close

Вход

Забыли?

вход по аккаунту

?

Resonant laser ablation: A mechanistic study of a novel solid sampling technique with detection by microwave-induced plasma atomic emission spectrometry

код для вставкиСкачать
R eso n a n t L a se r A b l a t i o n : A M e c h a n istic S tu dy
S o l id S a m plin g T ec h n iq ue
w ith
D e t e c t io n
by
of a
N ovel
M icrow ave
I n duced P la sm a A tom ic E m issio n S pec tr o m etr y
Peter Stchur III, Ph.D.
University of Connecticut, 2003
Laser ablation refers to the explosive process by which a solid sample is
vaporized though a violent laser-material interaction resulting in a plume of atoms, ions,
and molecules. It has been demonstrated that an increase in sensitivity and selectivity
of an analyte can be achieved by tuning the laser wavelength to match that of a resonant
gas phase transition of the analyte. Tuning the ablation laser to such a wavelength has
been termed Resonant Laser Ablation (RLA). Chapter 1 contains a review of the current
theory and applications o f RLA.
The analytical utility of RLA is hindered by the lack of understanding o f the
mechanism by which this enhancement occurs and an uncontrollable ablation process.
Chapter 2 outlines both resonant and non-resonant interactions in both the Chemistry
and Physics literature lending insight into factors that might have been overlooked in
the description of the currently accepted theory of RLA.
The data presented in Chapter 3 demonstrates the enhancement effect of trace
metals in stainless steel samples, and pure copper and aluminum samples in both
ablation wavelength scans and emission scans. Optical and SEM images illustrate
changes in surface morphology as the laser wavelength approaches the resonant
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
Peter Stchur III - University of Connecticut, 2003
wavelength o f the target material. Based on this date, a mechanism is presented in
Chapter 4 to account for the increase in atomic signal, as well as a distinct laser-material
interaction.
Chapter 5 explores a wide range of applications ofN ear Infrared Spectroscopy
(MRS) in various fields o f analytical chemistry, taking advantage of its inherent rapid
analysis time and minimal sample preparation. A novel PbS spectrometer is also
characterized using a variety of samples demonstrating its usefulness as an analytical
tool and also it’s shortcomings and limitations.
Appendix A outlines detailed experiments for the coupling of RLA with an
Inductively Coupled Mass Spectrometer (ICP-MS) through the use o f a novel cell
design. Appendix B and C outlines the RLA apparatus alignment and basic operating
principles o f the Optical Parametric Oscillator (OPO) laser, respectively. Appendix D
describes the use of computer-interfaced stepper motor for sample and attenuator
control.
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
R e s o n a n t L a s e r A b l a t io n : A M e c h a n is t ic S t u d y
S a m p l i n g T e c h n iq u e
w it h
D e t e c t io n
by
of a
N o v e l S o l id
M ic r o w a v e I n d u c e d P l a s m a
A t o m ic E m i s s io n S p e c t r o m e t r y
Peter Stchur III
B.S., Wilkes University, 1997
A Dissertation
Submitted in Partial for Fulfillment of the
Requirements for the Degree of
Doctor of Philosophy
at
University of Connecticut
2003
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
UMI Num ber: 3 1 0 1 7 1 5
UMI
UMI Microform 3101715
Copyright 2003 by ProQuest Information and Learning Company.
All rights reserved. This microform edition is protected against
unauthorized copying under Title 17, United States Code.
ProQuest Information and Learning Company
300 North Zeeb Road
P.O. Box 1346
Ann Arbor, Ml 48106-1346
Reproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
APPROVAL PAGE
Doctor of Philosophy Dissertation
Resonant Laser Ablation: A Mechanistic Study of a Novel Solid Sampling Technique
with Detection by Microwave Induced Plasma Atomic Emission Spectrometry
Presented By
Peter Stchur III, B.S.
Major Advisor _ _
Robert G. Michel
Associate Advisor
Robert K. Bohn
Associate Advisor
BrendaR. Shaw
University of Connecticut
2003
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
DEDICATION
To my wife Mary Anne
Ja te volim vise
iii
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
ACKNOWLEDGEMENTS
I would like to thank my advisor, Dr. Robert G. Michel, for his outstanding
guidance and support throughout my graduate career at the University of Connecticut.
He has always challenged and encouraged me to do my best work and for that I am
immensely grateful. I would also like to thank my associate advisors, Dr. Robert K.
'Springsteen.
Special thanks to "Machine Shop" Dave Osier for his expertise, creativity, sense
of humor, reliability, and friendship, which has allowed me to work more efficiently.
I express my appreciation to my former and present coworkers, Dr. Jack X.
Zhou, Dr. Karl X. Yang, Dr. Xiandeng Dan Hou, Dr. Suh-Jen Jane Tsai, Heather
Molyneux, Jeremy Koscielecki, Danielle Cleveland, and Shrimati Balram for their
support, ideas, and friendship.
I extend my gratitude to the Department of Chemistry for the financial support
during my graduate career. Also, many thanks Mark Marotte and to the office staff
including Joyce Rodriguez, Ann Delmastro, Wanda Hicks, Charlene Fuller, Emilie
Hogrebe, Dianne Tillman and Nancy Coogan.
I am extremely grateful for the support, understanding, and love from my
parents, brother and sisters, and especially my wife, Mary Anne.
iv
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
TABLE OF CONTENTS
1.
GENERAL IN TRO D U CTIO N .........................................
1.1
1.2
1.3
1.4
1.5
1.6
In t r o d u c t io n ..................................................................................................
R e s o n a n t L a s e r A b l a t io n ...................................................................................................... 3
M e t a l s ........................................................................................................
M e t a l O x id e s ................................................................................................................................ 10
S e m ic o n d u c t o r s ....................................................................................................................
12
....................................................................
O t h e r R L A A p p l ic a t io n s ...
1.6.1 Resonant Laser Induced Breakdown Spectroscopy ...............................................14
1.6.2 Selective Ion Source............................................................................................................ 15
1.7 C o n c l u s i o n
........................................................................
2.
1
14
A MECHANISTIC STUDY OF RESONANT LASER ABLATION.............. 18
2.1
2.2
2.3
2 .4
In t r o d u c t io n .................................................................................................................................18
L a s e r A b l a t io n v s . L a s e r I n d u c e d D e s o r p t io n ...................................................... 19
R e s o n a n t L a s e r A b l a t io n T h e o r y .................................................................................. 23
P r e c e d e n t s f o r L ID a n d L A ................................................................................................ 32
2.4.1 Surface Defects
......................................................................................................... 32
2.4.2 Polarons andExcitons...................................................................................................... 33
2.4.3 Surface Plasmon Resonance
.................................................................................... 35
2.5 C u r r e n t L ID M e c h a n is m s ............................................................................
2.5.1 Menzel-Gomer-Redhead Model..................................................................................... 36
2.5.2 Knotek-Feibelman M odel .................................................................................................39
2.5.3 Antoniewicz M odel
......................................................................................................39
2.6 R e s o n a n t R a d i a t i o n T r a p p in g ........................................................................................... 4 2
2 .7 C o n c l u s i o n .................................................................................................................................... 45
36
3. RESONANT LASER ABLATION MICROWAVE INDUCED PLASMA
(RLA-MIP) PROVIDES INCREASED ELEMENTAL DETECTION OVER
CONVENTIONAL LASER ABLATION..............................................
In t r o d u c t io n
..............................................................................
E x p e r im e n t a l ...............................................................................................
R e s u l t s a n d D is c u s s i o n ..........................................................................................................60
3.3.1 Optical and SEM Crater Images ................................................................................... 60
3.3.2 Spectral scans ........................................................................................................................ 66
3.3.3 Delay Time...............................................................................................................................66
3.3.4 Emission Spectra .................................................................................................................. 71
3.3.4.1 Pure A lu m in u m E m issio n S c a n s ....................................................
3 .3 .4 .2 M olyb d en u m in a N IS T S tain less S teel S a m p le
......................
3.3.5 Wavelength Ablation Scans ...............................................
79
3.3.6 Sample Angle and Polarization ............................................................
3 .4 M e c h a n is m s of R e s o n a n t L a s e r A b l a t io n ..........................................................
94
3.5 C o n c l u s i o n .....................................................................................................................................96
47
3.1
3 .2
3.3
v
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
47
50
73
75
85
4.
R E S O N A N T L A S E R A B L A T IO N - P A S T A N D P R E S E N T
4.1
4 .2
4.3
4 .4
4 .5
4 .6
5.
...........
98
In t r o d u c t io n
................ .............. ..............................................................
98
R L A M e c h a n is m s P r o p o se d b y Y a n g ...........................................................................98
R L A M e c h a n is m s P r o p o s e d b y H o u
...................... .................................. .
101
R L A M e c h a n is m s B a s e d o n P r e s e n t W o r k .............................................................. 105
P r o p o s e d M e c h a n is m B a s e d o n C u r r e n t W o r k ....................................................108
C o n c l u s i o n ..................................................................................................................................117
A REVIEW OF RECENT APPLICATIONS OF NEAR INFRARED
S P E C T R O S C O P Y , A N D O F T H E C H A R A C T E R IS T IC S O F A N O V E L P B S
C C D A R R A Y -B A S E D N E A R -I N F R A R E D S P E C T R O M E T E R ..................................1 1 9
5.1
5.2
5.3
5 .4
In t r o d u c t io n ................................................................................
119
C h e m o m e t r ic s .......................................... ............................................................................. 121
.............................................
126
In s t r u m e n t a t io n
L it e r a t u r e N IR A p p l ic a t io n s
.......................
129
5.4.1 Skin Applications ............................................................................................................. 129
5.4.2 Plastics Identification ....................................................................................................133
5.4.3 Hydrocarbon Samples ...................................................................................................138
5.4.4 Agricultural Applications ............................................................................................ 140
5 .4 .4 .1 Fruit s a m p le s ................................................................................................................. 140
5 .4 .4 .2 M eat s a m p le s ...................
145
5 .4 .4 .3 M ilk s a m p le s ...............................................................
146
5 .4 .4 .4 M isce lla n eo u s agricultural sa m p les....................
148
5.4.5 Pharmaceutical Applications ....................................
150
5.4.6 Biological Applications ................................................................................................ 156
5.5 C h a r a c t e r is t ic s o f a n o v e l P b S C C D a r r a y - b a s e d n e a r in f r a r e d
s pe c t r o m e t e r ........................................................................................................................................ 159
5 .5 .1
Experimental D esign ..................................................................................................... 159
5.5.2 Samples .................................................................................................................................166
5.5.2.1 M oistu re con tent in s k in ............................................................................................166
5 .5 .2 .2 W ater con tent in m e lo n s ............................................................
169
5 .5 .2 .3 P lastics id en tification ................................................
175
5.5.3 Discussion ........................................................................................................................... 180
5.5.4 Conclusion ..........................................................................
182
A P P E N D IX A .
R E S O N A N T L A S E R A B L A T I O N W I T H I N D U C T IV E L Y
C O U P L E D M A S S S P E C T R O M E T E R D E T E C T IO N U S IN G A N O V E L C E L L
D E S I G N ......................................................................................................................................................... 183
A. 1
A .2
A.3
A .4
A .5
A .6
A .7
In t r o d u c t io n .............................................................................................................................. 183
N o v e l R e s o n a n t L a s e r A b l a t io n IC P -M S C e l l D e s i g n ................................... 186
E ff ic ie n c y T e s t o f N o v e l R L A -IC P -M S C e l l D e s ig n ..........................................192
R e s o n a n t E ffe c t in L A -IC P -M S
............................................................................... 195
R L A -IC P -M S C a l ib r a t io n
......................................................................................196
Q u a n t it a t iv e M u l t i -E l e m e n t A n a l y s i s
......
201
C o n c l u s i o n .............................
202
vi
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
APPENDIX B.
RESONANT LASER ABLATION APPARATUS
ALIGNMENT.................................
B .l In t r o d u c t io n
........................
B .2 L a s e r A l i g n m e n t ...................
B .3 MEP E m is s io n A l i g n m e n t
....................
APPENDIX C.
BASIC OPO OPERATION
203
20 3
205
212
....................................
216
C .l I n t r o d u c t io n ............................................................................................................................. 2 1 6
C .2 W r it in g T a b l e s
...........................
217
C .3 B e a m S t e e r in g ...........................................................................................................................221
APPENDIX D.
D .l
D .2
D .3
D .4
COMPUTER-CONTROLLED STEPPER MOTOR
222
I n t r o d u c t io n ...............................
22 2
S o f t w a r e ...................................................................................................................................... 2 2 4
.............................................
226
A ttenuator C o n tr o l
S a m p l e H o l d e r R o t a t i o n ...................................................................................................22 8
REFERENCES
..................
229
vii
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
LIST OF TABLES
Table
2.1 A p p l ic a t io n s
Table
3.1: T he c o n st it u e n t s
of
RLA a n d
of
th eir sig n if ic a n c e ........................................................ 25
NIST SSS D845t
........56
3.2: M a j o r c o m p o n e n t s o f t h e e x p e r im e n t a l a r r a n g e m e n t f o r RLA-MIPAES...............................................
57
T a b le
Table
D. 1: S w it c h in g
s e q u e n c e fo r fo u r - p h a s e st e ppe r m o t o r .
v iii
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
..................... 223
LIST OF FIGURES
F ig u r e 2.1: A sc h e m a t ic of RLA in w h ich th e l e a d in g e d g e o f th e p u l s e ( a )
PRODUCES ATOMS, IONS AND PARTICLES, WHILE THE TRAILING EDGE OF THE PULSE
(B) RESONANTLY IONIZES THE ABLATED MATERIAL. BASED ON THE THEORY OF
29
W a t a n a b e e t a l . [3 6 ]......................
F ig u r e 2.2: S c h e m a t ic d ia g r a m o f th e v a p o r iz a t io n
W a t a n a b e e t a l . [3 6 ]....................
F ig u r e 2.3:
MGR m o d e l d e s c r ib in g
p r o c e s s a s e x p l a in e d b y
30
l a s e r in d u c e d d e s o r p t io n
[4 5 ,4 6 ].........................38
F ig u r e 2.4: A n t o n ie w ic z m o d e l d e s c r ib in g l a s e r in d u c e d d e s o r p t io n o f a n ato m
AND POSSIBLY A POSITIVE ION, WHERE M IS THE SUBSTRATE AND A IS THE
a d s o r b a t e . H e r e , a n in c id e n t ph o t o n b e c o m e s in s t a n t l y po sit iv e l y
IONIZED; THUS THE GROUND-STATE CONFIGURATION IS PROMOTED TO AN EXCITED
IONIC CURVE (M+A+)*. AS THE ION IS ATTRACTED TOWARD THE SURFACE, IT MAY
PICK UP AN ELECTRON FROM AUGER NEUTRALIZATION, WHEREUPON THE NEUTRAL
ATOM IS EJECTED FROM THE SURFACE VIA PAULI EXPULSION. THE NEUTRAL SPECIES
MAY ALSO BE IONIZED DURING ITS ESCAPE, (M + A +) CURVE, RESULTING IN
DESORPTION OF A POSITIVE ION.
..........
41
F ig u r e 3.1: E x p e r im e n t a l
a r r a n g e m e n t fo r
RLA.
.................... 52
F ig u r e 3.2: T h e r e d e s ig n e d RLA a r r a n g e m e n t a l l o w e d f o r e a s e o f s a m pl e
REPLACEMENT AND OPTICAL DETECTION ALIGNMENT. THE SET SCREWS AT (A), (B),
AND (C ) CAN BE LOOSENED TO ALLOW FOR ROUGH ADJUSTMENT OF M IP FOCUSING.
O n c e r o u g h a d ju s t m e n t h a s b e e n c o m pl e t e d , t h e s e t s c r e w a t (C ) c a n be
TIGHTENED WHILE FINE ADJUSTMENT WAS ACCOMPLISHED USING THE MICROMETER
......
MOUNT....
F ig u r e
53
3.3: T h e e x p e r i m e n t a l a r r a n g e m e n t f o r
R L A c o u p l e d w it h a n M IP f o r
ATOMIC EMISSION DETECTION..........................................................................
F ig u r e
3.4: O pt ic a l
55
im a g e s o f c r a t e r s fo r m e d o n p u r e d e o x y g e n a t e d c o pper
SAMPLES. A S THE LASER WAVELENGTH APPROACHES THAT OF THE RESONANT
ATOMIC GAS PHASE TRANSITION OF COPPER,
324.712 NM,
AN OUTER RING OF RE­
.............
62
SEM im a g e of c r a t e r fo r m e d o n p u r e d e o x y g e n a t e d
1000 SHOTS OF 0.75 M j PER PULSE ENERGY AT RESONANT
WAVELENGTH OF COPPER, 324.712 NM.........................................................................
63
DEPOSITED COPPER IS OBSERVED
F ig ur e
3.5: L in e
scan
COPPER SAMPLE USING
F ig u r e 3.6: SEM im a g e s of th e c e n t e r of th e c r a t e r s s h o w n in F ig u r e 3.5 r e v e a l
CHANGES IN SURFACE MORPHOLOGY OF THE CENTER OF THE CRATERS AT (A) 500 PM
ix
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
OFFLINE (324.254 NM), (B) 250 PM OFFLINE (324.504 NM), (C) 50 PM OFFLINE
(324.704 NM), a n d (D) o n l i n e (324.754 NM)........................................................................... 64
F ig u r e 3.7: O pt ic a l im a g e s o f c r a t e r s f o r m e d o n a p u r e , d e o x y g e n a t e d co pper
AS A FUNCTION OF POWER DENSITY BOTH ON- AND OFFLINE. THE TOP SERIES OF
CRATER IMAGES WERE PRODUCED FROM ABLATING 500-PM OFFLINE, 3 2 4 .2 1 2 NM,
AND ONLINE AT 3 2 4 .7 1 2 NM, REPRESENTED BY THE LOWER SERIES OF CRATER
IMAGES AT VARIOUS POWER DENSITIES. SEE TEXT FOR DETAILS.
