation equilibria can be
determined by applying Beer’s law to the solutions and reduction of the data through a

series of simultaneous equations for each pH. The molar absorptivities at each

70

wavelength for each species are available in the literamre.“ At a given pH, solution
absorption was determined at 430 nm, 590 nm and 630 nm, leading to the concentrations
of each form (see section A.1 in the Appendix for the derivation). The species
concentrations were then used to calculate values for the equilibrium constants K. and
K2. K. was determined to be (2.5 i 1.5)x10’2 moles/L, and K2 was determined to be (6.7
i 2.5)x10'3 moles/L. These values correspond to pK. = 1.67 i 0.29 and pKz = 2.20 i
0.19. The first pK. for another triphenylmethane dye, malachite green, which differs
from crystal violet by the absence of one dimethylarnino group from a phenyl ring, was
determined to be ~ 1.3,lo in reasonable agreement with the data These equilibria
constants were then used to calculate alpha values for each crystal violet species, which
in turn enable the determination of the acid concentration in the photoresist given an
absorption spectrum of crystal violet in that medium (see section A2 in the Appendix for
the derivation).

The changes in the steady state absorption response of crystal violet doped in a
polymer film were examined as a function of processing, and these data are presented in
Figure 4.12. An absorption spectrum taken just after spin coating shows the native,
deprotonated form of crystal violet with an absorption maximum at 590 nm. After
exposure of the film and subsequent post exposure bake, a decrease in absorption at 590
nm and an increase in the intensities of the 630-nm and 430-nm bands were observed.
This change is due to the protonation of crystal violet to both the monoprotonated and
diprotonated forms in response to the photo-generated acid. Re-examination of the film
several hours afier post-exposure bake shows a change in the crystal violet absorption

spectrum. The 430-nm band (H2CV3+) has decreased, and the 630-nm band (HCVZT) has

71

increased, indicative of a decrease in the extent of crystal violet protonation due to
photoacid diffusion and/or consumption in the film. One possible mechanism for proton
loss is long-term exposure of the film to air-home amines.W3 Thus, steady-state
absorption spectroscopy provides a facile means to quantify the photoacid production and
consumption as a function of processing.

4.3.6. Effect of probe presence on photoacid concentration during processing in
photoresist films using steady-state absorption

One of the concerns initially listed for the probe molecules was to be able to
account for the effect of its presence upon the reaction occurring in the resist. One way
to ascertain the effect of the probe upon deprotection efficiency is to monitor the steady-
state absorption of the photoresist film during processing.

The absorption spectrum of Apex-E 2408 DUV photoresist is shown in Figure
4.2. There are two features of note: one peak between 200 and 250 nm and another
between 250 and 300 nm. The first peak is characteristic of propylene glycol methyl
ether acetate (PGMEA), used as the solvent in the resist formulation; the second peak is
characteristic of poly(p-t-butoxycarbonyloxystyrene)-co-poly(vinyl phenol). It is this
second peak that exhibits a change in intensity with deprotection. As the acid diffuses
through the film during postexposure bake, the intensity at 278 nm increases due to the
cleavage of the pendant groups, which converts the t-butoxycarbonyl (TBOC) portions of
the copolymer to vinyl phenol groups. When crystal violet is introduced into the resist
formulation, some of the acid generated upon exposure is taken up by this dye to produce
the spectral shift that monitors acid concentration. This reduces the number of

deprotection events so that the increase in intensity at 278 nm is not as high as in the neat

72

resist. A comparison of the absorption spectra for the neat and doped photoresist as a
function of film processing is shown in Figure 4.13.

A series of experiments was carried out to estimate the effect of crystal violet
upon the extent of deprotection in the resist. Apex-E 2408 DUV photoresist was doped
with 0.1, 0.2 and 0.3 wt.% crystal violet. Neat and doped samples were spin coated onto
clean quartz slides at 3000 rpm to produce a thin film. An absorption spectrum of each
sample was collected after spin coating using the Hewlett Packard 8452A diode array
spectrophotometer. Then the samples underwent post-application bake (PAB) for 60
seconds at 90°C, exposure with a 200 W Hg(Xe) lamp (Oriel Corporation, Stratford, CT)
for 30 seconds, and post-exposure bake (PEB) for 90 seconds at 90°C. An absorption
spectrum of the samples was also taken after each of these processing steps.

The absorbance intensity at 278 nm for each sample after processing step was
ratioed against the absorbance intensity at 278 nm for the respective sample taken after
spin coating. This information is presented below in Table 4.1. No significant change
attributed to the deprotection reaction was noted until afier PEB. This is to be expected
since the acid is generated by exposure but does not start diffusing and reacting
significantly until the temperature is elevated during PEB. As the concentration of
crystal violet increased, the deprotection reaction efficiency decreased as predicted. This

efficiency was estimated as follows:

Idoped, PEB Tldoped.8pin “00% (4-5)

% efficiency =
Ineat, PEB T IneaLspin

where l is the absorbance intensity ratio at 278 nm for the specified sample (neat or

doped with crystal violet) at the specified processing step (after spin coating or after

73

PEB). This assumes that the neat samples are 100% deprotected, which is a good

approximation in this case where the slides are “overexposed.”

 

 

 

 

 

 

 

 

 

 

 

Table 4.1. Comparison of the absorbance ratio at 278 nm for the neat Apex-E 2408
DUV photoresist film to that of the crystal violet doped films during
various stages of processing.

Absorbance Ratio During Film Processing
Photoresist Alter PAB After exposure After PEB % efficiency
Sample deprotection
Neat 0969:0009 0.986:0.01 1 1.137:0013 100
0.1 wt.% CV 0976:0007 0991:0007 1.108:0.008 78
0.2 wt.% CV 0973:0001 0989:0001 1037:0006 27
0.3 wt.% CV 0971:0003 0983:0003 0986:0004 0

 

 

Thus, it is important to avoid too high a concentration of crystal violet in the
photoresist, else the entire p0pulation of photoacid will be sampled by the probe molecule
such that none will be left for the deprotection events. A concentration of 0.1 wt.%
crystal violet in the photoresist film is an acceptable compromise for the major
monitoring concerns. Over 75% of the photoacid continues deprotecting during PEB,
and the sensitivity of the probe molecule is not compromised (i.e., the absorption spectra
of the crystal violet registers the same acid sensitivity as seen in the 0.3 wt.% crystal
violet case) or lost below the detection limits of the spectrometer.

4.4. CONCLUSION

Spectroscopic techniques have been developed to characterize acid concentration
and mobility in chemically amplified resists during the exposure and post exposure bake.
Steady-state fluorescence of fluorescein is used to characterize the acid build-up during
exposure in situ. The acid-sensitive response was characterized by ratioing the intensities

of two interdependent peaks. By using this ratioing method, rather than absolute

74

intensities, the measurements are independent of detector settings, scattered and
background light, or small changes in fluorescein concentration (such as that caused by
photobleaching). Secondly, crystal violet is used to characterize the acid concentration in
the photoresist during post exposure bake. The absorption spectrum of crystal violet
exhibits three bands that each corresponds to a different protonated form. Therefore, the
shape of the absorption spectrum depends upon the local acid concentration, and this
dependence has been demonstrated both in aqueous solution and in polymer photoresist
films. These data are consistent with relatively high local proton concentrations in the
resist films subsequent to UV irradiation. The development of these experimental
techniques can then be used for a systematic study of acid concentration in situ during the
exposure and post exposure bake. For a given resist, a more thorough understanding of
the factors that affect acid concentration (e. g., post exposure bake temperature and
duration) can allow improved microlithographic pattern generation. In addition, this
information could be used for the development and optimization of new CARs designed

to control the mobility of the acid.

75

 
  
    

Hg(Ar) Pen Lamp '
with UV Filter

\
\\

 

 

 

Electronic Spectrometer
Shutter
Liquid Nitrogen
o I Cooled CCD
457.9 nm
Ar+ Laser

Figure 4.1. Experimental setup for in-situ fluorescence experiments to determine
photoacid generation during exposure.