........
....6 5
F ig u r e 3.8: A
sim pl ifie d p a r t ia l t r a n sit io n d ia g r a m o f t h e a l u m in u m a t o m .
....... 69
F ig u r e 3.9: O pt im iz a t io n o f d e l a y tim e a f t e r l a s e r p u l s e u s in g (A) RLA w ith
THE MIP o f f ; (B) RLA WITH THE MIP ON; (C) NON-RESONANT ABLATION WITH
MIP o f f ; a n d (D) n o n - r e s o n a n t a b l a t io n w it h M IP o n . S ee t e x t fo r
......
DETAILS
F ig u r e 3.10: O pt im iz a t io n
o f slit w id t h a n d
PM T
voltage.
70
(A ) 25 jam a n d - 1 9 0 0
V, (B) 25 fiM AND - 1 8 0 0 V, (C) 2 0 fiM AND - 1 7 0 0 V, (D) 15 j^MAND - 1 7 0 0 V, (E)
15 |hm a n d - 1 6 0 0 V . S ee
t e x t fo r d e t a il s ...........................
72
F ig u r e 3.11: E m is s io n s c a n s o f p u r e a l u m in u m s a m p l e w h il e a b l a t in g a t the
RESONANT ATOMIC (A) AND 5 0 0 PM OFFLINE (B) ......................................................................74
F ig u r e 3.12: A s im p lif ie d p a r t i a l t r a n s i t i o n d ia g r a m o f
th e
m o ly b d e n u m a t o m .7 6
F ig u r e 3.13: E m is s io n s c a n o f m o l y b d e n u m in a NIST s t a in l e s s stee l sa m pl e
WHILE ABLATION AT THE RESONANT ATOMIC TRANSITION OF MOLYBDENUM, 3 1 3 .2 5 9
NM .........
77
F ig u r e 3.14: M o l y b d e n u m in s t a in l e s s st e e l e m is s io n s s c a n s u s in g (A) RLA a t
3 1 3 .2 5 9 NM; (B) 5 0 0 PM OFFLINE AT 3 1 3 .8 0 0 NM; (c ) 1 NM OFFLINE AT 31 4 .8 0 0 ;
AND (D) 1.5 NM OFFLINE AT 3 1 4 .8 0 0 NM.........................................
F ig u r e 3.15: A b l a t io n w a v e l e n g t h s c a n s of m o l y b d e n u m
SAMPLES AT (A ) 4mJ, (B) 1.8 Mj, AND (C) 1 M J . ......
78
in s t a in l e s s stee l
F ig u r e 3.16: A n o v e r l a y o f m o l y b d e n u m e m is sio n in t e n s it ie s o b t a in e d a t
VARIOUS ABLATION WAVELENGTHS (BLUE) AND MOLYBDENUM ABLATION
WAVELENGTH SCAN (RED). ........
....8 2
...8 3
F ig u r e 3.17: A b l a t i o n w a v e l e n g t h s c a n s o f p u r e a lu m in u m s a m p le s a t (A ) 1 mJ,
(B) 0 .7 5 M l, a n d (C) 0.5 M j.........................................
84
F ig u r e 3.18: T h e
v a r ia b l e a n g l e s a m p l e h o l d e r c o n s is t e d o f a fee d t h r o u g h
sh a ft m o u n t e d t o a m ic r o m e t e r .
F ig u r e 3.19: T h e
........
p o s sib l e c o m b in a t io n s o f s a m p l e o r ie n t a t io n w it h r e s p e c t to
l a s e r p o l a r iz a t io n .
The
n e t r e s u l t of t h e se c o m b in a t io n s l e a d s to o n l y
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
86
THREE CASES: ORTHOGONAL (CH= GH=
AND 50% OF EACH POLARIZATION (BH=
Av = Ev), PARALLEL (AH= EH= Cv = Gv)
Hh = Dv = Fv = DH= FH= Bv = Hv) . .......88
F ig u r e 3.20: T h e o r etic a l plo t o f the s a m p l e p o sit io n v e r s u s e m is sio n in t e n s it y .
T h e d o t t e d line r e pr e se n t s e m is sio n in t e n s it y t r e n d if t h e e m issio n
PRODUCED BY THE LASER BEAM BEING ORTHOGONAL TO THE SAMPLE SURFACE WERE
A MAXIMUM, THE SOLID LINE REPRESENTS THE TREND THAT WOULD OCCUR IF THE
PARALLEL POLARIZATION WERE A MAXIMUM, AND THE DASHED LINE REPRESENTS NO
CHANGE IN EMISSION INTENSITY WITH SAMPLE POSITION.................................................
90
F ig u r e 3.21: E ffec ts o f s a m pl e a n g l e a n d p o sit io n w it h r e sp e c t to po l a r iz a t io n ,
WHILE ABLATING ONLINE, 3 0 8 .2 1 6 NM. ALL POINTS WERE OBTAINED AT THE SAME
SAMPLE POSITION, BUT SOME OF THEM ARE NOT SHOWN, IN ORDER TO CLARIFY THE
GRAPHS.....................
.91
F ig u r e 3.22: E ffec ts o f s a m p l e a n g l e a n d p o s it io n w it h r e s p e c t to p o l a r iz a t io n ,
WHILE ABLATING 5 0 0 PM OFFLINE, 3 0 7 .7 1 6 NM. ALL POINTS WERE OBTAINED AT
THE SAME SAMPLE POSITION, BUT SOME OF THEM ARE NOT SHOWN, IN ORDER TO
CLARIFY THE GRAPHS.............................................................................................................................93
F ig u r e 4.1: R e s o n a n t l a s e r a b l a t io n o f c h r o m iu m in a
b y Y a n g [6 3 ]...............................................................................
s t a in l e s s steel sa m ple
F ig u r e 4.2: R e s o n a n t l a s e r a b l a t io n o f m o l y b d e n u m in a s t a in l e s s stee l
SAMPLE BY H ou [6 4 ]...........................................................................................................................102
F ig u r e 4.3: O pt ic a l im a g e s o f c r a t e r s f o r m e d o n p u r e a l u m in u m s a m p l e s w hile
ABLATING (A ) 5 0 0 PM OFF THE RESONANT WAVELENGTH, 3 0 7 .7 1 6 NM, AND (B ) ON
THE RESONANT WAVELENGTH, 3 0 8 .2 1 6 NM USING A PULSE ENERGY OF 0 .4 MJ. (C )
AND (D ) SHOW CRATERS FORMED ON- AND OFFLINE USING A PULSE ENERGY OF 0 .9
MJ, RESPECTIVELY................................................................................................................................ 103
F ig u r e 4.4: R L A w a v e l e n g t h s c a n s o b t a i n e d u s in g a
F ig u r e 4.5: U n p u b l is h e d R L A
15-|im
s l i t w i d t h ................... 106
w a v e l e n g t h s c a n s o b t a in e d b y
H ou [7 2 ].
.............107
F ig u r e 4.6: F l o w d ia g r a m o f M G R m e c h a n is m r e s u l t in g in th e d e s o r p t io n o f a n
a t o m .......................................................................................................................................................... 109
F ig u r e 4.7: F l o w
io n o r a t o m
d ia g r a m o f
KF
m e c h a n is m r e s u l t in g in th e d e s o r p t io n o f a n
............................................................... ...................... .......................... ........... 110
F ig u r e 4.8: T h e A n t o n ie w ic z
m o d e l f o r th e d e s o r p t io n o f io n s a n d a t o m s ........... 111
F ig u r e 4.9: F l o w d ia g r a m o f R L A m e c h a n is m b a s e d o n c u r r e n t w o r k . S ee t e x t
FOR DETAILS............................................................................................................................................ 114
xi
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
F ig u r e 4.10: A s id e v i e w o f t h e l a t t i c e o f t h e s a m p le
IMPACT
...............
sur fa ce
u p o n l a s e r p u ls e
115
F ig u r e 4.11: (A ) N o n - r e s o n a n t a b l a t io n r e s u l t s in f e w e r d e s o r b e d a t o m s ,
COMPARED ITS ONLINE COUNTERPART (B ). SEE TEXT FOR DETAILS.................... .
116
F ig u r e 5.1: A n o n - s i t e w a s t e i d e n t i f i c a t i o n s y s t e m d e v e l o p e d b y F e l d h o f f e t a l .
T h e c o n v e y o r b e l t m o v e d a t a s p e e d o f 1 m /s a n d w i t h a n i n t e g r a t i o n tim e
OF 6.3 MS WHICH ALLOWED FOR RECORDING OF 158 SPECTRA PER SECOND............. 137
F ig u r e 5.2: T h e p o r t a b l e n e a r -in f r a r e d spe c t r o m e t e r . T h e lig h t is d ir ec te d
INTO THE SPECTROMETER THROUGH A FIBER OPTIC CABLE WHERE IT IS COLLIMATED
ONTO A DIFFRACTION GRATING, DIRECTED ACROSS TWO MIRRORS ONTO THE P b S
ARRAY WHERE 2 5 6 PIXELS CONVERT THE LIGHT INTO ELECTRICAL OUTPUT.................161
F ig u r e 5.3: T h e e x p e r im e n t a l a r r a n g e m e n t c o n s is t e d of a 5 0 0 -W t u n g s t e n b u l b
DIRECTED THROUGH A 0.5 CM SLIT IN THE SOURCE HOUSING, FOCUSED BY A
SUPRASIL LENS, PASSED THROUGH A SAMPLE, AND FOCUSSED ONTO THE END OF A
FIBER OPTIC CABLE..................
163
F ig u r e 5.4: W h e n t h e so u r c e is p l a c e d b e y o n d tw ic e th e l e n s ’s f o c a l l e n g t h ,
INCREASED INTENSITY WAS PRODUCED AT THE FIBER OPTIC CABLE. THE SOURCE
WAS PLACED ABOUT 15 CM FROM THE LENS................
165
F ig u r e 5.5: N IR s p e c t r u m of s k in l a y e r s in th e bic e p a r e a . A w a t e r s pe c t r u m is
SUPERIMPOSED TO ILLUSTRATE THAT WATER WAS THE ANALYTE RESPONDING IN THE
168
SKIN SAMPLES................................................................................
F ig u r e 5.6: A b s o r b a n c e f o r h o n e y d e w m e l o n slic es r e c o r d e d fo r
DRYING PERIOD AT EACH KNOWN WATER WAVELENGTH.........................
F ig u r e 5.7: W a t e r c o n t e n t a s a f u n c t io n
C a l ib r a t io n c u r v e s a t th re e k n o w n
1748 N M .....................................
a
7 1 - m in u t e
171
of tim e in h o n e y d e w m e l o n s .
w ater w a v e l e n g t h s:
1 6 9 8 ,1 7 1 8 ,
and
172
F ig u r e 5.8: M u s k m e l o n w a t e r c o n t e n t a s a f u n c t io n o f s a m p l e po sit io n a t the
THREE WAVELENGTHS ASSOCIATED WITH WATER: 1 6 9 8 , 1718, AND 1748 NM. HERE,
0 MICRONS DENOTES THE POSITION CLOSEST TO THE FRUIT’S SKIN, WHILE 500
MICRONS DENOTES THE POSITION NEAREST THE CENTER
.............................................174
F ig u r e 5.9: (A ) A n F T IR s p e c t r u m o f p o l y - e t h y l e n e t e r e p h t h a l a t e (PE T )
SAMPLES REVEALED A SPECTRAL FEATURE AT 1660 NM, WHILE A PEAK AT 1618 NM
WAS RELATIVELY SMALL BY COMPARISON. THIS IS A REPRESENTATIVE SPECTRUM OF
A 1.4 CM THICK SAMPLE OF PE T . (B ) A N IR SPECTRUM, WITH THE P b S ARRAY
SPECTROMETER, OF PE T SAMPLES SHOWS A SLIGHT WAVELENGTH SHIFT AT THE
CHARACTERISTIC 1 6 1 8-NM PEAK AND LACK OF RESPONSE NEAR 1660 NM
...... 178
xii
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
F ig u r e 5.10: (A ) C a l ib r a t io n c u r v e s , c o r r ec ted a n d u n c o r r e c t e d , a r e
PRESENTED FOR DATA COLLECTED FROM A BENCHTOP JASCO F T ® . FOR THE SAME
PLASTIC SAMPLES AT 1618NM. IT WAS ASSUMED THAT THERE WAS A 5% LOSS AT
EACH SURFACE, OR 10% FOR EACH LAYER THAT WAS BEEN ADDED FOR ANALYSIS.
(B ) C a l ib r a t io n c u r v e s , c o r r ec ted a n d u n c o r r e c t e d , f o r d a t a c o llected
FROM THE PBS SPECTROMETER AT 1618 NM. THE CORRECTED ABSORBANCE CURVE
ACCOUNTED FOR AN ESTIMATED 5% REFLECTION LOSS PER SURFACE IN THE
SANDWICHED PLASTIC. WHEN CORRECTED FOR REFLECTION, THE CALIBRATION
CURVE SLOPE APPROACHED ZERO, WHICH SUGGESTED THAT THERE WAS NO
179
ABSORBANCE BY THE PLASTIC. .................
F ig u r e A . 1: T r a d it io n a l cel l d e s ig n s u s e d in L A -IC P -M S
...............
CELL VOLUME RANGES FROM 0 .2 TO 12 CM3
F ig u r e A .2: N o v e l
c el l d e s ig n fo r u s e w ith
e x p e r im e n t s .
T y pic a l
184
R L A -IC P -M S ............................................... 188
F ig u r e A .3: T h e c e l l b o d y is c a l ib r a t e d in 3° in c r e m e n t s a n d t h e cell h o l d er
ALLOWS FOR COMPLETE ROTATION OF THE CELL BODY. THE BASE OF THE CELL
HOLDER CAN BE MOUNTED DIRECTLY TO AN OPTICAL BENCH.............................................189
F ig u r e A .4: (A ) T h e a r g o n s w e e p g a s e n t e r s t h e b o d y o f t h e c e l l t h r o u g h a V”
S w a g e l o k f i t t i n g b e f o r e p a s s in g t h r o u g h a s e r i e s o f b a f f l e s . T h e i n s e r t is
MOUNTED WITH A SERIES OF SET SCREWS SEALED WITH AN O-RING FITTING INTO (B ),
WHERE THE ARGON THEN PASSES THROUGH THE TUBE ARRAY, WHICH IS COMPOSED
3 1 6 STAINLESS STEEL HYPODERMIC TUBING x/z IN LENGTH................................................ 190
F ig u r e A . 5: E x p e r im e n t a l a r r a n g e m e n t f o r R L A -IC P -M S in c o r p o r a t in g
NOVEL CELL DESIGN........................................................................................
F ig u r e A .6: In - h o u s e
F ig u r e A .7: IC P -M S
the
191
b u il t t r a d it io n a l cel l f o r e f f ic ie n c y c o m p a r is o n ................ 194
g l a s s w a r e ...........................................................................................................198
F ig u r e A . 8: IC P -M S a r r a n g e m e n t fo r th e in t r o d u c t io n o f v a p o r iz e d so lid
SAMPLE...................................................................................................................................................... 199
F ig u r e A .9: IC P -M S
a r r a n g e m e n t fo r a q u e o u s s t a n d a r d a d d it io n c a l ib r a t io n .2 0 0
F ig u r e B . l : O v e r a l l R L A
F ig u r e B .2: A l ig n m e n t
e x p e r im e n t a l a r r a n g e m e n t .
......................
204
o f th e l a s e r b e a m t h r o u g h t h e f o c u s in g l e n s .....................2 0 6
F ig u r e B .3: T ilt / r o t a t io n p r ism st a g e k n o b a d j u s t m e n t s f o r d ir e c t in g the b e a m
THROUGH THE ELLIPSOIDAL MIRROR INTO THE ABLATION CHAMBER........................
208
F ig u r e B .4: A b u s in e s s c a r d is u s e d to e n s u r e t h a t t h e l a s e r b e a m is p a s s in g
THROUGH UNSCATHED (A ) AND IS NOT BEING CLIPPED AS IT PASSES THROUGH THE
ELLIPSOIDAL MIRROR AND ABLATION CHAMBER ( B ) ....................
...2 1 0
xiii
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
F i g u r e B .5: T h e g l a s s c e l l is r e p l a c e d w i t h a % r o u n d g l a s s c e l l a n d a
BUSINESS CARD IS PLACED OVER THE SAMPLE IN ORDER TO CHECK FOR BEAM
CLIPPING
F ig u r e B .6: T h e
211
.........
k in e m a t ic - m o u n t e d e l l ip so id a l m ir r o r k n o b a d j u s t s .....................21 3
F ig u r e B . 7: C r o s s h a ir im a g e a d ju s t o n m o n o c h r o m a t o r s l it . T h e in s e t
DEMONSTRATES POSSIBLE CROSSHAIR IMAGES. IMAGES RESEMBLING (A ) AND (B)
DEMONSTRATE THAT ONLY ONE AXIS IS FOCUSED, WHILE IMAGE (C ) SHOWS PROPER
FOCUSING OF BOTH AXES................................................................................................................... 2 1 4
F ig u r e C. 1: C o n f ig u r a t io n
F ig u r e C.2: W a v e l e n g t h
F ig u r e C.3: T a b l e
o f th e c o n t r o l p a n e l
F ig u r e D .l: F o u r - p h a s e
F ig u r e D .2: S te ppe r
..........
o f l a s e r e n e r g y d u r in g t a b l e w r it in g p r o c e d u r e s .
...2 2 0
m o t o r t r a n s l a t o r w it h c o n t r o l b o a r d m o d ific a t io n for
c o d e s a r e th e s a m e a s fo r t h o se s h o w n in
224
a n d b a u d r a t e se l e c t io n s c r e e n ...................
F ig u r e D .4: S t e ppe r
219
st e ppe r m o t o r w ir in g d ia g r a m ..................................................... 2 2 3
c o m p u t e r in t e r f a c e . T h e c o lo r
F ig u r e D . l ...................................
F ig u r e D .3: P o r t
217
r a n g e se l e c t io n s c r e e n ................................................................. 2 1 8
w r it in g d is p l a y .
F ig u r e C.4: O ptim iz a tio n
......
22 5
m o t o r c o n t r o l s c r e e n ............................................................................... 2 2 5
F ig u r e D .5: C o m p u t e r c o n t r o l l e d b r o a d b a n d a t t e n u a t o r . T h e b e a m pa t h
THROUGH THE ATTENUATOR IS SHOWN IN THE INSERT...........................................................2 2 6
F ig u r e D .6: L a s e r
a t t e n u a t io n a s a f u n c t io n o f st e p p e r m o t o r p o s it io n ............... 2 2 7
F ig u r e D .7: S t e ppe r
m o t o r m o u n t e d to th e s a m p l e h o l d e r ............................................. 2 2 8
x iv
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
I GENERAL INTRODUCTION
1.1 Introduction
The focus o f this thesis, Resonant Laser Ablation (RLA), is a variant of Laser
Ablation (LA) which has been gaining popularity as a method of sample introduction,
affording minimal sample preparation, localized microanalysis, and the ability to
analyze both conducting and non-conducting samples. There are several commercial
bench-top instruments currently available that couple to a mass spectrometer detection
system. These systems rely on high-powered, fixed wavelength lasers to ablate the
sample surface for analysis. The goal is to produce a low-density plume of material
with the same stoichiometric ratios of constituents, as found in the solid material.
However, fractionation may occur due to preferential ablation of a particular element
resulting in non-representative stoichiometric ratios in the plume. Fractionation is the
most significant drawback o f LA sampling and arises possibly from various elemental
properties including boiling and melting point, volatility, and ionization potential, and it
ultimately affects calibration and accuracy. The majority of the work done to diminish
fractionation effects has been through the exploration of laser parameters.
Mao et al. have shown that wavelength, as well as power density and pulse
width, affect the stoichiometric ratio of copper to zinc in the ablation of brass [1]. With
a 30 ns pulse-width laser, where thermal vaporization appears to be the dominant
process, the Zn/Cu ratio approaches stoichiometry at power densities, greater than 0.3
GW cm'2. With 3 ns and 35 ps pulsed Nd:YAG lasers, stoichiometric ablation can be
achieved at higher power densities, although thermal processes dominate ns ablation,
1
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
while the ps ablation seems to be governed by a nonthermal mechanism. They have
also shown that UV ps lasers provide greater emission intensity and wide fluence range
in order to obtain accurate stoichiometric analysis. However, UV ps lasers are not as
reliable and stable as the ns YAG laser.
Cromwell et al. have determined that fractionation is prevalent for ablation at
low laser fluence (< 1 J c m ') and with multiple overlapping laser pulses incident on the
same area of the sample surface [2]. The authors contend that it is possible, although
impractical, to predict which element may be preferentially ablated upon inspection of
solid-liquid binary phase diagrams for homogeneous materials, given the approximate
composition o f the sample, the thermodynamic properties of the analyte, and matrix
species, relevant phase diagrams, and knowledge of the heat-affected zone. Rather, due
to lack of matrix-matched standards, the authors suggest the use of a high fluence and
the avoidance of sampling previously ablated areas.
Eggins et al. have shown, using LA coupled with an inductively coupled plasma
mass spectrometer (ICP-MS), that during fractionation the more refractory elements are
condensed onto the surface before the more volatile ones [3], Also, a lower density
carrier gas gives an enhanced signal in the ICP-MS.
Figg and Kahr investigated fractionation as a function of wavelength and laser
irradiance [4], The authors observed significant fractionation, using 1064 and 532 mm
wavelengths, from a 50-Hz Nd:YAG laser, that was dependent upon power density.
The fractionation observed at these wavelengths was attributed to surface heating, thus
vaporizing the more volatile species, to which the authors related the metal oxide
2
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
melting points. However, ablation at 266 nm resulted in best overall LA-ICP-MS signal
stability.
Gunther et al. have studied fractionation in geological samples using a 193-nm
excimer laser as a function of repetition rate and power density [5]. Insignificant
fractionation effects were observed using modest power and low rep rates, below 10 Hz.
However, increasing the repetition rate improves the count rate of the ICP-MS, but
results in a non-linear increase in fractionation.
Gunther et al. compared the ablation behavior using a 266 nm Nd:YAG and a
193 nm ArF excimer laser coupled to an ICP-MS [6]. Both lasers had a pulse rate of 10
Hz and were attenuated to 0.4 and 0.5 mJ per pulse for the Nd:YAG and ArF lasers,
respectively. The authors note that significant fractionation occurred using the 266 nm
laser, while the 193 nm laser permitted controlled ablation. Therefore elemental
fractionation appears to be chiefly controlled by the wavelength of the laser.
1.2 Resonant Laser Ablation
Although laser ablation sample introduction is still in its infancy, LA is gaining
popularity with the availability of commercial instruments. However, rather than
disadvantages, there are also some potential advantages to inducing selective
fractionation, including improvement of sensitivity and selectivity of trace elements,
while decreasing non-selective background ablation. It has been demonstrated that by
tuning the laser wavelength to match that of a resonant atomic gas phase transition of a
particular analyte, the resultant analyte signal is greatly enhanced; hence the genesis of
resonant laser ablation (RLA). Although the mechanism by which RLA occurs is not
3
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
fully understood, which is discussed in Chapter 2, it is gaining popularity using tunable
dye lasers, and solid-state optical parametric oscillator (OPO) lasers that have a wide
tuning range spanning 220 - 2000 nm with a narrow line width.
Much of the work to be presented here relies on the use of tunable lasers for the
ablation process, although several parameters including repetition rate, pulse duration,
power density, and sample angle, vary between experiments. Typically, mass
spectrometers have been used for detection, since it has generally been believed that the
RLA process results in an increased number o f ions, which can be detected by a mass
spectrometer. Aside from inherent wavelength dependence, this chapter’s critical
review of RLA has been subdivided into the various aspects of the RLA process based
on bulk material, as well as, laser power density, laser polarization, sample angle,
matrix effects, and also alternative applications of the RLA’s selectivity.
1.3 Metals
Verdun et al. have demonstrated the utility of RLA for the analysis of cadmium,
copper, and molybdenum in thin metal-doped polymers and steel alloy samples coupled
to a laser microprobe mass analyzer [7]. The authors used a single pulse to achieve
both ablation and resonant ionization in order to investigate it’s analytical utility
compared to resonant ionization spectroscopy (RIS). In RIS experiments, an ablation
laser is directed normal to the surface resulting in the generation o f a atoms, ions, and
molecules; and a second laser, tuned to the resonant gas phase transition of the analyte
is directed parallel to the surface to cause resonant ionization. Using a 10-Hz Nd:YAG
pumped dye laser with a 25 ns pulse width and a power density of 109 - 1011 W cm'2, a
4
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
cadmium-doped polymer was resonantly ablated resulting in factor of five increase in
Cd+ signal intensity.
In order to test reproducibility and possible matrix effects, thin films of various
copper acetylacetonate-doped polymers were prepared. As the ablation laser was
scanned through two neighboring copper transitions at 324.75 and 327.39 nm,
noticeable increases in copper ion signals were observed. These results were compared
to those obtained for RLA o f copper in various matrices. The authors concluded that
differences in observed ion yield between samples were a result o f matrix effects,
particularly since the spot size and power density were consistent.
The authors propose a two-part mechanism by which the resonant enhancement
occurs. It is suggested that the leading edge of the pulse generates primarily neutral
atoms with a small percentage of ions. The ablation process is believed to be matrixdependent. The trailing edge of the pulse is then responsible for the resonant ionization
of the neutral atoms. This proposed mechanism is essentially that which is used to
describe the resonant ionization spectroscopy (RIS). In RIS experiments, two lasers are
used: the ablation laser, which is of fixed frequency, is directed incident to the sample
surface creating the dense plume of ablated material, while a second laser, directed
parallel to the surface, is tuned to the resonant gas phase transition of the analyte to
resonantly ionize the analyte. A detailed look at the characteristics of this mechanism is
discussed in Chapter 2.
Borthwick et al. have shown that RLA can be a reliable method for measuring
the ratio between ion and neutral atom yields [8]. Here, the authors coupled a
Nd:YAG-pumped dye laser with a pulse width of 10 ns to a time-of-flight mass
5
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
spectrometer (TOF-MS) for the determination of aluminum in NIST SRM standard
1263A and 1261A stainless steel samples. The ablation laser was directed at an angle
of 6° with respect to the surface with a power density of approximately 1 x 107 W cm'2.
The ablation wavelength scan, from 307.6 to 309.6, revealed two resonant transitions at
308.30 nm and 309.37 nm, each about 50 pm FWHM, which is similar to that obtained
by McLean et al. for the determination of aluminum in the AlGaAs semiconductor
sample [13]. In another experiment, the ablation laser was directed at the sample while
ablating for 500 laser shots at fixed wavelengths, both on on- and off-line, 308.3 nm and
307.5 nm, respectively. The on-resonance mass spectrum confirms that RLA provides
significant enhancement of the aluminum signal with estimated detection limits of about
5 ppm.
An enhancement o f the iron signal was also observed while scanning this
wavelength range. Although there is no wavelength in this range that connects the iron
ground state to an excited state, enhancements were seen at 307.63 and 308.47 nm,
which correspond to transitions in atomic iron fine structure, and suggested that iron
atoms are present in the ablation plume.
Using RLA, the ratio o f ions to neutrals, n+/n°, was measured by first recording
the ion yield when the laser was tuned to an atomic resonant transition and then
detuning to count the number of ions produced off-line. The authors contend that this
measurement is possible for two reasons: ionization of neutral atoms happens before the
atomic plume has had time to expand significantly, therefore providing a good
geometric overlap between the laser beam volume and the ablated neutral atoms; and
the efficiency o f ionization o f neutral atoms approaches 100%, since the photoexcitation
6
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
and photoionization steps are saturated at the fluences used. At low fluences, 0.4 mJ
per pulse, the on-line ion signal was about a factor o f four greater than that observed
off-line. As the laser fluence increased, the off-line ion yield seemed to increase
linearly as the on-line yield seems to plateau as the laser fluence approaches 2.0 m j per
pulse. This work is significant in that low power RLA leads to greater ion production,
thus greater efficiency, whereas off-line ablation suggests traditional laser ablation
processes.
Gill et al. studied the use of low powers for the RLA of copper samples
including bulk copper and thin film samples using a 10-Hz excimer-pumped dye laser
with a pulse width of 12 ns and a power density between 1.4 x 105 and 5.5 x 106 W cm'2
[9]. It was shown that RLA permitted the selective ionization, > 105:1, of trace analyte
from solid materials using power densities < 2 x 106 W cm'2. The authors attributed this
to the resonant nature o f the ionization process as well as an effective overlap of the
vaporized sample and the laser beam, spatially and temporally. The effects of spot-size,
surface-sample preparation and beam polarization were studied. The sample removal
rate of the copper thin film was calculated to be 100 fg per shot with confirmation by
both Auger spectroscopy and noncontact-surface profilometry measurements. Velocity
distributions were found to be Maxwellian, with an estimation of peak velocities at
about 1 x 105 to 2 x 105 cm s'1. In addition, they reported that laser polarization had a
strong contribution to the RLA process, and this was sample surface dependent. Based
on theoretical calculations and experimental results, the desorption process was shown
to be non-thermal in nature.
7
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
Eiden et al. have demonstrated the efficacy of RLA in copper, rhenium, and
nickel samples by semiquantitative analysis [10]. Their experimental arrangement
consists o f an excimer-pumped dye laser operating at 10 Hz with a pulse width o f 10 ns,
and a power density o f 1 x 107 W cm'2 incident at an 11° angle with respect to the
sample surface. The authors use a “2 + 1 ” ionization scheme, which corresponds to
photons to resonance + photons to ionize. Using this method for the determination of
trace impurities in rhenium, signals from iron, aluminum, copper, and chromium were
observed in concentrations o f 70, 5, 2, and 3 ppm, respectively. Virtually no signal
from the bulk material was observed, which was attributed to the excellent overlap of
the vaporized sample with the laser beam. However, sodium and potassium were
present in all spectra, possible due to their high volatility, low ionization potential, or
surface impurities. Analysis o f a high purity copper sample was performed to
demonstrate RLA’s analytical utility. During this analysis, copper, iron, silver and lead
were present in quantities below 0.1 ppm. Therefore, ultimate detection limits could
easily fall in the ppb range. A nickel sample was then analyzed for the trace detection
of technetium using the 2 + 1 resonant ionization scheme. The amount of trace
impurities present in the mass spectrum was attributed to the rough surface morphology
of the nickel sample.
The sensitivity or RLA was then characterized using a two-beam geometry,
similar to RIS. Here the authors proposed that this arrangement could be applied to
separately probe both the ablation and the ionization processes. In these experiments,
the output of the dye laser was split into two beams (-10% / 90% split), the weaker
beam was directed onto the surface at an angle of incidence of 11°, while the stronger
8
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
beam was passed across the surface at near grazing incidence with careful adjustment of
beam overlap. When both beams were present, a strong signal was observed (-100
ions/shot), while only very little signal (<1 ion/shot) with either sample blocked. The
authors have concluded from this that the second beam generated more ions than
created with a single beam. However, if the stronger beam was used to ablate the
sample, ion signals exceeding tens of thousands per shot were observed, that is, strong
ion signals were observed with the ablation beam intensity below the single beam
threshold. From this experiment it can be concluded that using a stronger beam for
ablation and the weaker beam for ionization results in a more efficient means of
resonantly ionizing the sample.
Campbell et al. have demonstrated the matrix-dependence of RLA of trace
manganese using TOF-MS detection [11]. Although details of the laser system used
were not included, the authors noted that an increase in pulse energy from 4 to 9 pJ
caused a delay in the signal by 400 ns, which resulted in broadening o f the mass peaks
and a consequent reduction in resolution of the system. This was overcome by
prefocusing the beam to produce an increase in spot size, so that the resolution was
similar to that of the 4-pJ pulse, but with a four-fold increase in ion yield. The authors
demonstrated an ion yield dependence on the mix of second harmonic and fundamental
dye wavelengths. Using only a prefocused 13-pJ beam at 289.19 nm produced twice
the number o f ions compared to ablation using 510 pJ of 560.38-nm radiation.
However, a mixture o f the two beams for two-photon ionization, 7.5 p j of 280.19 nm
and 500 pi o f 560.38 nm, produced an enhancement factor of two over the UV-only
case. Analysis of trace quantities of manganese in copper matrices proved inaccurate
9
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
due to the large variations in impurities across the surface. However, analysis of
manganese in a NIST zinc and an aluminum sample matrices allowed detection limits to
be calculated at the 50 ppm and 200 ppb level, respectively. Calibration curves were
established using aluminum samples of varying manganese concentrations. From this it
can be concluded that the analytical utility of trace elemental analysis using RLA is
matrix dependent.
1.4 Metal Oxides
Pang and Yeung have attempted to calibrate the ions produced from NaojWOs
and copper samples using RLA with respect to the intensity o f the acoustic signal
produced using a 10-Hz excimer-pumped dye laser with a pulse width of 25-ns and a
power density below 1.3 x 106 W cm'2 and an ion collector [12], For the wavelengthdependent studies, the ion signal was monitored with a gated integrator and boxcar
averager as a function of dye laser wavelength. The 10-nm wavelength scan
encompassing the resonant transition, at 589 nm, indicated a gain in ion signal of over
factor o f 70 when the wavelength matched that of the resonant gas phase atomic
transition of sodium, with a full width half maximum (FWHM) of the Lorentzian profile
o f 2.2 nm. From acoustic signals, it was determined that the same amount of material
was generated when the laser was both on- and off-line. Similar results were obtained
using copper samples, although the spectrum was slightly skewed towards the longer
wavelengths with a FWHM o f 1.4 nm.
The authors also noted that spectral broadening occurred as a function of power
density and higher gas pressures, which they attributed to collisional broadening, while
10
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
they discounted contributions from power broadening since similar power densities
were used in flames. The dense plume of atoms created at the initial moments
following the laser pulse was substantial enough to result in the formation of dimers,
indicating the high likelihood of collisional broadening, confirmed upon inspection of a
transition associated with the sodium dimer. The authors also note that the a red shift
occurred in the absorption maximum as the laser focal point increased, which could
have resulted from either a change in vibrational amplitudes of the dimers resulting
from changes in collision frequency or from plasma absorption, a prefilter effect, in the
plasma.
The ion signal produced was then calibrated with respect to the acoustic signal
and they determined that, at higher laser fluences, >5.5 x 104 W cm2, the calibration
curve rolled over. Also, as the spot size is increased by defocusing the laser, the slopes
of the calibration curves decreased, indicating a decrease in sodium ion production
caused by the increased dimer production.
The authors conclude that the single laser form ofRIS can be successfully
implemented and a very good linear correlation between acoustic signal and ion yield
was achieved over three orders of magnitude.
Gibson has reported RLA of europium and lutetium using a pulsed dye laser in
the 450 - 470 nm range. The enhancement effect was attributed to the resonant
photoionization or photoexcitation of atoms in the ablation plume. It was proposed that
neutral atoms of the analytes are excited by a one-photon resonant absorption process,
and subsequently preferentially ionized by a quasithermal collisional process. Atomic
transitions originating from atoms at high energy levels were also observed, thus
11
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
suggesting a significant population of ablated atoms with high internal energies. A
resonant enhancement of a factor of ten was observed for europium and lutetium ions,
compared to non-resonantly ablated ions from the concurrent laser pulse. However,
parameters such as wavelength, power density and ion extraction voltages were not
optimized for the estimation of detection limits.