76

J
N
I
I
d
'N

 

 

 

:
. abs 0
§
. ' fi'
“’- 2.
8 0.8. 2
g :
+3 ‘ 5

O -.
m 0.6 g
D (D
m - :3
o

'o

a) 0.4 - CD
.2 5'
E ‘ S"
8- O.2 d m
o :3
c 1‘
9’
F

 

 

0.0 A

T 1 I ' ‘

 

 

T l V I U I Y

I 1
200 250 300 350 400 450 500 550 600
wavelength, nm

Figure 4.2. Absorption (abs), excitation (ex) and emission (em) characteristics of
Apex-E 2408 DUV photoresist.

77

 

Figure 4.3. Molecular structure of fluorescein.

 

 

 

 

'4 r
1.0- .1,0 3
2
=: - ' g
as _
- 013- ~(t8 F:
8 m
o.
c 4 .

m a:
“e 0.6- r—O.6 3
o 6'
w 2
'0 0.4- -o.4 _.
an a
y. t a
To ‘ g
E 02 02 3'

h u d _ o
o 1‘
c I b m
'c

0'0 v I I I I r T T 1 1 r T 0'0

I I
350 400 450 500 550 600 650 700

wavelength, nm

Figure 4.4. Absorption and emission characteristics of fluorescein in Apex-E 2408
DUV photoresist.

78

08- -025

 

 

 

 

0.7- ’
‘ -(L20

3: 0.6-I L g
(D " 6.
>1 0'5“ _o 15 9-
:: - c;
(D '1 a
C
Q) 0.4-! _ s.
'E 1 010 g
c , " '
.9 0 3 '1 g
In -‘ . ‘5
g 0.2.. p
o «I -0.05 C

0:1-

0.0 I I Tfi I I I T I I I I F r l 0'00

1 I
460 480 500 520 540 560 580 600 620 640

wavelength, nm

Figure 4.5. Change in fluorescein fluorescence emission in Apex-E 2408 DUV
photoresist upon addition of sulfuric acid: before acid addition (—) and
afier acid addition (. . .).

79

 

   
  

 

 

 

 

 

—l:l— 0 sec
‘ —O— 10 sec
6000 - —A— 15 sec
, —v— 20 sec
. 5000 1 —<>— 30 sec
2
‘0 4
>2
.1: 4000 .
m
c
2 d
5 3000 1
c:
.9 .
8
'- 2000 —1
E
m 1
1000 - ’
V I T I V I I I V j
450 500 550 600 650 700

wavelength, nm

Figure 4.6. Fluorescence spectra of fluorescein in Apex-E 2408 DUV photoresist film
as a function of increased exposure time.

80

‘L12-
‘L10-

‘LOB-n 0' I.

Q 0
1.06- ~.o--£;~’"”‘°""’.""’
1.04.: . ’O.’. o
‘ o
1.02- r
1.00-

(198-'

 

(196-

intensity ratio (525 nm : 560 nm)

 

0-94 fl I I I I I I
20 4O 60

time, seconds

Figure 4.7. Intensity ratio at 525 nm and 560 nm of fluorescein in Apex-E 2408 DUV
photoresist film as a function of increased exposure time.

81

 

 

 

 

H3C\ [CH3
N
C Violet
“79°“
Inc—r?! Til-CH3
CH3 CH3
1.1+
H3C\ [CH3
ONH
0 HCV2+
C’ Green
km = 630 nm
“SC-ll! III—CH3
CH3 CH3
H4-
H3C\ [CH3
ONH
O H2CV3+
Yellow
(2‘ Am = 430 nm
H3C-NIH N-CH3

H3 ”3

Figure 4.8. Protonation/deprotonation equilibria for crystal violet. The corresponding
spectra are shown m Figure 4.11

82

 

 

 

 

1.0a -1.0 a
o
. W
a ‘ ' 3
' m
“l. 0.8- -0.8 g
4’ ca
2 . - a
co m
‘3 0.61 -0.6 g.
8 a:
(n
n I ----¢. D a.
‘5 a
'o 0.4- -0.4 -.
a a
.. ‘ _ (n
°° 3
E 02. -02 =3
3 :<
C ‘ P m
.Qo. F

0.0 "; 0.0

1 l U I I l T T I I f T 1 1 T I I I V

r
400 450 500 550 600 650 700 750 800 850 900
wavelength, nm

Figure 4.9. Absorption and emission characteristics of crystal violet in Apex-E 2408
DUV photoresist.

83

normalized absorbance, a.u.

normalized absorbance. a.u.

Figure 4.10. Absorption and emission characteristics of crystal violet in bulk acidic

 

acidic solution

1.0-

 

 

 

0.8 -

 

~1.0

-0.8

+0.6

l-O.‘

, 0.0

 

 

.°
:3

 

 

 

 

 

 

 

 

     

 

' I I I ' T ' I ' I V I fl I ' I
350 400 450 500 550 600 050 700 750 000
wavelength, nm
1.0 . basic solution f __ 1.0
1 \. '
I
0.8 . t. _ _ _, .. 0.3
. l .
\
0.5 .. \' _ 0.6
‘l
4 \ r
0.4 4 \ .. 0.4
0.2 q \w - 0.2
0'0 ' I ' I I I w 17 ' I ' I ' r I 0'0
350 400 450 500 550 600 650 700 750 800

wavelength, nm

(top) and basic (bottom) solutions.

84

we ‘Mgsuewl uolssgme pezueuuou

'n'e 'Ausuaiug uogssgwe pezueuuou

“6°: [H'] = 7.59x10" M
0.55 - .".

0.50 - ' '

0.45: 3
0.40: 3 [H’] = 1.55x10'2 M
035: 3' 3;“,
(L30: :' / 1 \
0.25: I ‘
°°2°‘ [H’] = 5.5x10'1 M I 1. .
0.15 q . .
0.10:
0.05.1
0.00

-0.05 f . ,
350 400

absorbance, a.u.

 

 

 

Y T V V

‘V fi V

I I I I 1
450 500 550 600 650 700

wavelength, nm

Figure 4.11. Absorption spectra of crystal violet in aqueous solution. [IT] for the
solution spectra are indicated on the plot.

85

After spin

 

 

0.14 - \
0.12 -«
. 2 h after PEB 5 d after PEB
0.10 - I!
I
. I o o ,I
3. o 08 Immediately i,
“a after PEB '
m \ 1‘
0 0.06- ‘\ I: l
c \ ’o\\ ’0‘. 0. o\
m ‘ ‘ ‘. --"".I .o ' °
D .{ .- u.‘-\.
‘- 0.04 "l f- ‘ I 'z: '2‘
o J \ ‘ o I ‘. 0‘
U) " ‘ I. o“ \\ ’ .° ' fl '.3
'0 . . . 0.0" .I .‘ "/50 at” K
m 0.02 -I I' o) bf o/ \°. ovv I fl], \\ .
w; \. .I \ -...~ -.>‘~~“~\~,,~” .
1 "V. I o ' o
v' \ ./ O
0.00 -
'0‘02 ' i ' i ' I ' I I I I I ' I
350 400 450 500 550 600 650 700

wavelength, nm

Figure 4.12. Absorption spectra of crystal violet in Apex-E 2408 DUV photoresist film

during and after processing.

86

Figure 4.13.

0.8 ..

 

 

 

 

 

 

. I
'I
0.7 - :I photoresist only:
I
' In. after spin
I
0.6- ‘1 | --- after PEB
I
. " l
3. I '
m 0.5-I
o. 1
U 0.4 u
I:
G 4
.D
L-
O
In
.0
m

 

 

 

 

250 300 350 400 450 500 550 600 650 700

wavelength, nm

 

 

0 7 u n
‘ with e mi violet:
0.6 q ry
after spin
0.5-l --- after PEB

 

 

 

 

absorbance, a.u.
O
U
4

 

  

 

    

 

V 1 V I V ' V I *7 I 1

I
250 300 350 400 450 500 550 600 650 700
wavelength, nm

Effect of 0.3 wt.% crystal violet in Apex-E 2408 DUV photoresist film
upon the absorbance spectrum before and after processing.

87

10.

11.

4.5. REFERENCES

. R.P. Haugland, Handbook of Fluorescent Probes and Research Chemicals, 6m ed.,

edited by M.T.Z. Spence and ID. Johnson, Molecular Probes, Inc.: USA, 1996.