1.5 Semiconductors
McLean et al. have shown significant enhancement in aluminum and gallium in
a mirror-finished thin layer AlGaAs semiconductor deposited on a GaAs substrate [13].
For the aluminum ablation wavelength scan, from 307.5 - 310 nm, an enhancement of
two orders o f magnitude was obtained using low fluences, typically about 100 pJ (1 x
107 W cm"2), and resulted in signals less than 50 pm FWHM. From this scan, two
resonant aluminum lines were clearly distinguishable, at 308.297 and 309.367 nm.
Similar results were obtained for the resonant excitation o f gallium. In this wavelength
scan, from 615.5 nm - 620 nm, the two resonance signals from gallium isotopes were
discernible with less than 50 pm FWHM and in their correct isotopic ratios.
Sodium, which is an impurity in the AlGaAs semiconductor, provided a means
of quality control to confirm the non-spurious nature of the RLA process. The above
experiments were repeated and no sodium signals were observed even with incident
angles up to 25°.
A calcium sample was then analyzed to study the effects, in a different matrix,
of scanning the incident wavelength compared to that obtained using the RIS
arrangement. The laser was directed at an angle of 3° and scanned through the resonant
12
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
transition as previously described. The resulting wavelength scan produced results
similar to previously obtained, with a 20 pm FWHM. To compare the effects of RLA
with that o f RIS, the calcium sample was ablated with a 1064-nm Nd:YAG laser, while
a second laser parallel to the sample surface directed through the laser plume 2mm
above the surface. The resonant laser was delayed by 2 ms with respect to the ablation
laser to obtain optimal overlap with the plume. The position o f the resonant transition
obtained using the RIS arrangement matched exactly that produced using the RLA
arrangement, thus indicating that the resonant excitation that occurred in RLA was very
similar to that which occurred in RIS - that is, nonselective ablation, followed by gasphase resonant excitation. The authors contend that if surface bonds play a role in the
resonant ablation process, peak broadening would be evident in the RLA wavelength
scan.
Wang et al. used the same experimental arrangement as McLean et al. [13] for
the study of RLA o f Ga in a mirror finish GaAs semiconductor sample [14], Wang et
al. varied the laser power density from the ion detection threshold, 0.39 m l per pulse,
up to 1 mJ per pulse (~ 8 x 106 W cm'2), while ablating at a grazing angle of 5°.
Interestingly, as the laser power density increased, the width o f the resonant peaks
increased asymmetrically towards longer wavelengths, while further increases led to the
disappearance of the resonant phenomenon and poor spectral resolution, probably due
to space charge effects and the onset of collisional processes in the ablation plume.
This red shift is similar to that observed previously by Pang and Yeung [12].
The sample angle was then varied between 5° and 8°. Although the sample
angle range was very small, the resulting ion signal increased significantly, a factor of
13
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
28 between 5° and 7°, using a fluence of 0.26 m J per pulse. Increasing the fluence to
0.61 mJ per pulse, while varying the angle between 5° and 8°, resulted in a 10-fold
increase across the angle range. In both cases marked asymmetrical broadening
towards longer wavelengths was demonstrated as shown previously. The authors
concluded that this asymmetrical profile resulted from optical collisions, which are
perturbations in the electronic levels of atoms which can result in resonant absorption of
non-resonant radiation. This phenomenon is even more pronounced by the increasing
density of neutrals near the surface.
1.6
Oth er RLA Applications
The analytical utility of RLA for trace metal analysis has been demonstrated.
However, there have been other applications, which exploit its usefulness, that have
several advantages over normal LA.
1.6.1 Resonant Laser Induced Breakdown Spectroscopy
In a typical (LIBS) experiment, a high-powered laser is used to produce a plume
o f ablated material, while the optical emission spectrum is used to provide information
about the elemental content of the material. Capitelli et al. have used RLA LIBS, citing
two main advantages: the time necessary for plasma formation is much less than that for
the use of non-resonant laser radiation; and the plasma produced requires less energy
[15]. Tsipenyuk and Davydov have demonstrated the potential of RLA when applied to
LIBS [16]. A frequency-doubled Nd:YAG laser, 532 nm, was used for the resonant
excitation o f copper, which has a resonant excitation 529.2 nm, with weaker lines at
532.3, 535.2, and 535.4 nm. Although this is not true RLA, the authors contend that by
14
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
slightly detuning the laser, the resonant excitation can be observed, however, this was
not demonstrated in the paper. Rather, using the 532- and 1064-nm ablation laser, the
shapes o f the emission spectra produced on a copper sample were virtually identical.
However, there was a significant increase in signal-to-noise ratio for the 532-nm
ablation, evident from the 578.2-nm copper line, which increased by a factor of six.
The authors also investigated the emission spectra o f a laser-induced plasma on
the surface o f a saturated aqueous copper sulfate solution. Although the presence of a
large hydrogen emission line was observed, the authors maintain that it can be used as
an inherent reference line for the determination of other elements.
The authors have demonstrated that by tuning the laser to the resonant gas phase
transition of the bulk material, the plasma produced allows for higher sensitivity, which
is advantageous in LIBS studies. However, the authors do not take advantage of the
longer-lived plume under the resonant conditions. This would allow for a longer
integration time, from which even higher signal-to-noise ratios would result.
1.6.2 Selective Ion Source
Gill et al. have used RLA to selectively generate chromium, iron, nickel, and
copper from a stainless steel sample for gas-phase ion molecular reactions with acetone
[17]. By utilizing an ion trap, the ablation was accomplished using an excimer-pumped
7
7
dye laser with a pulse width of 15 ns and a power density of 2.3 x 10 W cm ". RLA of
nickel in the NIST stainless steel sample was used to demonstrate the selectivity of the
technique, since no mass spectral peaks were observed for nickel or any other
component of the sample when the laser was off resonance. The net reaction for the
15
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
various metal ions [M+] with acetone is adduct or bi-adduct formation, as shown in
Equations 1 and 2, respectively.
M+ (g) + (CH3)2CO(g) -> M(CH3)2CO+ (g)
Equation 1
M(CH3)2CO+ (g) + (CH3)2CO (g) -> M{(CH3)2CO}2+
Equation 2
The results obtained for the various adducts and bi-adducts agreed with those previously
demonstrated in the literature, which indicated that RLA is an effective tool for the
selective generation of ions.
1.7 Conclusion
The enhancement effect of RLA has been shown on a variety of samples from
pure metals to semiconductors and from major constituents to trace impurities. The
magnitude of the energies to accomplish the enhancement has been shown to be
significantly lower than those of normal LA experiments, by factors of 10 - 100 times
less. This enhancement effect varies depending on the matrix of the sample; therefore
methods of calibration may be difficult without the use of matrix-matched standards
when employing mass spectral detection.
The resonant effect is more pronounced at lower power densities. An increase in
power density is accompanied by an increase in non-resonant ablation leading to a
decreased signal-to-noise ratio. At lower power densities RLA spectra are also much
simpler because the enhancement effect of trace analytes overshadows other
constituents in the bulk material.
16
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
It has been demonstrated that sample angle plays a role in the efficiency of
RLA, but this has not been investigated beyond only small ranges of near-grazing
sample angles. Further studies must be done to corroborate the results. In other studies,
arbitrary angles were chosen, with no explanation. The fact that enhancements have
been demonstrated to be a function of sample angle, as well as wavelength, indicates the
possibility of more efficient laser-material coupling, rather than a purely gas phase
mechanism to explain the enhancement. However, even though the observed signals
resulting from RLA exhibit angle-dependency and therefore suggests a possible lasermaterial interaction, there appears in the literature no deviations from the original
proposed mechanism. Such angle-dependency and polarization studies will be explored
in Chapter 3.
The analytical utility o f RLA has been demonstrated in a preliminary fashion,
although significant understanding of the mechanism of the RLA enhancement is
currently absent in the chemistry literature. A study of LA and RLA as described in the
physics literature is described in detail in Chapter 2. By a thorough optimization of
parameters, this technique could become a valuable solid-sampling analytical tool with
lower detection limits than reported for LA, while affording the advantages of LA
including, minimal sample preparation, analysis of a wide variety o f samples, rapid data
acquisition time, and both bulk and trace analysis, with the benefits of RLA which
include sensitivity and selectivity.
17
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
2 A MECHANISTIC STUDY OF RESONANT LASER
ABLATION
2.1 Introduction
Resonant laser ablation (RLA) has demonstrated excellent potential as an
analytical tool by selectively enhancing the analyte signal with respect to bulk material.
This technique has significant potential analytical utility for the determination of trace
metals, and the phenomenon has been widely demonstrated by coupling this technique
with mass spectrometer detection. However, its viability is hindered by the lack of
understanding of the mechanism by which this enhancement occurs. There are several
factors that must be explored in order to fully understand and exploit the usefulness of
this technique; therefore sufficient understanding of the underlying mechanisms behind
the RLA process is required. This review discusses such literature in the context of
both resonant and non-resonant interactions in terms of laser induced desorption (LID)
and laser ablation (LA), and lends insight into other factors that might have been
overlooked in the description of the theory.
Hagland describes the conceptual framework of the phenomenology of lasersurface interactions consisting of five distinct phases: (1) absorption of laser light by
excitation of electronic or vibrational modes of the solid; (2) competition between
localized and delocalized modes for the absorbed photon energy; (3) mechanisms of
laser-induced desorption from metals, semiconductors, and insulators; (4)
phenomenology of laser ablation as a function o f wavelength and pulse duration; and
18
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
(5) emission of photoelectrons, ions, and neutral species into the developing ablation
plume.
2.2 Laser Ablation vs. Laser Induced Desorption
According to Haglund, these two phenomena can be differentiated by describing
laser-induced desorption (LED) as dominated by microscopic mechanisms, and is a
much gentler process resulting in the emission of ions, atoms, and molecules without
any substantial disturbance in the surrounding surface. The term LID is often used to
describe desorption of chemisorbed molecules from the surface of the metal substrate.
Laser ablation (LA) is a macroscopic mechanism associated with a large-scale
disturbance o f surface and sub-surface geometrical and electronic structure, and under
some conditions may change the stoichiometry as well, due to preferential ablation [18].
The underlying difference between LID and LA is the laser energy required to complete
the process.
LID and LA are thus most accurately viewed not as a distinct phenomenon, but
as endpoints of a continuum, ranging from the ejection of isolated atoms, ions and
molecules from a surface, which remains essentially intact on the microscopic scale
(LID), to the massive removal of material and destruction of intermediate or long-range
order at the mesoscopic scale (LA). The LID process implies desorption resulting from
the localization of energy. The characteristic length scale associated with absorption of
an isolated photon is on the order o f dmiao « xVibVs, where Tvib is a typical phonon
vibrational period in the material, and vs is the speed of sound. Haglund’s definition
summarizes the idea that desorption must be a localized event, and that lattice instability
19
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
must be initiated before the incident photon energy is dissipated into non-local
vibrational modes of the solid. The length scale derived from this simple idea - ranging
from 0.1 to 10 nm - is consistent with other important microscopic length scales, such
as screening lengths for electron-hole plasmas in semiconductors, which is a measure of
the distance of Coulumb repulsion between carriers of like signs, and relaxation lengths
for polarions and self-trapped excitons in ionic insulators. At such length scales,
average measurements o f surface structure, such as low-energy electron diffraction,
give no evidence of any disturbance at all. [19]
Laser ablation, however, is associated with a characteristic mesoscopic length
scale of order dmes0 ~ ^iaserVs, where Tiaser is the duration of the laser pulse and vs is the
sound speed in the material [18]. This implies the idea that the excitation,
thermalization and lattice instability, which induce ablation, are “pumped” by the laser
pulse for its entire duration. This length ranges up to a few tens o f microns in some
cases, consistent with the idea that ablation implies the excitation and removal of a
substantial fraction of a monolayer, a volume of order 30 x 104 pm3, or a few tens of
microns in linear dimensions [18]. These size scales are consistent with typical
diffusion lengths associated with the thermal view of laser ablation, since diffusion
constants have a range dictated by density and the speed o f sound in the material. This
length scale is also one in which the thermal and mechanical properties of materials can
be said to be meaningful. As we shall see later, the bond-orbital theory makes a
straightforward connection between the microscopic and macroscopic length scales.
[19]
20
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
Haglund defines the power density range for LID between 1 x 103 - 1 x 107 W
cm'2 and between 1 x 107 - 1 x 10n W cm'2 for LA [19]. However, the RLA
phenomenon lies on the threshold between these two power density ranges and
therefore has characteristics of both LID and LA processes.
There are three distinct types of materials that behave differently under
irradiation with incident laser light: metals, semiconductors, and alkali-halides. The
physical characteristics of each of these materials are discussed here in turn.
A metal surface can be considered as a lattice o f ions immersed in a sea of
delocalized valence electrons of individual atoms occupying the metal valence-band
states. Upon photon impact, a sputtered particle escapes as a positive ion as electrons
stream from the surface. Picking up one or more of the escaping electrons can
neutralize the metal ion. Resonant electron transfer is assumed to be the most dominant
electron pick-up process, since one-electron interactions are much stronger than multi­
electron interactions [20]. In this process the ejected metal ion, having the same charge
state as in the solid, becomes neutralized by picking up an electron from the valence
band o f the solid. In addition, both conduction- and valence-band electrons may
participate in laser excitation, the former through a free-electron-like response, and the
latter an interband response that has a threshold corresponding to the energy separation
between the valence and conduction bands. The nearly free conduction-band electrons
absorb photons by direct heating of the electron gas well out into the infrared [19].
Therefore, the sputtering o f a neutral atom is regarded as the evolution in time of an
initially strongly coupled system consisting of a metal ion, in which an electron evolves
from an initial valence-band state through the conduction band to a final atomic state,
21
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
which can be described by the Menzel-Gomer-Redhead model and the KnotekFeibelman model, which will be discussed later.
The response o f insulators and semiconductors to optical excitation involves
both electronic and ionic contributions to the dielectric function. The variation of the
dielectric function in nonmetallic solids with wavelength is complex, with the real
(refractive) part dominating away from vibrational or electronic resonances and the
imaginary (absorptive) component in the ascendancy near resonances [19]. There are
virtually no free (conduction-band) electrons at room temperature in insulators and only
a very small number in semiconductors. Thus optical absorption in laser-irradiated
semiconductors and insulators at moderate to high laser intensities, in the range of MW
cm' to GW cm' , leads to the creation of electron-hole pairs rather than electron
'y
heating. However, at extremely high intensities, generally in excess of TW cm' , it is
possible to generate enough conduction-band electrons in insulators to generate
significant free-electron heating. The relaxation mechanism by which the electrons and
holes recombine, and the response of the lattice to the creation and motion o f free
charges and electron-hole pairs, depends in a complex way on two factors. These are
the ratio of the laser photon energy hv to the bulk bandgap energy Eg, and the electronic
structure and binding configuration of the solid and its incorporated defects [19].
Therefore, we can say that the processes of LID and LA depend not only on the
power density o f the incident radiation, but the properties of the target material, such as
the electron-lattice coupling strength, and how these two parameters interact to produce
the localization o f energy and instabilities within the lattice.
22
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
2.3 Resonant Laser Ablation Theory
Typical LA experiments use fixed-wavelength lasers to complete the ablation
process. It has been previously demonstrated that UV wavelengths provide several
advantages over visible and IR lasers, such as better depth control for polymer etching
[21,22]; better stoichiometry and smoother finish in film deposition [23,24]; and higher
ablation efficiency of metals which generates higher emission signals in the ablation
plume [25].
In contrast to typical LA experiments, RLA relies on tunable lasers, such as dye
lasers, or solid-state optical parametric oscillator (OPO) lasers, to match the laser
wavelength to known atomic and/or ionization transitions of analyte atoms in the
sample. It has been demonstrated that by tuning the laser wavelength to the resonant
gas phase atomic transition o f the analyte, less energy is required for analtye removal
while affording higher sensitivity and selectivity compared to its LA counterpart
[7,12,26]. A tabulated summary of the RLA work as seen in the literature is given in
Table 2.1.
Verdun et al. [7] have reported that RLA results are strongly dependent on the
nature of the sample matrix and demonstrated a five-fold enhancement for cadmium,
copper and molybdenum, with a particularly broad bandwidth of the resonance, around
0.4 to 0.7 nm. The authors suggest that this enhancement was a result of the proposed
two-step pulse mechanism in which the leading edge of the laser pulse produces a lowdensity plume of ablated material consisting of atoms, ions, and molecules. The trailing
edge of the pulse is then responsible for the resonant ionization o f the analyte, which
23
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
produces the enhancement. This mechanism of RLA, shown in Figure 2.1, has been the
accepted explanation of the enhancement of the signal from analyte ions. This
description is similar to that used to describe the resonant ionization spectroscopy (RIS)
in which two lasers are used. The ablation laser is directed normal to the sample
surface, while a second laser, tuned to the resonant gas phase transition, resonantly
ionizes the ablated material [27].
24
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Table 2.1 Applications of RLA and their significance.
element
sample matrix
laser parameters
Ablation wavelength
& ablation scan spectral features
detection
system
Na: 589 nm; Cu: 578 nm; 2.2 and
1.4 nm FWHM in vacuum,
respectively; FWHM increase with
He pressure.
MS
observations and conclusions
Over 70 fold ion signal
enhancement observed on-line at
589.0 nm for sodium compared
to off-line, although same
amount o f material is generated,
Matrix affected resonant
enhancement effects; 5-fold
enhancement resulting in LODs
o f 0.05 ppm Cu and 1 ppCd.
Proposed “one-pulse
mechanism” whereby pressure
o f the plasma dining the second
step caused broadening.
10-fold enhancement; 69Ga and
71Ga isotopically resolved. RIS
experiments concluded Ca atoms
were in the vapor state and the
effect o f surface bonds was not
important.
Na, Cu
Na0.7WO3
excimer pumped dye
laser, 25 ns pulse
width, 10 Hz, > 1.3 x
106 W cm'2
Cd, Cu,
Mo
thin film metal
doped polymer,
steel alloy
Nd:YAG pumped
dye laser, 15 ns pulse
width, 6 x 108 W cm'2
Cd: 228.8 nm, RLA found for 1071010 W cm'2; Cu: 324.75,
327.39 nm; Mo: 311.21,
313.26 nm; 0.4 to 0.7 nm
FWHM.
TOF-MS
Al, Ga,
Ca
AlGaAs
semiconductor,
Ca metal
sample
excimer-pumped dye
laser; 10 W cm'2.
Al: 308.297 and 309.367 nm; Ga:
618.644 nm; Ca: 600.125 nm. <50
pm FWHM.
TOF-MS
Al
NIST SRM
steel samples
Nd: YAG pumped
dye laser; 10 ns pulse
width; 107 W cm'2
Al: 308.29 and 309.37 nm; 50 pm
FWHM
TOF-MS
At low laser powers, on-line ion
signal is ~104 higher compared
to off-line. The plume was
believed to continue after the
laser pulse. LODs o f 5 ppm Al
Ga
GaAs
semiconductor
excimer-pumped dye
laser; 5 ns pulse
width; ~106 Wcm"2.
287.5 nm, one-photon process, 0.1
nm FWHM; 618.6 nm, two-photon
process, 15 pm FWHM.
TOF-MS
Ablation efficiency increases
with sample angle, but also leads
to broadening.
ref.
12
7, 26
28
8
29,14
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
element
Eu, Lu
lanthanide
oxide, Ln20 3
and hydrated
oxalate
Ln2(C204)3
laser parameters
excimer-pumped dye
laser; 12 ns pulse
width; 20 Hz; 107 109 W cm 2; 20 ps
delay.
ablation wavelength
& ablation scan spectral features
46 6 ,4 6 2 ,4 5 9 nm for Eu; 465, 451
nm for Lu; ~0.8 nm FWHM;
collisional and Stark broadening in
the dense plume.
Cu
bulk copper
and thin film
copper
excimer-pumped dye
laser; 12 ns pulse
width; 10 Hz; 0.3
cm'1 band width; <2
x 106W cm'2.
463.507 nm
Si, Fe; Fe,
Ag, Pb
bulk silicon,
rhenium,
copper, and
nickel
excimer-pumped dye
laser; 10 ns pulse
width; 10 Hz; 106107 W cm'2.
Fe: 447.7 nm; Ag: 470.0 nm; Pb:
450.0 nm; 0.3 cm'1 laser bandwidth
limited
to
o\
sample matrix
detection
system
TOF-MS
observations and conclusions
10-fold o f enhancement
observed on-line, compared to
off-line; increasing on-line
power density decreases
observed signal
ref
30
TOF-MS
Rate o f removal decreses with
number o f shots; increase in
energy leads to increase in ion
yield; maximum ion yield
obtained using s-polarized light
on smooth surfaces, and ppolarized light on rough
surfaces.
9
TOF-MS
RLA more dependent on laser
power than 2+1 multi-photon
ionization; gas phas ionization;
LODs in ppb range for all
elements;
10,31
11
32
Mn
manganese
foils, copper,
NIST zinc, and
aluminum
excimer-pumped dye
laser; ~106 W cm'2.
280.9 nm; 560.38 + 280.19 nm
TOF-MS
Incident angle o f 45 ; matrix
impurities effected ion yield;
LODs o f 200 ppb and 50 ppm
in aluminum and zinc,
respectively
Zn
zinc-doped
AlGaAs
semiconductor
excimer-pumped dye
laser; 6 ns pulse
width; 10 Hz; 107W
cm'2.
213.9 nm; FWHM 0.25 nm
TOF-MS
>10 fold increase when on-line
compared to off-line; LOD o f
less than 5 ppm.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
element
sample matrix
laser parameters
"Ni
nickel-plated
rhenium
filaments
nitrogen-pumped dye
laser; 7 nspulse
width; 107 - 108W
cm'2.
Cu
pure copper,
aluminum
sample
eximer-pumped dye
laser; 6 ns; 106W
cm'2
Ba
BaTi03, Ba2Lu-4Ti-.rO,
BaF2
excimer-pumped dye
laser; 12 ns duration;
20 Hz; 106 - 108 W
cm'2
Cr, Fe,
Cu, Ni
NIST stainless
steel sample
excimer-pumped dye
laser; 12 ns pulse
width; 106 W cm'2.
Cu
pure copper,
aqueous
solution o f
copper sulfate
Nd:YAG; 8 ns pulse
width; 1 - 3 Hz; 10s
- 109 W cm'2.
Hi
ablation wavelength
& ablation scan spectral features
471.76 nm
463.86 nm; FWHM 20 pm
455.2 nm
detection
system
observations and conclusions
ref
TOF-MS
2+1 ionization scheme
advantageous for RLA o f
nickel; electronic temperature
calculated to be -2600K
33
TOF-MS
Rapid increase in ionization
yield with increasing power
density, then levels off; LOD
sub-ppm
34
TOF-MS
Resonantly pumping ablation
plume leads to secondary
ionization suggests sufficient
collisional energy
35
Ion Trap MS
RLA for selective metal ion
source for gas-phase molecule
reactions with acetone; RLA
results in fewer total number o f
ions produced
17
RLA LIBS produced longerlived plasma allowing for a
longer integration time thus
producing more sensitive
results.
16
-> 4
473.17 for Cr; 447.5 for Fe; 463.51
for Cu; 471.7 for Ni
532 nm
OES
Watanabe et al. have proposed a theoretical model of this mechanism based on
Verdun’s proposed theory in order to predict RLA’s dependence of sensitivity and
selectivity on the incident laser power [36], The authors contend that this model can be
used to quantitatively predict resonant and non-resonant yields from RLA. The authors
separate the modeling into three subsections: the vaporization process, the ionization
process, and the space charge effect.
The vaporization accounts for the effect of recoil pressure produced by the
materials emitted early in the high-flux laser pulse, and assumes that the velocities of
the vaporized particles obey the Maxwell-Boltzmann distribution. Also, the
vaporization process accounts for the physical properties of the material including
specific heat, density, thermal conductivity, and temperature. The overall scheme of the
vaporization process is shown in Figure 2.2.
The ionization process is a based on both resonant and non-resonant ionization
components, depending on probabilities of the atom of interest being in ground state,
excited state, and ionization state, as well as, the cross section for two-photon
absorption, the cross section for photoionization from excited state, the stimulated
emission rate, and the photon flux.
28
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
6 ns
X'+ +
X
X+ X-+ +
X
+
X X,
X
X
(A)
+ X
+
(B)
Figure 2.1: A schematic of RLA in which the leading edge of the pulse (a) produces
atoms, ions and particles, while the trailing edge of the pulse (b) resonantly ionizes the
ablated material. Based on the theory of Watanabe et al. [36].
29
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
sample
vapor
the equation o f
heat conduction
the equation for
gas dynamics
the temperature
on sample surface
vapor density near
sample surface
the particle flux
_JV
emitted from
/* —— \ /
sample surface \ j
the particle flux
flowed into
sample surface
Figure 2.2: Schematic diagram of the vaporization process as explained by Watanabe et
a l [36],
30
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
The authors note that while using a time-of-flight (TOF) mass spectrometer to detection
ions produced by RLA, the space charge effect has little influence on the ion detection
efficiency due to expansion of ions by the Coulomb force between themselves. When
applying this theory to RLA o f a stainless steel sample the laser was tuned to a twophoton excitation plus one additional photon for ionization (2+1 scheme) for copper at
463.510 nm, the ion yield curve was similar to that obtained previously in the literature
[37]. The elemental selectivity produced from RLA decreased with increasing laser
power due to an increase in the non-resonant ionization processes, such as collisional
ionization.
Although this model shows comparatively good agreement between
19
9
experimental and theoretical results at power densities less than 2.5 x 10 W m ', it
fails to account for the increase in neutral atom density that has been demonstrated by
optical emission spectroscopy, which will be discussed in Chapter 3, and subsequently
discounts the possibility of any direct resonant laser-material interaction. There is some
physics literature that has been overlooked in this theory, which accounts for lasermaterial interactions and is dependent on the nature of the bulk material.
There has been much speculation on the mechanism by which the resonant
enhancement in the RLA process occurs, but there is no single mechanism that clearly
accounts for all variables that could possibly be involved in the process. Based on the
power density of the incident radiation, it is believed that the RLA process lies between
the LA and LID processes, possibly incorporating details from both phenomena.
Furthermore, different mechanisms exist for each of the three types of materials: metals,
nonmetals, and semiconductors. However, common to all types of materials, the
31
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
antecedents o f desorption include surface defects, excitons, and surface plasmon
resonance.
2.4
Precedents for LID and LA
2.4.1 Surface Defects
Surface defects arise due to crystal imperfections in the lattice structure, which
can be neutral, positively or negatively charged, and result in localization of energy.
The energy of the defect can be located anywhere in the band gap of the material.
Haglund states that defects play three important roles in LID and LA: (1) they
may act as recombination centers for electrons and holes, thus removing electron-hole
pairs from the potential drivers for desorption or ablation; (2) by weakening local
bonds, they may serve as “nucleation centers” for rapid multiplication of desorption or
ablation sites; and (3) they may store absorbed energy at a lattice site for long periods of
time in localized modes, thereby contributing to the possibility o f bond-breaking [19].
LID may be initiated from perfect surface sites, or from steps or kinks near an
otherwise perfect site, or from surface sites which have been altered by other
disturbances, such as the presence of mechanical defects. However, in all these cases,
the spatial density o f ejected particles above the surface is negligible, and there are only
local changes in surface structure and composition, too small to be seen with lowenergy electron diffraction or even in a scanning electron micrograph. Moreover, both
the energy spectra and the dynamics of ejected particles mirror the characteristic of the
particular site from which the particle was ejected [18].
Laser ablation, on the other hand, is invariably associated with large-scale
disruption o f the surface as initially constituted, with substantial alterations in either
32
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
surface composition or geometrical structure or both. Ablation is also associated with
the formation of a plume o f ejected material which can have a significant degree of
ionization which follows the laws of plasma and gas dynamics. The fluences required
for the laser ablation threshold range from 10'2 —10 J cm'2, or an intensity o f 107 - 109
W cm'2, well below the threshold of 1011 W cm"2 at which electron avalanches would
be initiated in a wide-bandgap material.
The structure o f the surface is also a critical factor in the ablation process.
Rough surfaces exhibit a large number of steps, kinks, adatoms, terraces, etc. Because
of the low coordination number, atoms can be released preferentially from these sites.
Gotz et al. explain that at the defects, population of a repulsive electronic energy
surface must be accomplished to initiate desorption [38].
2.4.2 Polarons and Excitons
There are several characteristics involved in the principle elementary excitations
impacted by the coupling of laser to solid, namely, polaritons, polarons, and excitions.
These excitations provide the conceptual basis on which to form a first-principles
picture of optical excitation and relaxation, based on the coupling of these excitations to
phonons. Two types of phonons exist: acoustic and optical. Acoustic phonons are
quantized lattice vibrations or acoustic disturbances. In a monoatomic lattice, phonons
oscillate in phase with each other; however, in a polyatomic lattice phonons may
oscillate in or out of phase with each other. Optical phonons are the out-of-phase
modes, which are infrared active and can be excited by laser light, while excitation of
acoustic phonons occurs via lattice melting [19].
33
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
The polaron is an elementary excitation or quasi-particle consisting of a charge
carrier and its accompanying charge cloud. In weak-coupling materials such as the
compound semiconductors, scattering of free carriers is more probable than smallpolaron formation or trapping, whereas the reverse is true for strong-coupling solids.
Scattering o f free carriers is important to desorption and ablation in two ways. First,
inelastic scattering slows carriers to the point where they may be captured, thus creating
electronic defects, which may lead to desorption or ablation. Second, scattering of
laser-excited free carriers by phonons results in an increased kinetic energy which may
produce rapid local heating; thermalization times required to bring energetic electrons
to the bottom o f the conduction band are of the order 1 ps.
Polaron size is strongly correlated with electron-lattice coupling strength. One
usually speaks of “large” polarons with radii on the order o f 100 - 150 nm in weakcoupling solids, and of “small” polarons with radii a factor o f two or three smaller in
strong-coupling solids. The small polaron appears to be a precursor to the lattice
distortion known as a self-trapped exciton, already implicated in laser-induced
desportion from strong-coupling solids such as the alkali halides and alkaline earth
fluorides. The exciton can be viewed as a hydrogenic atom comprising an electron and
hole, bound together by their mutual Coulombic attraction, modified by the dielectric
constant of the medium and by the effective masses of the electron and hole.
Dependent on excitation energy, excitons and/or electron-hole pairs can be excited.
Self-trapped holes are formed as a result of the stabilization o f a polaron by the Vk
center, an X2 ' molecule localized on the site of a single halogen ion. The electrons and
holes can recombine to produce singlet and triplet excited excitons. These excitons can
34
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
be localized on several perfect lattice sites and they become immobile, at low
temperatures, through self-trapping. In the process of self-trapping, or afterwards, they
can decompose into a pair o f neutral Frenkel defects: an F center (an electron trapped
by a halogen vacancy) and an H center (an interstitial halogen atom in the form o f X 2 ' at
a single halogen site, where X is a halogen). The H center is much more mobile than
the F center and the two defects can separate by diffusion. If the separation distance is
larger than several lattice constants, these defects are stable at low temperatures. At
room temperature, most of these defects recombine to restore the perfect lattice, but
some of them may form more stable aggregates responsible for radiation damage of
these materials [39].
2.4.3 Surface Plasmon Resonance
A surface plasmon resonance (SPR) arises from the coupling of incident
radiation with delocalized electrons on the surface of metals. The excitation quanta are
generically referred to as plasmons. The plasmon is a collective oscillation of the
electron gas, and is also important under certain circumstances, even in semiconductors
where an electron-hole plasma can be formed [19].
By inducing SPR, a considerable field enhancement occurs at the metal surface
resulting in direct ejection o f atoms [40,41], Mechanistically, it is unclear whether this
desorption is a direct result o f surface-plasmon excitation, without coupling to the
phonon bath, or whether the excitation energy is converted to heat to cause a rise in
particle temperature which results in thermal desorption [42]. A roughened surface, i.e.
increased number of defects, can create such conditions because the broken symmetry
allows more efficient coupling of the electric field of the incident light to the surface
35
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
plasmons [43]. Discrete particles can also provide the conditions for the energy
coupling [40,44], The SPR phenomenon is also both wavelength- and polarizationdependent. However, the broad wavelength dependency is not related to the atomic
resonant gas phase transition of the metal. Shea and Compton have demonstrated that ppolarized light best couples to surface plasmons [43], which is similar to results
obtained by Gill using low power RLA of roughened copper surfaces, as noted in Table
2 .1 [9 ].
2.5
Current LID Mechanisms
There are three mechanisms to describe the LID process, all of which explain
the desorption process in terms of metal oxide and organic molecules adsorbed on metal
substrates. In both the long-standing Menzel-Gomer-Redhead (MGR) model [45,46]
and the Knotek-Feibelman (KF) model [47], the desorption is initiated by an incoming
photon by first creating an electronically excited state of the surface complex. If the
potential energy curve (PEC) associated with the excited state is both repulsive and
sufficiently long-lived to survive surface quenching, then “standard” desorption can
occur and this is independent o f the specific nature of the electronic state responsible for
the PEC [48].
2.5.1 Menzel-Gomer-Redhead Model
The Menzel-Gomer-Redhead (MGR) model describes desorption induced by
electronic transitions of an atom or molecule on the target surface by impact of either
electrons or incident photons [45,46]. An incident photon results in removal of an
36
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
electron from one o f the stable levels of the surface species resulting in a charged or
excited species, whose interaction energy curve with the solid may be very different
from that o f the ground state neutral species. This excited or ionized species may have
a shallow attractive potential well or no attractive region at all. The system will then
relax by causing the excited species to progress away from the surface, which reduces
the system potential energy and imparts an equivalent kinetic energy to the excited
species. Without further intervention, the ion or excited neutral will appear in the gas
phase with a kinetic energy range as shown in the Figure 2.3.
Production of a neutral species is possible via Auger or resonance neutralization
process, in which an electron from the surface tunnels into the vacant energy level of
the excited species. This results in the return of the excited species to the ground state
potential energy curve with some kinetic energy.
If this kinetic energy is greater than the potential energy difference between the
vacuum level and that represented by the ground state curve at the point of de­
excitation, then the atom or molecule can escape from the surface as a neutral species.
These processes will be much more efficient for species that are found close to the
surface in the adsorbed ground-state, because the probability of these neutralization or
de-excitation processes is strongly dependent on the distance between the excited
species and the surface.
37
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
+
kinetic energy range
desorption as
0
desorption as
excitation
Figure 2.3: MGR model describing laser induced desorption [45,46].
38
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
2.5.2 Knotek-Feibelman Model
The Knotek-Feibelman (KF) model describes electron- or photo-induced
desorption via core-hole Auger decay resulting in an atom-specific, valence sensitive,
probe of surfaces [47]. In this model, an incident photon creates a core-hole in the
lattice structure by ionization at the metal-core level. This core-exciton then decays
either by way o f an intra- or an interatomic Auger transition. Coulombic repulsion then
leads to atomic emission, where the ion may capture a delocalized electron located on a
surface defect, leading to desorption of a neutral species. This mechanism is similar to
that proposed by MGR, however, the KF model also proposed that an atom leaving the
surface may also undergo a “double Auger” process resulting in production of a positive
ion.