J. Slavik, Fluorescent Probes in Cellular and Molecular Biology, CRC Press, Inc.:
Ann Arbor, 1994.

G. Pohlers, J .C. Scaiano, and R. Sinta, “A Novel Photometric Method for the
Determination of Photoacid Generation Efficiencies Using Benzothiazole and
Xanthene Dyes as Acid Sensors,” Chem. Mater., 9, 3222 (1997).

G. Pohlers, S. Virdee, J.C. Scaiano, and R Sinta, “Aromatic Monoazines as
Fluorescent Sensors for Photoacid Generation in Thin Polymer Films,” Chem. Mater.,
8, 2654 (1996).

RM. Dentinger, B. Lu, J .W. Taylor, SJ. Bukofsky, G.D. Feke, D. Hessman, and RD.
Grober, “On-wafer photoacid determination and imaging technique for chemically
amplified photoresists,” J. Vac. Sci. Technol. B, 16(6), 3767 (1998).

U. Okoroanyanwu, J.D. Byers, T. Cao, S.E. Webber, and CG. Willson,
“Deprotection Kinetics of Alicyclic Polymer Resist Systems Designed for ArF (193
nm) Lithography,” Micro and Nana-patterning Polymers, edited by H. Ito, E.
Reichmanis, O. Nalamasu, and Y. Ueno, American Chemical Society, Washington,
DC, 1998.

S. M., Tuthill, O. W. Kolling, K. H. Roberts, “Photometric and visual titration of
certain alkaloids in glacial acetic acid using malachite green as indicator,” Anal.
Chem, 32, 1678 (1960).

M.M. Martin and L. Lindqvist, “The pH Dependence of Fluorescein Fluorescence,” J.
Luminescence, 10, 381 (1975).

H. Leonhardt, L. Gordon, and R. Livingston, “Acid-Base Equilibria of Fluorescein
and 2’,7’-Dichlorofluorescein in Their Ground and Fluorescent States, " J. Phys.
Chem, 75(2), 245 (1971).

E. Bishop, “Indicators for Non-Aqueous Acid-Base Titrations,” Indicators, edited by
E. BishOp, Pergamon, New York, 1972.

J. B. Conant, T. H. Werner, “The determination of the strength of weak bases and
pseudo bases in glacial acetic acid solutions,” J. Am. Chem. Soc, 52, 4436 (1930).

88

12. CA. Mack, “Lithographic Effects of Acid Diffusion in Chemically Amplified
Resists,” Microelectronics Technology: Polymers for Advanced Imaging and
Packaging, edited by E. Reischmanis, C.K. Ober, S.A. MacDonald, T. Iwayanagi,
and T. Nishikubo, American Chemical Society, Washington, DC, 1995.

13. L.F. Thompson, “Resist Processing,” Introduction to Microlithography, 2"‘1 ed.,

edited by L.F. Thompson, C.G. Wilson, MJ. Bowden, American Chemical Society,
Washington, DC, 1994.

89

Chapter 5

FREE VOLUME IN CHEMICALLY AMPLIFIED PHOTORESISTS

5.1. INTRODUCTION

As discussed in Chapter 1, in order to predict adequately the behavior of
chemically amplified photoresist, it is crucial to understand and characterize the mobility
of the photo-induced acid that provides the sensitivity to pattern development on
semiconductor substrates. A portion of this task is being able to quantify the acid
concentration during processing, which was addressed in the previous chapter. The other
portion is to determine the effect or impact of available free volume upon the diffusional
mobility of the photoacid during post exposure bake. First of all, free volume increases
with the higher temperature of the bake, thus increasing the diffusivity of the photoacid

according to the following relationship:l
-5, (5.1)
D T = D ex
( ) o P[ RT J

where D is the diffusivity, Do is a reference diffusivity, Ed is the activation energy for

 

diffusion, R is the gas constant, and T is the absolute temperature. In addition, there is
“chemically produced” free volume arising from the deprotection reaction, which
liberates carbon dioxide and isobutylene. These small molecules diffuse out of the film
relatively rapidly (the characteristic time for diffusion through the one micron film is on
the order of a second), creating free volume that is transient in nature. This “transient”
free volume is eliminated by the relaxation of the polymer chains, leading to a decrease in

the thickness of the film. In general, free volume provides the acid greater mobility and

90

freedom to attack the next available pendant TBOC group. Thus, diffusion of the
photoacid should be greater as the number of deprotection reactions increases. Recently,
J akatdar er al. have developed a model to correlate the deprotection-induced thickness
loss of the photoresist film as a function of post exposure bake;2 however, no research has
been undertaken to understand the correlation between the free volume in the photoresist
films and the photoacid diffusional mobility.
5.1.1. Utility of molecular probes

In addition to measuring the concentration of acid as discussed in the previous
chapter, fluorescence can be used to probe the local environment of the system. There
are a number of probe molecules that exhibit an increase in the fluorescence signal and
excited-state lifetime as the local viscosity is increased. As the molecular rotations
become hindered, non-radiative pathways for energy transfer are less efficient; therefore,
more of the excited state energy must be dispelled through radiative (fluorescent)
pathways. Other free volume sensitive probes are based upon photoisomerization and
intramolecular excimer formation. With a small amount of a viscosity-sensitive
molecule, the micro-viscosity or free volume of the system can be measured. Many
researchers have exploited this technique to gather such information in polymer systems.
For example, Francis Wang et al. have used l,3-bis-(l-pyrene) (BPP) to monitor the
polymerization of methyl methacrylate in situ.3 Moorjani and coworkers further
demonstrated the utility of fluorescence monitoring when dealing with diffusion-
controlled reactions in media of different viscosity.4 Victor and Torkelson have also used

molecular probes to measure the free volume available in polymer structures.5

91

5.2.1. Studies of free volume evolution in CARs

In this chapter, spectroscopic techniques are identified that may be used to
characterize acid mobility in chemically amplified resists during the post exposure bake
steps. To measure the free volume changes during post exposure bake, ground-state
recovery measurements were used to characterize the excited-state lifetime of crystal
violet. As is shown below, the lifetimes of the excited state for this triphenylmethane
“propeller” molecule increase as the local free volume decreases, and molecular rotation
is hindered. In addition, the effects of the post exposure bake time upon the overall film
thickness and macroscopic free volume were investigated using dynamic film thickness
measurements.

5.2. EXPERIMENTAL

5.2.1. Materials

These studies were performed using a commercial photoresist, Apex-E 2408
DUV (Shipley Company, Marlborough, MA). The dye crystal violet (Sigma Chemical
Company, St. Louis, MO) exhibits a sensitivity to free volume in its excited state
lifetime. It was used to determine free volume generation during PEB using ground state
recovery measurements. All chemicals were used as received.
5.2.2. Methods
5. 2. 2. I. Ground-state recovery measurements with photoresist films during processing

The picosecond pump-probe laser spectrometer6 used for the ground-state
recovery measurements is shown in Figure 5.1. In these experiments, a mode-locked CW
Nd:YAG laser (Coherent Antares 76-S) produces 30 W of average power (1064 nm, 100

ps pulses, 76 MHz repetition rate). The output of this laser is frequency-doubled to

92

produce ~2.l W of average power at 532 nm. The second harmonic light is used to
excite two cavity-dumped dye lasers synchronously (Coherent 702-3). Both lasers
operate with Rhodamine 66 laser dye (Kodak). The output of each laser is ~100 mW
average power at 8 MHz repetition rate with a pulse that produces a 7 ps F WHM
autocorrelation trace using a three plate birefringent filter. The pump laser wavelength
was set to 589.5 nm, coincident with the absorption maximum of the first singlet
transition of crystal violet. The probe wavelength was chosen to lie within the absorption
band of crystal violet. The probe beam was polarized at 54.7° with respect to the pump
beam to eliminate molecular reorientation contributions to the signal. The time
resolution of this system, ~10 ps, is determined by the cross-correlation between the
pump and probe laser pulse trains. Detection of the transient signals was accomplished
using a radio and audio frequency triple-modulation scheme, with synchronous

”'9 Each reported time constant is the average of at least five

demodulation detection.
individual determinations that are themselves the average of seven to ten time-scans.
Samples for these experiments were prepared in the following manner. Resist
solution with 0.1 wt.% crystal violet was spin coated at 2000 rpm onto quartz slides and
processed according to standard procedures. One set of samples experienced no
processing subsequent to spin coating, while a second set of samples was processed
through the post-exposure bake. The samples were mounted in a manually driven
translation stage and the two laser beams were co-focused on them. To minimize the

effects of photobleaching and thermal degradation, the slide was translated as the sample

was scanned, and neutral density filters were used to attenuate the incident light.