2.5.3 A ntoniew icz Model
Antoniewicz proposed a modification of the MGR and KF models, in which
transitions between bonding states lead to neutral atom or ion desorption from metal
surfaces [49]. This mechanism has been successful in describing electron stimulated
desorption from physisorbed layers. An incident photon or electron results in instant
ionization. The ion then experiences the attractive image potential and progresses
towards the substrate. Electron tunneling from the substrate then neutralizes the ion.
The kinetic energy that the ion had at the time of neutralization is unchanged, which
means that the total energy of the neutral is the kinetic energy before neutralization,
plus the potential energy of the lower curve at the position of neutralization. If this total
39
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
potential energy is greater than the binding energy, then the neutral atom is desorbed
after recoiling from the substrate.
Antoniewicz also proposed a two-electron process to explain the desorption of
positive ions. This is illustrated in Figure 2.4, where the ground-state configuration
(M+A) is promoted to an excited ionic curve (M+A+)*. Similarly, the excited ion
moves toward the substrate for neutralization, placing the system high on the groundstate (M+A) curve. Pauli expulsion then dominates at very small separations, and will
deflect the hot neutral from the substrate, causing it to escape [50]. In this mechanism,
removal o f material destroys the localization o f electrons at the surface, and therefore,
the immediate dispersion o f the energy of incident photons is lost. As the localization
of electrons increase, the partially desorbed atoms in the bulk begin to take on
characteristics of that o f gas-phase atoms; as a result, wavelength dependency will
increase, although broader than that would be expected of atomic vapors. Since the
probability o f electron hopping processes at short distances is finite, the neutral species
may be reionized by resonant electron tunneling into the substrate during its escape,
which yields an desorbing ionic species, as shown by the curve crossing with the
(M+A+) curve (Figure 2.4).
40
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
(M + A)
(M + A)
at ground state
Figure 2.4: Antoniewicz model describing laser induced desorption o f an atom and
possibly a positive ion, where M is the substrate and A is the adsorbate. Here, an
incident photon becomes instantly positively ionized; thus the ground-state
configuration is promoted to an excited ionic curve (M+A+)*. As the ion is attracted
toward the surface, it may pick up an electron from Auger neutralization, whereupon
the neutral atom is ejected from the surface via Pauli expulsion. The neutral species
may also be ionized during its escape, (M+A+) curve, resulting in desorption o f a
positive ion.
41
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
2.6 Resonant Radiation Trapping
Power densities on the order of 107 W cm'2 or greater are typically sufficient to
induce a laser ablated plume with high electron temperature (104 — 105 K) and electron
densities (1015— 1019cm'3) [51]. Upon induction, the plume begins to expand and
cool, resulting in a massive continuum bremsstrahlung emission and ion-electron
recombination [52]. It has been demonstrated that the plasma, including the relatively
cool stage, can survive for up to 25 ps [53]. In the case of low power RLA, the delay
time could be prolonged by radiation trapping in the gaseous phase of analyte atoms, in
which several mechanisms can occur simultaneously.
When excited gaseous analyte atoms decay from a high-energy state to ground
state, the resonance radiation emitted by one atom can be readily re-absorbed by a
neighboring atom in the ground or lower energy state. This cyclic absorption/re­
emission process can occur for quite some time until the photons escape from the
atomic vapor [54], This process is called radiation trapping, and is also referred to as
imprisonment o f resonance radiation, line transfer, radiative transfer of spectral lines, or
radiation diffusion. This phenomenon is markedly enhanced when the analyte atoms
constitute the majority of the solid sample.
When photons are not trapped in the gaseous vapor of analyte atoms, the rate of
excitation, Eexc, is equal to the rate of decay, given by Equation 2.1:
Eexc = A 2in2
Equation 2.1
42
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
where A 2i is the resonant transition from a higher-energy state to ground state and n2 is
the density of the upper state atoms. However, in the case of radiation trapping, the
resonant photon undergoes a number of cyclic emission/absorption processes, given by
the trapping factor, g0, resulting in an increase of the excited state density given by
Equation 2.2:
E exc
= n2A2I/g0
Equation 2.2
Thus, under the condition o f radiation trapping, the excited state density is increased by
a factor of g0 [54],
There are two distinct parameters that determine radiation trapping: (1) the
strength o f the absorption, which is much stronger in RLA than normal LA, which thus
requires less laser energy. However, laser-induced plasma formation occurs on a much
shorter time scale compared to that formed using non-resonant radiation [15]; and (2)
the spectral lineshape of the absorption, which is dependent on the linewidth of the
laser.
At the onset o f the resonant plasma, there are only two dominant processes: (1)
the resonant absorption of a laser quantum by theanalyte and (2) the stimulated
emission of an excited atom due to the irradiationas given by Equations 2.3 and 2.4,
respectively,
A + co -» A*
Equation 2.3
A* +
Equation 2.4
co —> A + 2 co
43
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
where
co is the energy of the resonant state. Associative ionization may occur as a
result o f collisions between resonantly excited atoms, given by Equation 2.5.
A* + A* —» A 2 + e
Equation 2.5
The dominance o f ionization by associative ionization has confirmed that
neutral, excited, and ionized species of the analyte can be observed by varying the time
of observation after the laser pulse into a dense atomic vapor [55,56], After 200 ns,
neutral and excited atoms are observed, with considerable ionization occurring after 400
ns and beyond 600 ns, complete ionization is evident with the disappearance of the
neutral absorption features.
The physical effects of radiation trapping are also more predominant in pulsed
laser experiments where observation o f photon emission is delayed [54]. Radiation
trapping must therefore be considered to have a substantial impact on the mechanism of
pulsed RLA, resulting in a longer-lived laser ablated plume. In fact, lifetimes of laser
ablation plumes of up to 100 ps have been observed due to radiation trapping [57]. This
is very attractive for applications of laser-induced breakdown spectroscopy (LIBS), in
which the measurement o f line emission intensity provides information about the
elemental content of the material [15]. The increased lifetimes would allow for a longer
integration time of the emission signal resulting in a higher signal-to-noise ratio.
However, similar to fluorescence, this lifetime is also dependent on the energy
density of incident radiation. If the excitation pulse is sufficiently strong to saturate the
44
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
optical transition, the emission decay occurs much faster than that of a natural decay.
This is an expression of the radiation trapping in an optically bleached medium, which
leads to nonlinear plasma dynamics and massive bremsstrahlung. If the energy density
of the incident pulse is decreased, radiation trapping and plasma dynamics can be
considered uniform and explained by a linear theory [58].
Once the plasma is formed, and the resonant transitions of the analyte have
reached saturation, the resonant radiation is efficiently transferred into the vast amounts
of free electrons o f the plasma. In addition to plume dynamics, there is also the
interaction between the laser induced plasma and the sample surface as well. Since the
resonant plasma is sufficiently long-lived, the resonantly excited atoms may interact
with the surface resulting in sputtering of analyte atoms near the sample surface,
subsequently resulting in an increase in selective analyte removal from the bulk.
2.7 Conclusion
Since RLA lies between the power density range of desorption and that o f
ablation, it is likely to possess characteristics of both phenomena. O f the mechanisms
and precedents described here, it is apparent that the subtleties of this process are
mainly matrix- and power density dependent with minimal contributions from angle and
polarization. Based on the mechanisms and precedents o f the desorption and ablation
phenomena described here, there are three critical inconsistencies that prevent a
complete description o f the RLA process by any single theory. First, the currently
accepted mechanism proposed by Verdun does not take into account a direct laser-
45
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
material interaction; rather the resonant enhancement effect is thought to occur once the
material has been vaporized.
Secondly, although the MGR and KF models account for the laser-material
interaction, they fail to account to for any plasma-material or plasma-laser interactions
that may exist. These models imply the use of lasers o f very low power densities,
which may be insufficient to initiate the formation of the plasma present on the sample
surface. However, such plasmas are evident in RLA.
Finally, the radiation trapping mechanism may be involved in the resonant
enhancement effect observed; however again, the mechanism does not take into account
a laser-material interaction, which may be critical in the understanding of the RLA
mechanism. Another inconsistency exists since this mechanism may explain the
resonant enhancement at high analyte atom densities, but not when the analyte is
present in trace quantities.
It is evident from the precedents of ablation and desorption, in order to fully
exploit the inherent advantages of RLA, power density and polarization of the laser,
sample angle, and surface condition must be taken into account. The experiments in
Chapter 3 explore these possible mechanisms and precedents using RLA coupled with
microwave induced plasma atomic emission spectroscopic detection. A mechanism
based on the data presented in Chapter 3, drawing on current mechanisms discussed
here, is postulated in Chapter 4, with a critical review of previously postulated
mechanisms based on experiments conducted in this lab.
46
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
3. RESONANT LASER ABLATION MICROWAVE INDUCED
PLASMA (RLA-MIP) PROVIDES INCREASED
ELEMENTAL DETECTION OVER CONVENTIONAL
LASER ABLATION
3.1 Introduction
Traditional methods o f trace metal analysis of solid samples involve the tedious
process of sample dissolution and preparation. In addition to the risk o f sample
contamination, these often labor-intensive methods may require human exposure to a
potentially hazardous sample, as well as hazardous dissolution reagents. Thus the need
for a reliable, rapid, and sensitive sample introduction method, which can accommodate
a wide variety o f solid samples. Laser ablation has the potential of becoming a
universal sample introduction method, which can be coupled to a spectroscopic method
of detection.
Laser ablation refers to the explosive process by which a solid sample is
vaporized though a violent laser-material interaction resulting in a plume o f atoms, ions,
molecules, and clusters. The use of laser ablation (LA) as a form of sample
introduction has been gaining popularity over the past several years due to several of its
attractive features, which include minimal sample preparation requirements, ability to
analyze both conducting and non-conducting materials, localized microanalysis, and
also surface analysis.
Typical LA experiments rely on high-powered, flxed-wavelength lasers to
initiate the massive ejection of atoms, ions, and molecules from the sample surface. In
most experiments, the ions are then directed into a mass spectrometer for detection.
47
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
This non-selective technique has only recently begun to see commercial success,
although it has yet to become a widely used analytical technique due to several factors
including uncontrollable ablation processes and the lack o f matrix-matched standards
for solid sampling. However, it has been shown that an increase in ablation efficiency
and selectivity can also be achieved by tuning the laser wavelength to match that of a
resonant gas phase transition of the analyte, which is the genesis of resonant laser
ablation (RLA). The selectivity of this phenomenon has been demonstrated by the
enhancement of aluminum in a stainless steel sample with respect to other major
constituents as shown by Borthwick et al. using resonant laser ablation mass
spectrometry [59]. They have also shown that lower laser energies are required to
achieve a significant enhancement when the laser is on the resonance line (online)
compared to that obtained off the resonance line (offline). As the energy of the laser
increases both on and offline, the ion signal increases and the efficiency of the ablation
process appears to be similar at energies exceeding 2 mJ.
The RLA phenomenon can be considered as a form of preferential ablation,
whereby fractionation is induced, which has been a drawback to normal LA where
stoichiometric ratios are desired. In doing so, signals of trace constituents may be
selectively enhanced and quantified. There has been extensive research demonstrating
preferential vaporization o f material. Mao et al. have shown that wavelength as well as
power density and pulse width affect the stoichiomertric ratio o f copper and zinc in the
ablation of brass [1]. With a 30 ns laser, where thermal vaporization appears to be the
dominant process, the Zn/Cu ratio approaches stoichiometry at higher power densities,
•j
greater than 0.3 GW cm' . With a 3 ns and 35 ps pulse Nd: YAG lasers, stoichiometric
48
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
ablation can be achieved at higher power densities, although both thermal and
nonthermal processes exist using the former, while thermal processes seem negligible
using the latter. They have also shown that UV ps lasers provide greater emission
intensity and wide fluence range in order to obtain accurate stoichiometric analysis.
However, ps lasers are not as reliable and stable as the ns YAG laser.
Cromwell et al. have determined that fractionation is prevalent for ablation at
low laser fluence (< 1 1 cm"2) and with multiple overlapping laser pulses incident on the
same area of the sample surface [2]. The authors contend that it is possible, although
impractical, to predict which element may be preferentially ablated upon inspection of
solid-liquid binary phase diagrams for homogeneous materials, given the approximate
composition o f the sample the thermodynamic properties of the analyte and matrix
species, relevant phase diagrams, and knowledge of the heat-affected zone. Rather, due
to lack of matrix-matched standards, the authors suggest to use instead a higher fluence
and avoid sampling previously ablated areas.
Eggins et al. have shown, using LA-ICP-MS, that during fractionation the more
refractory elements are condensed onto the surface before the more volatile. Also, a
lower mass density o f the carrier gas gives an enhanced signal in ICP-MS.
The effects o f laser polarization and surface finish versus atomic or ionic signal
are well known in the resonant laser ablation literature. Allen et al. have shown that
smooth surfaces produce a maximum signal for p-polarized light, whereas rough
surfaces produce the maximum signal for s-polarized light [60].
The data presented here illustrate the enhancement effect of trace metals in stainless
steel samples, and pure copper and aluminum samples in both ablation laser wavelength
49
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
scans and emission scans. A microwave-induced plasma (MIP) was used to detect the
atoms produced. Also, optical and SEM images are presented which exemplify changes
in surface morphology as the ablation laser wavelength approaches the resonant
wavelength of the target material, and which lend insight into possible mechanisms.
Interpretations o f the data presented here contradict the currently accepted two-step
mechanism that has been set forth by Verdun et al. to describe the RLA phenomenon.
Similar to the RIS theory, they describe the laser pulse in terms of the leading edge of
the pulse and the trailing edge. It has been proposed that the leading edge of the pulse
is responsible for non-selective normal ablation, while the trailing edge resonantly
ionizes the low-density plume of ablated material. However, in contrast with ion signal
enhancement, we have demonstrated the enhanced generation of atoms using detection
by optical emission spectroscopy. Also, surface characteristics of the craters indicate a
possible enhanced laser-material interaction, which is also not covered in the accepted
theory.
3.2 Experimental
A schematic diagram of the experimental arrangement is shown in Figure 3.1 for
resonant laser ablation coupled with MIP atomic emission detection. The ablation
chamber is similar to that described by Uebbing et al. [61]. The Beenakker MIP cavity
was water cooled, and the plasma was sustained by low-pressure argon gas between 500
mTorr and 2 Torr [62]. The sample was inserted into the chamber from below and was
mounted to a stepper motor. The holder is off-centered with respect to the center of the
cavity by 5 mm so that it could be rotated to obtain a fresh sample surface. The
50
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
distance between the sample surface and the bottom o f the Beenakker cavity was about
2 to 3 mm. A second sample holder, which will be described in detail later, was
interchanged to allow for control of the sample angle and can be rotated to change the
sample orientation with respect to the laser polarization. The glass plasma tube with a
silica window was inserted into the cavity, and the plasma tube and the glass cell were
sealed to the cavity with viton 70 O-rings. Ultra high purity argon gas was introduced
tangentially into the glass cell to sweep vaporized material into the microwave cavity
and also acted as the plasma support gas. A roughing vacuum pump was used to
provide a low argon pressure, which was about 800 mTorr for the experiment. A 2.45
GHz microwave power generator powered the microwave with an operating range from
30 to 50 W. An impedance matching tuner was employed to obtain minimum reflected
power before being transferred into the Beenakker cavity through a coaxial cable. From
previous work conducted in this lab, the microwave power was not found to have a
great affect on the emission intensity. A microwave power o f less than 50 W was used
to minimize heating effects to protect the O-rings. The system was redesigned from
that employed by Yang and Hou [63,64], so that the sample can be easily replaced
without disturbing the MIP, while keeping the integrity of the O-ring seals and allowing
sufficient vacuum to be obtained quickly after sample replacement. The entire ablation
chamber was mounted on a rail system to ensure proper focusing of the MIP with
respect to the detection system, as shown in Figure 3.2.
51
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
tunable OPO laser
ellipsoidal mirror
monochromator
vacuum
microwave cavity
sample cup
sample
argon inlet
I
glass chamber
Figure 3.1: Experimental arrangement for RLA.
52
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
Figure 3.2: The redesigned RLA arrangement allowed for ease of sample replacement
and optical detection alignment. The set screws at (A), (B), and (C) can be loosened to
allow for rough adjustment o f MEP focusing. Once rough adjustment has been
completed, the set screw at (C) can be tightened while fine adjustment was
accomplished using the micrometer mount.
53
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
The 355-nm output of an injection seeded, Q-switched Nd:YAG laser (10 Hz)
was used to pump an optical parametric oscillator (OPO) laser. The fundamental output
wavelength range o f the OPO was 440 - 1840 nm, however, when equipped with a
frequency doubling option, this range was extended down to 220 nm. In this range, the
output energy was o f the order of 1 - 11 mJ/pulse with a linewidth on the order of 2 - 5
pm and a pulse width of 6 ns. The OPO laser system has been previously characterized,
and details about the characteristics of it can be found in reference [65].
The laser beam was first attenuated and then reflected 90° vertically by use of a
right angle silica prism and passed through the hole in the center of the ellipsoidal
mirror before entering through the silica window of the MIP cavity. A biconvex lens,
f/50 cm, was utilized to focus the beam onto the surface of the sample, resulting in a
spot size of 250 pm in diameter. The sample was then ablated and swept into the MIP
cavity, and the emission from the MIP cavity was collected with the ellipsoidal mirror,
and focused onto the entrance slit of a monochromator. A cut-off filter was used to
reduce the scattered and stray laser radiation.
The detection system consisted of a photomultiplier tube (PMT), a preamplifier,
a boxcar integrator, and a computer for data collection. The experimental setup is
shown in Figure 3.3. The samples used were NIST Spectrographic Stainless Steel
Standard (D845), pure aluminum, and deoxygenated copper samples. The constituents
of NIST SSS D845 are listed in Table 3.1. The major components of the experimental
arrangement are listed in Table 3.2.