93

5. 2. 2.2. Matti-wavelength interferometry

Multi-wavelength interferometry was performed using the Inspector Model 800
(SC Technology, Fremont, CA). The instrument was used to observe changes in the
macroscopic free volume of the photoresist as the deprotection reaction occurs. Multi-
wavelength interferometry is used to determine film thickness in semi-transparent films
by measuring the interference between reflections of incident light from the film top
surface and from the film interface with the substrate, in this case the quartz slide. A
generalized schematic of this method is shown in Figure 5.2. Thus, small changes in the
film thickness due to the effects of the deprotection reaction can be recorded. In these
experiments, the time-evolution of the free volume (averaged over the area illuminated
by the optical fiber probe) was characterized by continuously monitoring the resist film
thickness during the post exposure bake.

The instrument was calibrated in the reflectance mode using a clean quartz slide.
Film thickness measurements were calculated using the Prolight data acquisition package
with the LithoPak applications module, which accompanies the Inspector Model 800, and
the Cauchy coefi'rcients provided by the Shipley Company in the technical data sheet for
the photoresist. Samples for these experiments were prepared by spin coating the neat
photoresist at 3000 rpm onto quartz slides and by processing with a post application bake
for 60 seconds at 100°C, an exposure time for 60 seconds under a 200 W Hg(Xe) lamp
(Oriel Corporation, Stratford, CT), and a post exposure bake for 90 seconds at 100°C. A
second set of samples were prepared as controls to which all processing steps, save

exposure, were applied.

94

5.3. RESULTS AND DISCUSSION

5.3.1. Study of microscopic free volume changes in photoresist films using ground-
state recovery

Crystal violet is a member of the triphenylmethane dye family, with three
identical N,N-dimethylaniline rings bound to a central carbon atom. The structure of
crystal violet is presented in Figure 5.3. The predominant conformation of crystal violet
in solution is a propeller shape, having D3 symmetry, with each of the three phenyl rings
tilted slightly out of the molecular plane. The phenyl rings rotate around their bonds to
the central carbon atom in what is believed to be a barrierless process. Ring rotation
along the excited state potential energy surface provides access to efficient non-radiative
decay channels. Studies have indicated a strong dependence of the excited state lifetime
on solvent viscosity, which is indicative of essentially no intramolecular rotational barrier
to ring rotation.'°'” Crystal violet was chosen as a probe of the local environment in the
photoresist thin films under examination here because of its combined sensitivity to local
acid concentration and viscosity.

The recovery time of the ground-state population of crystal violet is measured
instead of its fluorescence lifetime because of the characteristically low fluorescence
quantum yield of this family of molecules. The ground-state recovery measurement is
equivalent to a fluorescence lifetime measurement for the triphenylmethanes. The
ground state recovery measurement is accomplished by detecting the transient change in
absorption of the probe pulse resulting from excitation by the pump laser pulse (Figure
5.4). The difference in arrival time at the sample between the pump pulse and the probe

pulse is controlled mechanically.

95

An exponential decay of the ground-state recovery signal for the spin coated,
doped resist film is observed, and these data are shown in Figure 5.5. A fit of the decay
to a sum of two exponentials results in a 25-ps time component and a lZO-ps time
component. The behavior observed is similar to fluorescence decay experiments using
malachite green, another triphenylmethane dye at similar temperatures in highly viscous
systems.”‘'3 A double exponential decay has also been observed for the ground state
absorption recovery behavior of malachite green in a solid SiOz-Zroz matrixM and in
polymethyl methacrylate (PMMA) and polyvinyl alcohol (PVA) films."-" Therefore,
the observation of two exponential decay components for a triphenylmethane dye is not
new, although there is an ongoing debate in the literature over its physical origin. None-
the-less, there is a well-established relationship between the values of both decay time
constants and the local viscosity of the medium, with the constants increasing with
increasing microviscosity. The values of the two time constants observed here are similar
to the time constants obtained by Ben-Amotz and Harris,10 Canva er al. '4 and Ippen,
Shank and Bergman.”

A linear calibration curve (quality of fit as reflected by the R2 value was 0.98) was
developed, shown in Figure 5.6, that relates the short time constant from the ground state
recovery time of crystal violet in a series of viscous liquid to the matrix viscosity. The
ground state recovery measurements were accomplished using the pump-probe laser
configuration modified with a recirculating flow loop to accommodate the viscous liquid
samples. The liquid was pumped through the coils of a temperature bath maintained at
300 K and into a 0.1.cm square flow cell. From the flow cell, the liquid entered a

reservoir before repeating the pump cycle. The samples contained 0.1 wt.% crystal

96

violet. The viscosities of ethylene glycol and propylene glycol were acquired from the
literature.‘8 The viscosities for the remaining data points were measured using a
Brookfield Model DV-I+ Viscometer (Brookfield Engineering Labs, Inc., Stoughton,
MA). Due to the highly viscous nature of the liquids, it was necessary to use the small
sample adapter with spindle #18 for the diols and spindle #34 for glycerol in order to
obtain accurate readings.

The resist before exposure and post exposure bake exhibits free-volume
characteristics similar to that of glycerol. A fit of the excited-state lifetime decay (a
representative plot of which is shown in Figure 5.7) to a sum of two exponentials results
in 25 i 4 ps for the short-time component (1:1) and 115 :1: 18 ps for the long-time
component (12). Thus, the effective free volume of the photoresist in this stage of the
processing is 141.0 cP. As free volume is created in the film during post exposure bake,
a decrease in the ground-state recovery times of crystal violet due to the increased
freedom available for its ring rotations is predicted. However, as that free volume
collapses due to polymer relaxation, the time constants will increase as the ring rotations
of the crystal violet embedded in the film are restricted.

5.3.2. Multi-wavelength interferometry

As mentioned at the start of this chapter, there is a change in film thickness
associated with the post exposure bake. During the deprotection reaction, which mainly
occurs during post exposure bake, two small molecules are created by the breakdown of
the TBOC pendant group that is cleaved by the photoacid. These gas molecules, namely

carbon dioxide and isobutylene, diffuse out of the film, thus creating free volume.

97

However, as the polymer chains relax, the film will collapse into itself and decrease the
overall film thickness and the available microscopic free volume.

In order to differentiate between solvent volatilization and polymer relaxation, the
film thickness behavior of the exposed photoresist samples was compared to those that
had not undergone exposure. There is a definite difference in shape between the two
thickness loss curves collected during the post (exposure) bake step, which confirms that
the mechanism by which film reduction occurs is related to the deprotection reactions
during post exposure bake. In the case of the control samples, which were not exposed,
the film thickness reduction is best described by a linear fit (Figure 5.8, top). This profile
is consistent with volatilization of solvent remaining in the film after the post application
bake. In contrast, the film thickness reduction of the exposed samples is best described
by a second-order exponential decay (Figure 5.8, bottom). This suggests that two
mechanisms are occurring during post exposure bake: relaxation of the polymer chains
facilitated by the evaporation of the gases produced during deprotection and volatilization
of any remaining solvent.

This explanation for the decrease in film thickness during post exposure bake is
further confirmed by examination of the diffusional properties for carbon dioxide and
isobutylene. The diffusion coefficient of the small molecules in the thin film can be

described using the Stokes-Einstein equation:l9

 

D = kT (5.2)
6m1r

where k is the Boltzmann constant, T is the absolute temperature, 11 is the viscosity of the
medium, and r is the radius of the diffusing molecule. For carbon dioxide, this radius is

0.165 nm; for isobutylene, this radius may be approximated using that of iso-butane,

98

which is 0.25 nm.20 Thus, given a bake temperature of 100°C and an effective viscosity

of 141 cP for the photoresist film initially, the diffusion coefficients for carbon dioxide
and isobutylene are 1.17x10‘7 cmZ/s and 7.8x10‘8 cmz/s, respectively. These diffusion
coefficients can then be used to estimate a characteristic diffusion time (1:) using the
following expression:

1} (5.3)

where L is the length for diffusion. In the case of the photoresist films, the longest
vertical diffusion length is one micron. Therefore, the diffusion time for these small
molecules is on the order of one second or less. With this time scale for diffusion, there
is adequate time during the post exposure bake for the polymer chains to relax and fill the
free volume created as the small molecules diffuse from the film.

In addition to the difference in film thickness loss profiles, there is a difference in
the total thickness lost during the bake times for the two sample sets. With the
unexposed samples, the thickness reduction observed was less than 1% of the total film
thickness measured at the beginning of the bake time. For the exposed samples, the
thickness reduction observed was greater than 3%. Thus, the greatest film thickness loss
during post exposure bake is due to the polymer chain relaxation.

Similar studies, conducted at Advanced Micro Devices (Sunnyvale, CA),
corroborate these results, as well as lend insight into the effect of bake time upon the
thickness loss profile. Resist was spin coated onto silicon wafers, and each wafer was
subdivided into a grid. Each grid section was exposed using a KrF excimer laser at a

different dose. The wafers were then subjected to different post exposure bake times, and

99

the film thickness of each grid section was measured thereafter. In Figure 5.9, these data
are shown for the Apex-E 2408 DUV photoresist at a bake temperature of 100°C. If the
film does not undergo post exposure bake, the activation energy for deprotection is not
realized; thus, no change in thickness is observed. At shorter post exposure bake times,
less thickness loss occurs regardless of the exposure dose because there is not enough
time for the photoacid to effect more deprotection events. As the post exposure bake
time is increased, less energy expenditure is necessary to effect a higher number of
deprotection reactions, as manifested by the steepness in the exponential curves. Similar
trends are observed for a second Shipley resist, UV5-0.6 positive DUV photoresist, at a
bake temperature of 130°C (Figure 5.10).
5.4. CONCLUSIONS

Spectroscopic techniques designed to characterize acid mobility in chemically
amplified resists during post exposure bake have been developed. To measure the free
volume changes during post exposure bake, ground-state recovery measurements using
the triphenylmethane dye crystal violet as a molecular probe were used. The lifetimes of
the excited state of crystal violet increase as the local free volume decreases and
molecular rotation is hindered. Secondly, dynamic film thickness measurements were
applied to understand the physical phenomena behind the thickness loss observed during
post exposure bake. The results indicate that the majority of the thickness loss is due to
the relaxation of polymer chains facilitated by the diffusion of gases out of the film afier
deprotection events take place and not due to volatilization of solvent. The further
application of these experimental techniques will provide a systematic study of free

volume evolution in situ during post exposure bake. For a given resist, a more thorough

100

understanding of the factors that affect free volume evolution and acid mobility (e. g., post
exposure bake temperature and duration) can allow improved microlithographic pattern

generation.

101

 

+ » t

‘ cm'tydnpeddyelasers
410rm-aDrm

 

 

 

 

 

Figure 5.1.

Experimental setup for ground state recovery experiments to determine
microscopic free volume in the photoresist films.

102

Bifurcated Fiber Optic Cable

  
    
 
 
    

Incident
Light

\

F llm of Thickness, T Reflected

Light

 

 

 

Figure 5.2. Experimental setup for dynamic film thickness measurements to observe
macroscopic free volume evolution.

103

9‘3
9 N—CH;

|.|

CH3

Figure 5.3. Molecular structure of crystal violet.

  
 
 
 

Before Pump

After Pump

absorbance, a.u.

 

 

wavelength, nm

Figure 5.4. Schematic representation of the transient change in ground state
absorption of crystal violet in response to excitation by a pump laser pulse.

104

‘L0

(18

O
O)

AVIV, a.u.
O
.5

0.2

0.0

Figure 5.5.

 

 

 

 

L o
1'1: 25 ps
,. 1 ‘ T2 120 p8
_ 5 . ,
1' 711‘
1 d"- l I
- T r 'Imru-I:Liu
I 4 l 1 I L I 1 l
0 100 200 300 400
time, ps

Instrumental response function (m) and experimental ground state
recovery data (—) for crystal violet in the Apex-E 2408 DUV photoresist

film.

105

25 - Glycerol I
20 "' .0 .
Propylene Glycol . - 'I 1.3 Butanedlol
u, b 1. "1:4 Butanediol
9- 15 — '
E:
s ..
P I. 1.3 Propanediol
10 - '
5 _ . ' . I Ethylene Glycol
1 I l I 4 I l I 1 l l I l I l I

 

 

0 20 4O 60 80 100 120 140 160

viscosity, cP

Figure 5.6. Linear calibration curve relating the short time constant, 1., from the
ground state recovery experiments of viscous liquids doped with 0.1 wt.%
crystal violet to the matrix viscosity.

106

AV/V, a.u.

1.0

0.8

0.6

0.4

0.2

0.0

Figure 5.7.

 

 

 

 

 

 

Instrumental response function (~--—) and experimental ground state

recovery data (—) for crystal violet in glycerol.

107

9200 -.

9150-

film thickness, A
Q
3
O
I

9050 -

 

unexposed film

 

 

 

 

 

sooofi.,,rfij
o 10 20 30

I V‘fi‘r' I 1 fr ‘fi

. ' 1
40 50 60 70 80 90 100

bake time. seconds

9300 a

9
9200 . o
9100 -

9000 -4

film thickness, A

8900 .-

 

- exposed film

 

 

 

 

 

eaoo , . ,
o 20

T I V 1 f I T

j
40 60 80 100

bake time. seconds

Figure 5.8. Change in Apex-E 2408 DUV photoresist film thickness as a function of
bake time at 100°C for unexposed (top) and exposed (bottom) films.

108

 

 

 

 

 

 

 

 

8100 d \-n “;‘||l|||l|'l"'lll'lIIIlIIHI'I”minimN.H.”..H”:”mm!“I ”Ump-
‘ ‘ cl 0 s PEB
800° ‘ e 10 s PEB
‘ v 25 s PEB
790° ' A 50 s P
4 EB
a; ‘ o 75 s PEB
8 7800 4 I 90 s PEB
5 .,
.9 7700 _
5 l
g 7600..
a ‘ —"Hugfim:'umtl'
7500 - lune‘utut lllll '
7400 ..
I V I V 1 U I 1 I I I ' I ' l '
0 20 40 60 80 100 120 140

exposure dose, mJ

Figure 5.9. Change in thickness of Apex-E 2408 DUV photoresist film as a function
of exposure dose and post exposure bake time.

109

7050 -

  
     
   
   

       
  

    

    
   
 

 

 

 

 

 

 

7000 - - l
. .o v) m “5;; . d
w w x‘ ”7.)"? v 9.0.“. c“... 0
695° - o a".
+ \ n 7 .
4 6900 - '1' 9 o
m" ‘
m 6850 .
d)
c a:
fl 6800 -
E . I O s PEB
E 6750« o 10sPEB , LA
:3 A 25 s PEB W
6700 J ‘b 57
. v 50 s PEB V
6650 _ O 80 s PEB V
. " o
6600 1— ' j I I T U ]
0 2 4 6 8

exposure dose, mJ

F igure,5.10. Change in thickness of UV5-0.6 positive DUV photoresist film as a
function of exposure dose and post exposure bake time.

110

10.

11.

5.5. REFERENCES

. Z. Tadmor and CG. Gogos, Principles of Polymer Processing, John Wiley & Sons,

Inc., New York, 1979.

N.H. Jakatdar, J. Bao, C.J. Spanos, R. Subramanian, B. Rangarajan, “Physical
modeling of deprotection-induced thickness loss,” Proceedings of SPIE, 3678, 275
(1999)

F .W. Wang, R.E. Lowry and W.H. Grant, “Novel excimer fluorescence method for
monitoring polymerization: l. Polymerization of methyl methacrylate,” Polymer,
25(5), 690 (1984).