54
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
lens
attenuator
optical trigger
Nd:YAG
A
prism
OPO
>
pre-amp
PMT
mono­
chromator
ellipsoidal
mirror
cuto f filter
RLA
chamber
generator
boxcar
A/D
^
computer
Figure 3.3: The experimental arrangement for RLA coupled with an MIP for atomic
emission detection.
55
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
Table 3.1: The constituents of NIST SSS D845*
conc.
element
come. (%)
element
cone. (%)
element
(w/w%)
Mn
0.77
Cr
13.31
P
0.2 -0.3*
Si
0.52
Mo
0.92
S
0.01 - 0.02*
Cu
0.065
Nb
0.11
Fe
83.2*
Ni
0.28
C
0.06-0.1
total
100
*homogeneity satisfactory for the certified elements
* uncertified values
56
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
Table 3.2: Major components of the experimental arrangement for RLA-MEP-AES.
GCR 250, Spectra Physics Laser (Mountain View, CA);
Nd:YAG laser
injection seeded, repetition rate, 10 Hz; output pulsed energy
more than 500 m l at 355 nm; pulse duration, 6 ns.
Quanta-Ray MOPO 730, Spectra Physics Laser (Mountain
OPO laser
View, CA); fundamental output wavelength range 440 - 1840
nm; can be frequency doubled down to 220 nm.
Model 935-3, Newport Corporation (Fountain Valley, CA);
attenuator
aperture 3 x 9 mm; spectral range 200 -2 1 0 0 nm.
laboratory constructed consisting of a fiber optic cable fitted to
optical trigger circuit
a side-on IP28-type PMT (Hamamatsu, Japan) mounted in a
LD-R2 housing (CVI Laser Corp., NM) driven at -300V.
Custom made by Aero Research Assoc., Inc. (Port Washington,
ellipsoidal mirror
NY); 90 off-axis; diameter, 89 mm; fi = 140 mm, fi = 260 mm;
aluminum reflective surface with magnesium fluoride overcoat.
Model 218, McPherson (Acton, MA); focal length: 1/3 meter;
monochromator
1200 groove; bandpass, 0.4 nm; wavelength range, 105 - 1000
nm.
57
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
PMT Housing Model 0893-492, Products for Research, Inc.
(Danvers, MA); PMT Tube Model 9893Q with quartz window,
photomultiplier
and 8 mm cathode, (EMI). ~25% quantum efficiency in 350 420 nm range.
Housing fitted with pulsed dynode chain for
operation with laser detection [66].
Model W 1 0 0 BTB wide band, LeCroy Corp. (Chestnut Ridge,
preamplifier
NY)
Models SR250, SR280, and SR275, Stanford Research Systems
(Sunnyvale, CA), for integrator, system mainframe, and display
boxcar integrator
modules, respectively; or Model 162 with gated integrator 165,
Princeton Applied Research (Princeton, NJ)
Model P5-60, Gateway 2000 (North Sioux City, SD), software
computer
written in Microsoft VisualBasic
Laboratory constructed based on design by Beenakker [62],
microwave cavity
Aluminum housing with 91 mm inner cavity diameter.
Model Microtron 200, Electro-medical Supplies, (Greenham)
microwave power
Ltd. (Wantage, Oxfordshire, England) 200 W maximum output
generator
power.
58
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
microwave tuning
Model 1878B, Maury Microwave Corporation (Ontario, CA)
stubs
flow meter and flow
Model 1355-8506, and Model 8802, respectively; Brooks
controller
Instrument Division, Emerson Electric Co. (Hatfield, PA)
pressure gauge
Series 275, Granville Phillips Co. (Boulder, CO)
59
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
3.3 Results and Discussion
3.3.1 Optical and SEM Crater Images
■In order to determine whether or not a wavelength-dependent laser-surface
interaction existed, craters were formed on a pure, deoxygenated copper sample by
ablating for 1000 shots at 0.75 mJ per pulse (2.5 x 107 W cm'2) over a 1-nm wavelength
range encompassing the resonant gas phase transition of copper, from 324.212 to
325.212 nm. These craters, 250-pm in diameter, were then imaged using both an
optical microscope as well as a scanning electron microscope (SEM).
From the optical images, shown in Figure 3.4, an outer ring of re-deposited
material is formed as the ablation laser approaches that of the resonant transition of
copper, 324.754 nm, thus demonstrating the increased efficiency o f online ablation.
This ring dissipates as the wavelength o f the ablation laser progresses beyond the
resonant gas phase atomic transition. This ring of re-deposited material was verified
using the line scan function o f SEM, which indicates a build-up o f material around the
crater, as shown in Figure 3.5. It should also be noted that the center o f the crater
cannot be considered a true crater, but rather a disruption in the morphology o f the
sample surface.
Upon closer inspection of the center of these craters using SEM, as shown in
Figure 3.6, distinct changes in morphology are evident as the resonant line of copper is
approached. At 500 pm offline, small beads of copper are observed, appearing more
wave-like at 250 pm and 50 pm offline, and finally, significant melting is observed at
the resonant line, 324.754 nm. This may indicate that the RLA process is the result of a
thermal mechanism. However, from both sets of photos, it is evident that the RLA
60
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
mechanism may involve a wavelength dependent interaction between the laser and the
sample. This phenomenon is inconsistent with the current accepted mechanism, which
suggests the resonant excitation occurs as a purely gas-phase process. Alternatively, or
in addition, the formation of the laser-induced plasma, which was observed to be larger
and brighter as the laser wavelength approached that of the resonant line, may be
responsible for re-deposition of the material observed around the craters, as well as the
increased melting as noted in Figure 3.6.
The efficiency of the RLA process was then compared to that achieved 500 pm
offline by the observation of craters formed under various laser power densities. The
optical images in Figure 3.7A indicate that at high power densities, using 2-mJ laser
energy per pulse, which corresponds to 7 x 107 W cm’2, the ablation efficiency appears
to be roughly equal, with both craters exhibiting the outer ring of copper re-deposition.
However, as the power density was decreased using an optical attenuator, to 1 mJ per
pulse (corresponding to 3 x 107 W cm'2) the thermal effects are markedly higher for the
online case. This is evident in the offline case where disappearance of the outer ring is
noted, represented by Figure 3.7B. As the power density is further decreased to 2.5 x
107, 2 x 107, and 9 x 106 W cm'2, Figure 3.7C, D, and E, respectively, the thermal
effects remain higher than those of the offline counterpart.
61
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
324.254
324.729
324.504
324.742
324.804
324.654
324.754
324.854
324.704
324.764
325.004
324.779
326.254
Figure 3.4: Optical images o f craters formed on pure deoxygenated copper samples. As
the laser wavelength approaches that of the resonant atomic gas phase transition of
copper, 324.712 nm, an outer ring o f re-deposited copper is observed.
62
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
a— ww
■
■
Figure 3.5: Line scan SEM image of crater formed on pure deoxygenated copper sample
using 1000 shots o f 0.75 mJ per pulse energy at resonant wavelength of copper, 324.712
nm .
63
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
Figure 3.6: SEM images o f the center of the craters shown in Figure 3.4 reveal changes
in surface morphology o f the center of the craters at (A) 500 pm offline (324.254 nm),
(B) 250 pm offline (324.504 nm), (C) 50 pm offline (324.704 nm), and (D) online
(324.754 nm).
64
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
offline ablation (324.212 nm)
(A)
(B)
a
<§>
(Q
(D)
(E)
online ablation (324.712 nm)
Figure 3.7: Optical images of craters formed on a pure, deoxygenated copper as a
function o f power density both on- and offline. The top series o f crater images were
produced from ablating 500-pm offline, 324.212 nm, and online at 324.712 nm,
represented by the lower series of crater images at various power densities. See text for
details.
65
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
The apparent thermal effects as demonstrated in the SEM images may be a
result o f the radiation trapping effects, as previously discussed in Chapter 2. The
increased size and lifetime o f the laser-induced plasma, confirmed by Tsipenyuk and
Davydov [16], leads to increased plasma temperatures, which may result in back
heating, localized heating directed back to the sample surface. Furthermore, increased
vaporization due to the increased temperature of the plasma may lead to increased
deposition of metal as seen in the optical images as rings of re-deposited material.
3.3.2 Spectral scans
Similar to fluorescence measurements, two types of scans were obtained:
emission wavelength scans and ablation wavelength scans. The emission scans were
acquired by ablating the material at a fixed wavelength, while the monochromator was
scanned through a small range of wavelengths, encompassing the atomic emission line.
The ablation scans were recorded in a similar manner by setting the monochromator to
the atomic emission line, while scanning the ablation laser through a small range of
wavelengths. For these experiments, the delay time was optimized, which is the time
span between the firing o f the laser pulse and the detection of the atomic emission
signal.
3.3.3 Delay Time
A simplified Grotarian diagram for aluminum is shown in Figure 3.8. For a pure
aluminum sample, the resonant line at 308.216 was used for resonant laser ablation,
while the atomic emission was measured at 394.403 and 396.153 nm. For the delay time
scans the atomic emission at 396.2 nm was observed across a range of delay times after
66
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
the laser pulse. Figure 3.9A indicates that the aluminum emission signal is present up
to 15 ms after the resonant ablation laser pulse when the MIP was not employed and at
least up to 20 ms when the MIP was switched on, Figure 3.9B. Since the pulse duration
o f the laser is only about 6 ns, it can be concluded that the aluminum atoms are re­
excited after being ejected from the sample surface due to collisions with the excited
argon species in the MIP, resulting in a signal enhancement of about a factor of three.
Radiation trapping may also play a dominant role, especially in this case where the
target material is a pure sample. The concentration of vaporized aluminum atoms is
sufficient enough to entrap resonant radiation, initiating the cyclic absorption/emission
process, which inevitably would result in a longer-lived signal.
The delay time was then observed 1-nm off the resonant atomic transition of
aluminum at 307.259 nm. Figure 3.9C indicates a much shorter-lived atomic emission
compared to its online counterpart, Figure 3.9A. However, as in the online ablation
case, when the MIP is turned on, the lifetime of the atomic emission is drastically
increased. The differences in emission lifetime and intensity between Figure 3.9A and
C might indicate an increased production of atoms when the laser is tuned to the
resonant gas phase transition of aluminum, or a more energetic ablation plasma,
possibly resulting from radiation trapping as discussed previously. Radiation trapping
might be effective at the high atom densities that might result from ablation of pure
aluminum but might not be effective at low analyte atom densities that might result
from the resonant ablation of trace species. This increased lifetime o f atomic emission
shown here agrees with the previous work o f Tsipenyuk and Davydov [16]. It is not
clear whether the resonant enhancement effect is a result of production of more atoms
67
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
or radiation trapping. Another possibility stems from resonant excitation to high lying
Rydberg states, which would take longer for the decay of atoms to cascade to the lower
energy levels being observed.
68
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
30784
25348
VO
«N
00
O
CO
wo
t- H
CO
o
vd
ON
m
CO
Os
CO
112
Figure 3.8: A simplified partial transition diagram of the aluminum atom.
69
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
pj 0.6
0.4 C4
<N
0 .8 -
VO
)0 0.2
& 0 .4 ta
*
0
■ rH
Os
5
10
15
20
delay time (ps)
delay time (us)
2.5
.1
6.0
l
4 .0 -
CG
-
m
<D
0
5
10
15
20
20
delay time (|is)
delay time (fas)
Figure 3.9: Optimization o f delay time after laser pulse using (A) RLA with the MIP
off; (B) RLA with the MIP on; (C) non-resonant ablation with MIP off; and (D) non­
resonant ablation with MIP on. See text for details.
70
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
3.3.4 Emission Spectra
While the ablation laser was held at a fixed wavelength, the atomic emission of
the sample was detected by scanning the monochromator through a 5-nm wavelength
range, at a rate of 1 nm min'1, encompassing the analyte atomic emission line and
possible close-lying emission lines of other constituents. Based on the delay time
optimization previously discussed, a delay time of 120 ns was employed with a gate
width o f 30 ns. A monochromator slit width of 15 pm with a PMT voltage of -1600 V
gave the highest signal-to-noise ratio. The optimization of these parameters is shown in
Figure 3.10, where a pure aluminum sample was ablated at the resonant line, 308.259
nm, while an emission wavelength scan was recorded. A laser energy o f 2 mJ per pulse
was used, corresponding to a power density of 7 x 107 W cm'2, which was focused on
the sample to result in a 250-pm spot size.
71
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
M
u 393.5
394.5
395.5
396.5
397.5
398.5
393.5
394.5
wavelength (nm)
395.5
396.5
397.5
398.5
wavelength (nm)
<u
.9
S3
O
■ *“ i
393.5
394.5
395.5
396.5
397.5
398.5
393.5
394.5
wavelength (nm)
395.5
396.5
397.5
398.5
wavelength (nm)
33
i'
S3
O
<D393.5
394.5
395.5
396.5
397.5
398.5
wavelength (nm)
Figure 3.10: Optimization o f slit width and PMT voltage. (A) 25 pm and -1900 V, (B)
25 pm and -1800 V, (C) 20 pm and -1700 Y, (D) 15 pm and -1700 V, (E) 15 pm and
-1600 V. See text for details.
72
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
3.3.4.1 Pure Aluminum Emission Scans
Atomic emission scans on a pure aluminum sample were obtained at the
aluminum resonance line, 308.216 nm, and 500 pm offline, at 307.716 nm.
For these
experiments, the laser was directed normal to the sample surface for each of these
wavelengths, while the monochromator was scanned through a five-nanometer range to
encompass both emission lines at 394.403 and 396.153 nm. Although the laser
wavelength difference between both scans is only 500 pm, the resonant enhancement,
shown in Figure 3.11, is demonstrated to be an increase of a factor o f 8 and 13 for the
emission lines at 394.403 and 396.153, respectively. The ratio of the intensity
difference between the two emission lines, obtained while ablating online, is roughly
what would be expected based on the atomic transition probabilities [67].
73
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
relative emission intensity (a.u.)
3.53.0- — offline @ 307.716 nm
— online @ 308.216 nm
2.5-
(A)
2. 0 -
1.51. 0 0.50.0 n
393.3
394.3
395.3
396.3
397.3
398.3
wavelength (nm)
Figure 3.11: Emission scans of pure aluminum sample while ablating at the resonant
atomic (A) and 500 pm offline (B).
74
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
33.4.2
M olybdenum in a NIST Stainless Steel Sample
The resonant enhancement effect was then tested by investigating the emission
signals generated from trace metals in a NIST stainless steel sample. The emission and
ablation wavelengths were chosen based on the simplified Grotarian diagram of
molybdenum, Figure 3.12. Here, the monochromator was scanned through the range
from 395.10 to 400.10 nm, while ablating at a series of wavelengths around the
molybdenum wavelength. As shown in Figure 3.13, several elements in the stainless
steel sample are evident in the emission scan, as these elements emit in this wavelength
range, although at a much lower intensity.
The molybdenum emission in the stainless steel was observed at a series of
ablation wavelengths. In Figure 3.14, the molybdenum emission signal decreased as the
ablation wavelength progressed further offline in 500-pm increments towards the red,
shown in panels A through D.
75
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
31913
26321
25872
25612
Figure 3.12: A simplified partial transition diagram of the molybdenum atom.
76
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
^
S3
1.4
S
1.2
a<D
Mo (I)
0.92%
Co (I)
2.90%
.a
v (0 Cr (I)
3.04% 7 .8 2 %
0.6
<D
<U
>
0.4
0.2
313
»h
397.5
398.5
399.5
400.5
401.5
402.5
wavelength (nm)
Figure 3.13: Emission scan of molybdenum in a NIST stainless steel sample while
ablation at the resonant atomic transition of molybdenum, 313.259 nm.
77
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
1.4
1.4
7 7 0.4
&
-
398.5
399.5
400.5
401.5
397.5
402.5
C
398.5
399.5
400.5
401.5
402.5
wavelength (nm)
1.4
1.4
(D)
0.6
0.4
-
-
0.2
0.2
397.5
398.5
399.5
400.5
401.5
402.5
397.5
wavelength (nm)
398.5
399.5
400.5
401.5
402.5
wavelength (nm)
Figure 3.14: Molybdenum in stainless steel emissions scans using (A) RLA at 313.259
nm; (B) 500 pm offline at 313.800 nm; (C) 1 nm offline at 314.800; and (D) 1.5 nm
offline at 314.800 nm.
78
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
3.3.5 Wavelength Ablation Scans
Wavelength ablation scans were obtained by holding the monochromator at the
fixed emission wavelength, while scanning the ablation wavelength through the
resonant wavelength of molybdenum for a stainless steel sample and over a range of 3
nm. These scans, shown in Figure 3.15A through C, were obtained using 4 m l , 1.8 mJ,
and 1 mJ, respectively. It should be noted that as the laser energy decreased, the
resulting peak became narrower. The full width at half maximum value of the ablation
wavelength scans is about 1 nm for the 4 mJ pulse energy and narrows to about 500 pm
for the 1.8 mJ pulse energy. As the laser energy decreases, the resulting peak becomes
narrower indicating a decrease in non-resonant ablation at the wings of the spectral
profile. As the laser energy is further decreased to 1 mJ, the spectral features are
reduced to a series o f emission spikes, with the higher intensity spikes being centered
around the molybdenum resonant wavelength.
An overlay o f the molybdenum emission intensity at various ablation
wavelengths, Figure 3.14, and the ablation wavelength scan, Figure 3.15A, reveals good
correlation between the data obtained by these two methods. This is demonstrated in
Figure 3.16.