8K. Moorjani, B. Rangarajan, and AB. Scranton, “Effect of viscosity on the rate of
photosensitization of diaryliodonium salts by anthracene,” Photopolymerization:
Fundamentals and Applications, edited by AB. Scranton, C.N. Bowman, and R.W.
Peiffer, American Chemical Society, Washington, DC, 1997.

J .G. Victor and J.M. Torkelson, “On Measuring the Distribution of Local Free
Volume in Glassy Polymers by Photochromic and Fluorescence Techniques,”
Macromolecules, 20, 2241 (1987).

Y. Jiang, S. A. Hambir, G. J. Blanchard, “Synchronous pumping of two dye lasers
using a single UV excitation source,” Opt. Commun., 99, 216 (1993).

P. Bado, S. B. Wilson, K. R. Wilson, “Multiple modulation for optical pump—probe
spectroscopy,” Rev. Sci. Instrum, 53, 706 (1982).

L. .Andor, A. Lorincz, J. Siemion, D. D. Smith. S. A. Rice, “Shot-noise-limited
detection scheme for two-beam laser spectroscopies,” Rev. Sci. Instrum, 55, 64
(1984).

G. J. Blanchard, M. J. Wirth, “Measurement of small absorbances by picosecond
pump-probe spectrometry,” Anal. Chem, 56, 532 (1986).

D. Ben-Amotz, C. B. Harris, “Torsional dynamics of molecules on barrierless
potentials in liquids. 1. Temperature and wavelength dependent picosecond studies of
u'iphenylmethane dyes,”.l. Chem. Phys, 86(9), 4856 (1987).

D. Ben-Amotz, C. B. Harris, “Torsional dynamics of molecules on barrierless
potentials in liquids. 11. Test of theoretical models,” J. Chem. Phys, 86(10), 5433
(1987).

111

12.

13.

14.

15.

16.

17.

l8.

19.

20.

J.Y. Ye, T. Hattori, H. Inouye, H. Ueta, H. Nakatsuka Y. Maruyama, M. Ishikawa,
“Glass transition of associated solvents studied by fluorescence measurement of
doped chromophores,” Phys. Rev. 8, 52(13), 8349 (1996).

J. Y. Ye, T. Hattori, H. Nakatsuka, Y. Maruyama, M. Ishikawa, “Microscopic
dynamics of the glass transition investigated by time-resolved fluorescence
measurements of doped chromophores,” Phys. Rev. 8., 56(9), 5286 (1997).

M. Canva, G. Le Saux, P. Georges, A. Brun, F. Chaput, J.P Boilot “Time-resolved
saturated absorption recovery in malachite green-doped xerogel,” Chem Phys. Lett..
176(5), 495 (1991).

K. M. Abedin, J. Y. Ye, H. Inouye, T. Hattori, H. Sumi, H. Nakatsuka, “Diffusive
torsional dynamics of malachite green molecules in solid matrices probed by
fluorescence decay,” J. Chem. Phys, 103(15), 6414 (1995).

K. M. Abedin, J. Y. Ye, H. Inouye, T. Hattori, H. Nakatsuka, “Local dynamics in
solid matrices investigated by malachite green optical microprobes,” J. of
Luminescence, 64, 135 (1995).

E. P. Ippen, C. V. Shank, A. Bergman, “Picosecond recovery dynamics of malachite
green,” Chem. Phys. Lett.. 38(3), 611 (1976).

Lange 's Handbook of Chemistry, 14th ed., edited by J.A. Dean, McGraw-Hill Book
Company, New York, 1992.

J .E. Guillet, “Mass Diffusion in Solid Polymers,” Photophysical and Photochemical
Tools in Polymer Science, edited by M.A. Winnik, D. Reidel Publishing Company,
Hingharn, MA, 1986.

L.H. Sperling, Introduction to Physical Polymer Science, 2“‘1 ed., John Wiley & Sons,
Inc., New York, 1992.

112

Chapter 6

CONCLUSIONS AND RECOMMENDATIONS

6.1.SUMMARY OF RESULTS

In this body of research, spectroscopic methods have been developed to gain in-
situ and on-line information for two different polymer systems: thermoplastic polymer
composites and chemically amplified resists. With these techniques, it is possible to
conduct quality control and process optimization efforts during actual polymer
processing. Spectroscopic techniques such as fluorescence and absorption are attractive
for in-situ and on-line characterization because pertinent processing information can be
gathered in a quick and timely manner.
6.1.1. Cure monitoring of polymer composites

First of all, a method was developed to monitor polymer cure based upon the
fluorescence response of a specially chosen probe molecule.l The use of an optical
spectrometer to implement this method provides an inexpensive, non-intrusive, and
portable means of cure monitoring.2 This method is based on the introduction of trace
quantities of a solvatochromic probe into the system. These molecules do not lose
sensitivity at the gel point and can be implemented easily using optical fibers. Two such
molecules are phenoxazone 660 and pyrene, whose fluorescence spectra change in
response to the conversion of double to single bonds during the polymerization. In the
case of the phenoxazone 660, the emission spectra blue-shifts with increasing polymer
conversion; in the case of pyrene, the relative intensities of two interdependent peaks in

the emission spectra invert. The fluorescence cure monitoring scheme using these probe

113

molecules was successfully demonstrated in a variety of curing polymers, including
methyl methacrylate, styrene and Derakane” vinyl ester resins.
6.1.2. Characterization of chemically amplified photoresists

Secondly, highly sensitive, time-resolved spectroscopic techniques were
developed to address two basic, yet significant, questions for reaction-diffusion
phenomenon in chemically amplified resists: (i) how much acid is in the system and (ii)
where is the acid going. 'This was an extremely challenging task because the photoresist
films are typically one-micron thick, making it difficult to detect optical signals during
processing. These difficulties were in part overcome by carefully selecting probe
molecules that would allow inquiry into the status of acid concentration and free volume.

To answer the first question, focus was placed on two stages in the resist
processing: exposure and post exposure bake. Steady-state fluorescence of the acid-
sensitive molecular probe fluorescein was used to measure the photoacid generation
during exposure. With this technique, the concentration of acid was characterized as a
first order exponential in response to illumination with 248-nm light. Steady-state
absorption of a second acid-sensitive molecular probe, crystal violet, was used to measure
proton concentration during post exposure bake. This technique allows the quantification
of acid concentration throughout the deprotection reaction regime based on the alpha
charts for three possible states of crystal violet.

In answering the second question, focus was placed on the development of free
volume during the post exposure bake and its effect upon acid mobility; this was
investigated on two different physical scales: macroscopic and microscopic. Dynamic

film thickness measurements based on multi-wavelength interferometry confirm the

114

difference in thickness reduction mechanisms between solvent volatilization and polymer
chain relaxation during post exposure bake. The major film thickness loss occurs as the
polymer chains relax, causing the film to collapse into itself and decrease the overall free
volume provided by the small molecules, created during the deprotection reaction, which
quickly diffuse out of the film. The nature of the free volume evolution on a microscopic
level was probed using ground state recovery experiments with the molecular probe
crystal violet, which also exhibits a free volume-sensitive behavior in its transient
response. This technique allows the correlation of the resist microenvironment to an
effective viscosity. As this viscosity increases, the free volume available for acid
diffusion decreases.
6.2. SUGGESTIONS FOR FUTURE WORK

With this foundation laid, further research is recommended in order to apply fully
the techniques developed in this body of work. Specifically, quantitative correlation
studies and industrial application strategies are suggested in furthering the work
completed on cure monitoring of polymer composites. In the area of characterization of
chemically amplified resists, the extension of the spectroscopic techniques to quantify
acid loss and free volume evolution is recommended, as well as application of the
techniques to forthcoming l93-nm resist technology.
6.2.1. Cure monitoring of polymer composites

Although the thermocouple studies show that the fluorescence emission of the
probe molecules does indeed follow the polymerization reaction, it would be advisable to
correlate the results with well-accepted, quantitative techniques such as differential

scanning calorimetry and/or Raman spectroscopy. In differential scanning calorimetry,

115

the overall heat of reaction for the polymerization is measured at a constant temperature
(see Figure 6.1, top). By integrating the area under the resulting curve (when the
exothermic peak are pointed up), the conversion over time is obtained (see Figure 6.1,
bottom). In order for this measurement to have quantitative significance, the heat of
polymerization must be known for the polymer under investigation, a number of which
are reported by Sawada.3 However, the qualitative shape of the conversion curve is
adequate to compare the fluorescence response of the molecular probe to the
polymerization in the case of those monomers, such as the Derakane” resins, for which
this information is not readily available. In order to accomplish these experiments, a
special adapter for the Perkin-Elmer DSC 7 (Norwalk, CT) must be constructed so that the
fluorescence monitoring technique may be applied simultaneously. Basically, the adapter
should consist of a quartz window with an SMA connection for an optical fiber, which
would be placed over the sample pan. Using a bifurcated optical fiber, the excitation
light can be directed onto the sample while the fluorescence emission is collected and
sent to the spectrometer. Some considerations for the sample include: (i) a high
concentration of the molecular probe (say, 10'3 wt.%) to enhance the fluorescence signal,
(ii) the largest sample size permitted in the pan (namely, 30 mg) to prevent the sample
from concentrating around the edges of the pan out of the sampling area of the optical
fiber set-up, and (iii) a less volatile monomer, such as one of the more viscous Derakaneo
resins, to discourage evaporation of the monomer during the experiment.

A second possible method of verification is Raman spectroscopy, which is based
upon a frequency shift of inelastic scattered light. Thus, it is well suited to observe

polymerization reactions in which there is a conversion from double to single bonds.‘

116

This technique has been used to follow conversion of many polymer systems, including
acrylates,s styrenef’7 vinyl ethers,"’9 and epoxies.lo Two sets of samples would be
necessary for these experiments: one set containing only the monomer and thermal
initiator for Raman spectroscopy and the other set with the molecular probe added for
fluorescence spectroscopy. It is advisable to conduct the experiments in this manner,
separating the real-time collection of Raman spectra from that of fluorescence spectra,
because the fluorescence from the molecular probes may overwhelm the Raman signal.
Finally, since preliminary studies have shown that the excitation laser beam may
photodegrade the molecular probe, it is important that the intensity of the beam be
reduced with neutral density filters to alleviate the problem.

In addition to comparing polymer conversion directly to the fluorescence intensity
ratios, it is recommended that specific control strategies involving this technique be
developed. Integration of this method into various industrial processing scenarios
presents a special challenge. On the one hand, there is the high-speed web processing
format in which thin films are applied and photopolymerized. This is in contrast to the
traditional molding process in which thick parts of varying geometry are thermally cured.
In the former case, algorithms must be developed to correlate the position of the sample
with the signal obtained so that corrective action may be taken on the appropriate section
of film. In the latter case, the concern is sampling port placement so that areas in which
cure is problematic are monitored appropriately.

Finally, although this method is applicable with glass fiber-filled composites, no
work has been done with carbon fiber-filled or pigmented systems. These are

challenging systems in that the fibers and pigments may absorb or reflect the monitoring

117

signals. An investigative study in this area to tailor the monitoring method to these
systems is suggested, since it would allow the technique to encompass a broader range of
industrial processing situations.
6.2.2. Characterization of chemically amplified resists
6.2.2.1. 248-nm resisb

The ultimate goal of research in this area is to obtain accurate physical parameters
for the reaction and diffusion in chemically amplified photoresists, which may be
incorporated into a model in order to predict pattern mask outcomes. The acid
concentration calibration curves for crystal violet absorption allow a great flexibility in
examining various aspects of the photoacid diffusional behavior. First of all, this
technique can be applied to the systematic study of acid loss due to airborne bases and/or
evaporation. This is accomplished by the reversibility of the protonated states of crystal
violet. Secondly, the diffusion times of the acid can be determined via spatial resolution
of this method. As a portion of the photoresist sample is exposed, a second portion a
known distance away may be monitored for acid incursion. If the photoresist
components are combined in the lab, rather than using a commercial formulation, then
the effect of the acid counterion on acid diffusion may be assessed by varying the
photoacid generator. Finally, as suggested by the fluorescence work in Chapter 4, a
similar set of calibration curves may be developed for the fluorescence emission response
of crystal violet in order to work with photoresist samples on silicon (rather than quartz)

substrates.

118

The relationship between the viscosity extracted from ground state recovery data
of crystal violet (n) and the translational diffusion coefficient of a diffusing moiety was
first introduced in the previous chapter:

kT (5.2)

D =
61mr

 

Here, the (spherical) diffusing entity under consideration is a proton with its associated
counterion. The relationship between D and the free volume of the polymer matrix has

been treated by Ganesh et al. ” and is given below as Equation 6.2.

‘ . (6.2)
131 = Do, {e‘E’RT I{exp[-V—y—[E]Vu +g52v2,]]}
FH
Q = 1711M” (6'3)

VZJMZJ

VF” is the average hole free volume per gram of mixed polymer, V” is the specific
critical hole free volume of species i required for diffusion to occur, Q (Equation 6.3) is
the ratio of the critical molar volume of the proton in the polymer matrix, 7 is the overlap
factor which accounts for cooperative nature of the diffusing proton and the surrounding
matrix, M” is the molecular weight of the diffusing species, M2; is the molecular weight
of a polymer matrix segment that can occupy the same space as the diffusing moiety, in.
and to; are the weight fiactions of the diffusing species in the matrix and the matrix itself,
E is the activation energy for diffusive motion, and Do. is an effective Arrhenius

prefactor. Thus, through this analysis of the ground state recovery data, it is possible to

correlate the microscopic evolution of free volume in the photoresist as a function of

119

 

processing time, which in turn will allow insight into the diffusional cross-section
available to the photoacid.
6.2.2.2. 193-nm resists

With the recent advent of the l93-nm resist films on the commercial market,
microelectronics companies are pressed to understand the correlation between acid
diffusion in these photoresists and resulting line defects, such as line end shortening and
iso-dense bias, before full-scale production begins in the next few years. These
processing problems will be more prominent than in the 248-nm resists since the shorter
exposure wavelength will enable smaller feature sizes and more compact devices. In
addition, the resist formulation has been significantly altered to obtain a transparent
spectral window at 193 nm by removing the benzene rings, yet retaining the high carbon
to hydrogen ratio with unsaturated ring systems to obtain acceptable etch rates (see
Figure 6.2).12 Thus, optical and kinetic characterization studies must be completed in
order to optimize use in microlithographic processing.

ln-situ and on-line characterization can be used to address these imminent
concerns with the l93-nm photoresists. Absorption and FTIR spectroscopies can be
adapted to determine the kinetics of chemical amplification in these photoresist films on
the wafer. Fluorescence and absorption spectroscopies used in conjunction with
judiciously chosen probe molecules can track photoacid concentration in the resist film
during processing. Finally, ground-state recovery measurements can be applied to
monitor the changes in film viscosity throughout the baking cycles and correlated to acid

diffusion.

120

70-

604

heat flow. ng
‘6’
1

 

 

 

V I

V I V

r I
120 140 160

U I V I T Tfi I V

I
0 20 40 60 so 100
time. minutes

 

energy released, mW-min

 

 

V I T I V ' f r f T V

r
0 20 4O 80 80 100 120 140 160

V I TV 1 fi

time, minutes

Figure 6.1. Differential scanning calorimetry plots for polymerization of Derakane”
411-C50 at 70°C: heat of reaction (top) and conversion (bottom).

121

O\O/O 0\0/0
a “A m
e 2 wk
OH ' O
0 0 0M0

Figure 6.2. Chemical structure of the latest l93-nm resist formulation.

122

10.

ll.

12.

6.3. References

G.J. Blanchard, J.L. Jessop, A.B. Scranton, “Method and apparatus for in situ, non-
invasive polymer cure determination,” US Patent #5633313, May 27, 1997.

G]. Blanchard, J .L. Jessop, A.B. Scranton, “Apparatus for in situ, non-invasive
polymer cure determination,” US Patent #5707587, January 13, 1998.

H. Sawada, “Thermodynamics of Polymerization. I,” J. Macromol. Sci—Revs
Macromol. Chem, C3(2), 313 (1969).

H.G.M. Edwards, A.F. Johnson and LR. Lewis, “Applications of Raman
Spectroscopy to the Study of Polymers and Polymerization Processes, " J. Roman
Spectrosc., 49A, 475 (1993).

J. Clarkson, S.M. Mason and KP]. Williams, “Bulk radical homo-polymerisation
studies of commerial acrylate monomers using near infrared Fourier transform Raman
spectroscopy,” Spectrochimica Acta, 47A, 1345 (1991).

E. Gulari, K. McKeigue and K.Y.S. Ng, “Raman and FTIR Spectroscopy of
Polymerization: Bulk Polymerization of Methyl Methacrylate and Styrene,”
Macromolecules, 17, 1822 (1984).

B. Chu, G. Fytas and G. Zalczer, “Study of Thermal Polymerization of Styrene by
Raman Scattering,” Macromolecules, 14, 395 (1981).

E.W. Nelson and AB. Scranton, “Kinetics of Cationic Photopolymerizations of
Divinyl Ethers Characterized Using In Situ Raman Spectroscopy,” J. Polym. Sci:
Part A: Polym. Chem., 34, 403 (1996).

E.W. Nelson and AB. Scranton, “In Situ Rarnan Spectroscopy for Cure Monitoring
of Cationic Photopolymerizations of Divinyl Ethers,” J. Raman Spectrosc., 27, 137
(1996).

KB. Chike, M.L. Myrick, R.E. Lyon, S.M. Angel, “Raman and near-infrared studies
of an epoxy resin,” Appl. Spectrosc., 47(10), 1631 (1993).

K. Ganesh, R. Nagarajan and J .L. Duda, “Rate of Gas Transport in Glassy Polymers:
A Free Volume Based Predictive Model,” Ind. Eng. Chem. Res, 31, 746 (1992).

CG. Willson, Resists for Deep UV Lithography, SPIE’s International Symposium on
Microlithography, March 18, 1999.

123

APPENDIX

124

A.1. DETERMINATION OF EQUILIBRIUM CONSTANT S
In order to develop a calibration curve that can be used to determine the acid
concentration in the photoresist thin films, the effect of acid concentration upon the
absorption spectrum of crystal violet was obtained using a buffered solution series (see
Figure A.l). From these spectral data, it was possible to calculate the concentration of
the three crystal violet species (cv = [cv*], cc, = [Hcv’j and Cy = [Hzcv3*]) in solution

using the application of Beer’s Law to mixtures:l

A430 = 8V(430)b°v + 80(430)b°o + 8“430)va (A-I)
A630 = evlo30)bcv + 30(630)b°o + 3Y(630)b°v (A2)
A590 = 8vr590)b¢v + 50(590)b¢o 4' 3Y(590)b°v (A3)

where, for a given acid concentration (or pH), A) is the absorption at wavelength j, Sjj is
the extinction coefficient of species i at wavelength j, b is the pathlength, and c, is the
concentration of species i. The three wavelengths (430, 590 and 630 nm) were chosen
because they represent the wavelengths of maximum peak absorption for the three
species of crystal violet under consideration: yellow, violet and green, respectively. The
extinction coefficients for these wavelengths and species were obtained from the
literature.2 Extinction coefiicients for the violet and yellow forms at 430 nm were
extrapolated from these data (see Figure A.2).

Equations A.l - A.3 were written in matrix form, as shown below, and solved by

Gauss reduction to obtain the three species concentrations at each acid concentration:

125

7.14x102 0 2.83x10‘ A430 (M)
1.62x10‘ 1.12x105 0 = A630
9.32xlo‘ 5.6x10‘ 2.8)thl A590

The equilibrium constants that relate the concentrations of the three crystal violet species
were then determined using the equations described in Chapter 4 and the species

concentrations calculated from above:

 

K _[HCV2*][H+] (4.3)
- 3+
[H2CV ]

K _lCV*liH*l (t4)
2- 2+
[HCV ]

The reported equilibrium constants are the average of these calculated values for the nine
acid concentrations evaluated.
A.2. DETERMINATION OF ALPHA PLOTS
From these equilibria data, it is then possible to construct alpha plots, which
enable the determination of acid concentration based on the concentration distribution of

the three crystal violet species. The following set of equations was solved algebraically

to obtain the alpha values (at), a. and (12) for this system:l

 

[Hcv2+ ][H”] (4.3)
K = = 0.025 mole
[CV*][H*] (4.4)
K = —— = 0.0067 moi
[HCV2+] %
[CV]ml = [CV*]+[HCV2+]+[H2CV3+] (A5)

126

= [H2CV3+] (A.6)

[CVItotal
a _ [HCV2+] (A.7)
‘ [Ci/1m
[cvt] (A8)
(12 = ——
[CVItotal
(10 H1, +a2 =1 (A.9)

Substitution and rearrangement of the seven equations above yields the following
expressions for the alpha values:

a __ K,[H*] (A.lo)
" [tr]2 +K,K, +K,[H*]

 

a _ K,K, (A.ll)
’ [H*]2+K,K2+K,[H*]

 

r10 =l—aI --a2 (A.lZ)

Figure A.3 shows the relationship among the three species of crystal violet as a function
of acid concentration based upon computation of the alpha values. Thus, with an
absorption spectrum of crystal violet in the photoresist film, it is possible to calculate the
individual concentrations of the three crystal violet species and determine the acid

concentration in the film using the alpha plot.

127

 

 

 

 

 

 

 

 

 

 

 

 

 

 

0.36- HCV”
. —pl-l . 0.26 f;
0.30. --- prise.“ ;' ‘1
: '1
.1 ..... pH 8 1.05 :’ .I.\..'.
--—-- H: 1.23 ,I I t:
. 0.241 p ’c '- 0“:
q - ----- pH 81.52 .' i "I.
G '1 I! I'3'".
o 0.18-1 . .’ I .' 1.1.1.
g Hzcv° '.-' I! :13
to 1 - . - .1;
‘9 2!:
o 0.12‘ '
to
.D 11
Q
0.06-
0.00-1
firfirfi'r'r'rrrfr'i
300 350 400 450 500 550 600 650 700 750
wavelength. nm
0.60- CV’
0,55:1 —pH s1.a1 ,3
050; --- pH = 2.25 its,
0.5; ----- vH'i-os ‘t
1 -.-.- H s 5.12 F 2’
. “0‘ p j: lHCV
3. - 1'
6 0.30.
g 1
m 0.25:
‘3 0.20.
3 .
0.10‘
0.05-
0.00.
'o'osrr'rrr'r'r'rrr'ffr
300 350 400 450 500 550 600 650 700 750
wavelength. nm
Figure A.l. Change in absorption spectrum of crystal violet with increasing acid

concentration in buffered solution series.

128

 

 

 

 

 

 

 

 

 

 

 

100000 .
violet I
to 00000 .t '
. I .
a
c d
.2 I
,2 60000 -1 .
*
o .
8
40000 ..
g extrapolated c p .
‘13 ‘ 5
.g 20000 _ 3!
x . I
o , J:
0 q . ...... 1 ' I
' I 1— T ' I ‘ I j T ' I
400 450 500 550 600 650 700
wavelength, nm
30000 1
. I.
25000 . -\
w. . extrapolated 1:
5 20000 . '.
'6 9
E l -.
a, .
0 15000 - .
o q i
C .
.3 10000 -
U ‘.
.E ' I.
32' 5000 . '. Yellow
0: I'.
I1 '.'.
0 -1 ' ' I I I I I I I I
' T fl T m 1 fl 1 r u ' 1
400 450 500 550 600 650 700

wavelength, nm

Figure A.2. Extrapolation of extinction coefficients for CV+ (violet species, top) and
HzCV3+ (yellow species, bottom) based upon reported literature values.

129

 

a fraction

 

 

 

 

Figure A.3. Change in relative amounts of CV+ (a2), HCV” (0n) and 1-12CV3+ (do)
with decreasing acid concentration.

130

A.3. REFERENCES

. D.A. Skoog, D.M. West and F .J . Holler, Analytical Chemistry: An Introduction, 5th
ed., Saunders College Publishing, Chicago, 1990.

. J .B. Conant and TH. Werner, “The determination of the strength of weak bases and
pseudo bases in glacial acetic acid solutions,” J. Am. Chem. Soc, 52, 4436 (1930).

131