When the laser beam couples with the surface of the material, the interaction is
generally believed to be broadband; therefore distinct spectral differences at the
resonant line are thought to be highly unlikely. However, the data presented here
clearly demonstrate that there does exist spectral enhancement at the resonant line at
moderate laser energies. Based on the increased atomic emission intensity at the onset
of normal ablation as shown in Figure 3.9C, it is a possibility that material is being
79
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
removed from the surface of the sample resulting in the formation o f defects such as
steps and terraces, as discussed in Chapter 2. Located on the edges o f these defects may
exist partially desorbed analyte atoms that have available orbitals, which can interact
with the resonant photons. As an atom begins to desorb it may progressively increase
its resonant interaction with the laser beam and the rate of desorption may then increase
to the point where more efficient desorption takes place under resonant conditions [50].
This would lead to more analyte atoms being removed from the surface.
At higher laser power, nonresonant processes dominate which may explain the
increase in FWHM o f the ablation profile with increased energy. Dai et al. have
demonstrated a similar effect when employing RLA on a copper sample using TOF-MS
[34]. As the laser power is decreased, the RLA process dominates, which agrees with
results obtained by Borthwick et al. in the RLA of pure aluminum samples [59],
However, ablation wavelength scans of pure samples were much more difficult
to obtain. Here, a pure aluminum sample was ablated while scanning the laser from
307.7 to 308.7 nm, encompassing the resonant gas phase transition for aluminum at
308.216, while detecting the atomic aluminum emission at 396.2, according to the
Grotarian diagram in Figure 3.8. These ablation wavelength scans are shown in Figure
3.17A, B, and C as a function of laser energy corresponding to 1,0.75, and 0.5 mJ,
respectively. As the laser energy is decreased, the emission signals at 307.8 and 308.6
decrease, while the aluminum atomic emission signal at 308.2 remains. However,
further studies must be conducted at lower energies and a wider scan range to explain
both the spurious emission signals at 307.8 and 308.6 nm and also the linewidth of the
80
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
aluminum emission peak, which is much narrower than the molybdenum emission
in the stainless steel samples, shown in Figure 3.15.
81
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
10
3
(A)
-
311.8
312.3
312.8
312.3
312.8
313.3
313.8
314.3
314.8
'iS
oo
as
t>
CO
15
*
.aPS
o
W
W
• f»H
Q-
<D
<D
>
■t----------- 1------------ 1—-------- r
311.8
313.3
313.8
314.3
314.8
3TS
1
311.8
“T
312.3
— i-----------f-----------s----------- r
312.8
313.3
313.8
314.3
314.8
ablation wavelength (nm)
Figure 3.15: Ablation wavelength scans of molybdenum in stainless steel samples at (A)
4mJ, (B) 1.8 mJ, and (C) 1 mJ.
82
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
2.5
-1 6
-
12
Mo @ 398.3 nm
Mo/Co emission
— resonant laser ablatioq
0.5 -
313
313.5
314
314.5
315
ablation wavelength (nm)
Figure 3.16: An overlay o f molybdenum emission intensities obtained at various
ablation wavelengths (blue) and molybdenum ablation wavelength scan (red).
83
Reproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
3.0
(A)
2.5
2.0
1.5
1.0
0.5
0 4—
307.7
<n
g
307.9
308.1
308.3
308.5
308.7
307.9
308.1
308.3
308.5
308.7
307.9
308.1
308.3
308.5
308.7
1.6
307.7
1.0
0.8
0.6
0.4
0.2
04—
307.7
ablation wavelength (nm)
Figure 3.17: Ablation wavelength scans of pure aluminum samples at (A) 1 mJ, (B)
0.75 mJ, and (C) 0.5 mJ.
84
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
3.3.6 Sample A ngle and Polarization
It has been previously demonstrated that sample angle [68] and polarization [9]
may have a dramatic effect upon the observed ion signal detected. However, only a 3°range o f angles was included and these variables were never altered simultaneously
within the same experiment. In order to study the effects of these variables, the sample
holder previously used was replaced with one that would allow precise angle control.
The sample cup was mounted on a hinge and a shaft was inserted through the sample
holder fitted with an O-ring in order to keep the integrity of the vacuum. The shaft was
fitted with a calibrated micrometer to allow precise angle control from 0° to 90°, shown
in Figure 3.18.
85
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
feed-through
shaft
hinged sample cup
micrometer
Figure 3.18: The variable angle sample holder consisted of a feed through shaft
mounted to a micrometer.
86
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
Since the polarization of the laser beam is >98% horizontally polarized, it has
been experimentally determined that rather than changing the polarization of the laser,
the orientation of the angled sample can be changed, thus effectively changing the
polarization of the laser relative to the sample. This is illustrated in Figure 3.19, where
A - H represents the position of the sample with respect to a fixed point on the ablation
chamber and the subscripts h and v represent the polarization of the laser, horizontally
and vertically, respectively. O f the possible combinations of sample orientation and
laser polarization, only three cases result: orthogonal (Ch = Gh - Av = Ew), parallel (Ah =
Eh = Cv = Gv), and 50% of each (Bh = Hh = Dv = Fv = Dh = Fh = Bv = Hv).
87
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
Figure 3.19: The possible combinations of sample orientation with respect to laser
polarization. The net result of these combinations leads to only three cases: orthogonal
(Ch = Gh = Av = Ey), parallel (Ah = Eh = Cv = Gv) and 50% o f each polarization (Bh = Hh
= Dv = Fv = Dh = Fh = Bv = Hv).
88
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
If in fact the laser polarization did play a role in the emission intensity produced,
one o f two cases would result: either the polarization that is orthogonal to the sample
surface would produce a maximum, or conversely, the parallel polarization would
produce a maximum. However, if laser polarization has no effect on the efficiency of
the ablation process, there will be no change in emission signal as a function of sample
position. These three cases would produce an emission profile vs. polarization plot that
would look like a sine wave, as shown in Figure 3.20.
Experimentally, when the laser is tuned to the resonant gas phase transition of
aluminum, the emission profile produced a maximum using the orthogonal polarization,
as shown in Figure 3.21. When the laser is directed at the surface at a 15° angle, the
emission profile closely follows the theoretical sine wave, as given by the solid line in
Figure 3.20. As the sample angle is increased, the emission intensity increases
indicating a more efficient laser-material interaction. It should be noted that when the
sample angle is 60 , the emission scan at position G and H could not be acquired due to
limitations in the cell geometry. Since the sample holder is off-center with respect to
incident laser radiation, certain orientations of the sample resulted in the laser pulse
striking the sample cup in which the sample is placed. Also because of cell limitations,
the emission intensity versus sample position is not a true sine wave as proposed, which
can be due to sample transport into the microwave cavity. However, it is evident from
this figure that an increase in sample angle and orthogonal polarization enhances RLA
efficiency.
89
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
12
53
CS
10
(U
8
S3
6
O
'55
ra
4
a>
>
I13
2
0
A
B
C
D
E
F
G
H
sample position
Figure 3.20: Theoretical plot of the sample position versus emission intensity. The
dotted line represents emission intensity trend if the emission produced by the laser
beam being orthogonal to the sample surface were a maximum, the solid line represents
the trend that would occur if the parallel polarization were a maximum, and the dashed
line represents no change in emission intensity with sample position.
90
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
1 4
- p .-
A
B
C
D
E
F
G
H
sample position
Figure 3.21: Effects o f sample angle and position with respect to polarization, while
ablating online, 308.216 nm. All points were obtained at the same sample position, but
some o f them are not shown, in order to clarify the graphs.
91
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
When the ablation laser is tuned 500 pm offline, 307.716 nm, the emission
spectra produced with respect to sample orientation and angle follow the same trend,
however, the differences between maxima and minima are significantly reduced, as
shown in Figure 3.22. It appears as though the observed change in signal occurs as a
function o f polarization both on and offline, which indicates that the mechanism of the
resonant desorption is not a polarization dependent phenomenon. Indeed, the
polarization-dependent ablation phenomenon is superimposed on both non-resonant and
resonant ablation. Although there may exist desorption contributions due to surface
plasmons resonances, which are polarization dependent, they are broadly wavelength
dependent, on the order of 100 nm [41], therefore may have no primary contributions to
the RLA phenomenon, where only a 3-nm ablation wavelength is employed.
92
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
2.5
►*0 degrees
15 degrees
2s-30 degrees
►-45 degrees
is- 60 degrees
cj
03
53
a
.s
53
.2
’ as
as
to
<o
_>
ts
B
B
G
D
H
sample position
Figure 3.22: Effects o f sample angle and position with respect to polarization, while
ablating 500 pm offline, 307.716 nm. All points were obtained at the same sample
position, but some o f them are not shown, in order to clarify the graphs.
93
Reproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
3.4 Mechanisms o f Resonant Laser Ablation
Several theories and models have been proposed to explain resonant
enhancement effects in RLA, as outlined in Chapter 2. However, there exists no
satisfactorily convincing mechanisms that can explain RLA processes completely, and
none of these proposed mechanisms have been fully explored experimentally. It
appears that many factors, such as laser wavelength, power density, pulse duration,
polarization, characteristics of analyte elements, and sample matrix, affect laser ablation
mechanisms.
Verdun et al. proposed that RLA is a multi-step process in which the leading
edge of the pulse generates a low-density plume of ablated material. This gas phase
plume is then resonantly excited/ionized by the trailing edge of the pulse, thus
selectively generating ions [7]. This theory infers that the enhancement effect occurs in
the gas phase. If this generally accepted theory were accurate, the craters produced
when the ablation laser is online would be very similar to those produced offline.
However, the crater photos shown here appear to be drastically changing as the ablation
laser wavelength approaches the resonant gas phase transition, therefore, the lasermaterial interaction and other possible processes must be explored in any proposed
theory. In trace metal analysis, the resonant enhancement is clearly observed, which
implies a resonant interaction at the sample surface, however the resulting craters are
expected to appear different in a macroscopic sense.
94
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
Also, Verdun’s theory fails to account for the increased atomic emission signal,
shown experimentally in this thesis, as opposed to an ion signal, produced when
ablating at the resonant atomic gas phase transition of an element.
The data presented here suggests that a material-laser interaction should be
considered in order to better describe the RLA phenomenon. This has been
demonstrated by the marked increased in atomic emission signal using the orthogonal
polarization while increasing the sample angle. The present author can see no reason
why gas phase atoms and ions would respond to different laser polarizations or changes
in sample angle. However, surface plasmons, which are an angle-dependent
phenomenon, may play a role in the overall and non-selective removal of material.
However, due to the broad wavelength dependence of surface plasmons, it is very
unlikely these resonances play a significant role in the resonant desorption process, as
the general trend o f the polarization vs. relative emission intensity were similar both on
and offline.
Of the desorption mechanisms described in Chapter 2, the Antoniewicz model
may partially account for the resonant effect, although none of the mechanisms
described earlier implicitly address resonant desorption. The role of defects on the
sample surface also supports the theory, in that partially desorbed atoms located at
initial defect sites on the sample surface, and those created by normal ablation, absorb
the resonant incident radiation causing Pauli expulsion [50]. This theory supports the
fact that the excitation scans are relatively broad compared to gas-phase excitation
scans. A mechanism based on the experimental work here is discussed in Chapter 4,
with an overview o f past mechanisms postulated based on work in the RGM lab.
95
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
3.5 Conclusion
The inherent advantages of RLA for its unique sample introduction capabilities
were explored, resulting in increased sensitivity and selectivity for trace metal analysis
using RLA with MDP-AES detection. The data presented here indicates a distinct lasermaterial interaction, substantiated by the changing craters as a function of wavelength
and the relatively broad linewidth of the ablation laser wavelength scans, compared to
gas-phase excitation spectra. The trend in emission signal as a function of laser
polarization was similar for both nonresonant and resonant ablation, which indicates
that the enhancement effect is not polarization dependent and therefore does not
contribute to the RLA phenomenon. The sensitivity and selectivity of RLA make it a
viable alternative to traditional laser ablation sample introduction techniques, by
simplifying emission spectra, which potentially could decrease matrix effects and
interferences. The analytical utility of this technique includes higher sensitivity, while
using lower energies.
In addition to direct RLA, the phenomenon may be probed by conducting matrix
assisted laser desorption and ionization (MALDI) experiments. The principle behind
MALDI is that the analyte, usually proteins or biological samples, is embedded in a
matrix, while the laser is used to desorb and ionized the analyte. Extremely low
energies are required for this process, which makes RLA-MALDI an attractive
alternative. Here, the laser can be tuned to match a resonant gas phase transition of a
matrix component. This may be applied to trace metal analysis, as in the case of
96
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
stainless steel. Here, tuning the laser to that of a major constituent’s resonant line, iron,
while monitoring masses of trace elements may prove to be of value and compared to
that obtained using direct RLA of the analyte.
The author has also planned future experiments to exploit the efficacy of this
technique by applying it to an ICP-MS detection system for simultaneous multi-element
quantification. Appendix 1 contains an outline of proposed experiments and detailed
designs for a novel ablation chamber for use with ICP-MS detection.
97
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
4. RESONANT LASER ABLATION ~ PAST AND PRESENT
4.1 Introduction
Significant progress had been made since the initial RLA work for
determination o f trace metals in stainless steel samples previously performed in the
RGM lab by Yang [63,69] and Hou [64]. Upgrades to the instrumentation, including
the laser ablation chamber design as well as the chamber mount, have enabled greater
flexibility together with ease of sample replacement. These improvements have also
contributed to the reproducibility of the data. However, due to the flexibility, alignment
procedures for both excitation and detection can be quite a cumbersome task to the new
user. Therefore, detailed alignment procedures as well as basic laser operation and
maintenance are outlined in Appendix 2.
This chapter will outline the evolution of RLA’s development in the RGM lab
leading to some major discoveries that changed previous presumptions about the
mechanism by which the resonant enhancement occurs.
4.2 RLA Mechanisms Proposed by Yang
Initial RLA studies performed by Yang, revealed very narrow spectral
linewidths o f the excitation spectrum as the laser was scanned through a range of
wavelengths encompassing the resonant line of a chromium [63]. The resulting FWHM
of the individual peaks was on the order of several picometers, which was believed to
be a function o f the linewidth of the laser, thus indicating RLA’s high selectivity for
analyte excitation/removal. A typical wavelength ablation scan is shown in Figure 4.1.
98
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
53
d
'w'
Mo, Fe, V
<§)
0. 8 -
CO
S
.1 0 . 6 S3
o
0.4
<u
S>
0 .2 -
'S
300
300.5
301
301.5
302
302.5
303
303.5
ablation wavelength (nm)
Figure 4.1: Resonant laser ablation of chromium in a stainless steel sample by Yang
[63].
99
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
The ablation wavelength scans were very irreproducible. From the data, it was
speculated that the laser-material interaction was very selective, thus producing spectra
resembling that of gas-phase spectra. Yang correlated the data with narrow band solidstate photoexcitation as seen in the literature. However, those molecules were adsorbed
on the surface o f alkali halide matrices, to which resonant IR radiation corresponding to
symmetric stretching frequencies, was used to desorb various molecules [70,71].
However, the data obtained by Yang does not fit the model o f the surface of
stainless steel, which can be regarded as metal ions immersed in a sea of electrons.
Classically, the spectral absorption of the metal species is thought to be broad band due
to strong electron-lattice interactions in the body of the solid. If significant
delocalization were present, primarily due to defects, spectra of this type would be
expected at the onset o f resonant ablation, however, a much broader under-structure
would accompany the ablation wavelength scan.
It was postulated that, according to the data acquired by Yang, that normal laser
ablation disrupts the sample surface resulting in localized free atoms, while photons that
may be in resonance with localized atoms on the surface can result in selective removal
of atoms from the surface. Hence, resonant laser ablation is also accompanied by
nonresonant ablation. The species generated by nonresonant ablation would be carried
into the MEP, excited, and emit light, thus resulting in an increased baseline of emission,
as well as an increased signal from the analyte.
Another theory to explain the sharp peaks observed by Yang is that which has
been proposed by Verdun [7]. According to this theory, see Chapter 2, the leading edge
o f the laser pulse indiscriminately ablates the solid sample into the gaseous phase.
100
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
Once these particles are in the gaseous phase, the distance between adjacent particles is
much larger; therefore, the interaction between particles is much weaker. The trailing
edge o f the resonant laser pulse can then be selectively absorbed by the analyte resulting
in an increased population of atoms in the MEP.
Although these proposed theories may account for the observations and trends in
the data by Yang, further efforts by Hou gave further insight into the enhancement
occurs.
4.3 RLA Mechanisms Proposed by Hou
Hou had successfully reproduced similar spectra to those obtained by Yang in a
stainless steel sample, although molybdenum was the analyte and the scan range was
reduced to 1 nm [64]. A typical spectrum is shown in Figure 4.2. According to Hou,
the baseline emission was again attributed to nonresonant ablation, while the resolution
of the peak widths was attributed to the spectral resolution of the detection system.
However, this would only apply to scans obtained by scanning the monochromator,
rather than the laser.
The significant advances from Hou’s work arose from the craters produced as a
function o f wavelength. Here, craters were formed on a pure aluminum sample by
ablating both at the resonant wavelength of aluminum, 308.216 nm, and 500 pm offline,
307.716 nm using both 0.4 mJ and 0.9 mJ pulse energy. The two sets of craters formed
are shown in Figure 4.3.
101
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
crs 1.2
<§>
Cr.N a
312.8
313.2
313.4
313.6
313.8
ablation wavelength (nm)
Figure 4.2: Resonant laser ablation of molybdenum in a stainless steel sample by Hou
[64].
102
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
Figure 4.3: Optical images of craters formed on pure aluminum samples while ablating
(A) 500 pm off the resonant wavelength, 307.716 nm, and (B) on the resonant
wavelength, 308.216 nm using a pulse energy of 0.4 mJ. (C) and (D) show craters
formed on- and offline using a pulse energy of 0.9 mJ, respectively.
103
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
Since the craters had significantly different shapes both on and offline, it is
difficult to quantify the power density, however, distinct differences are evident with
ablation wavelength. Regardless, significant change in crater shape with wavelength
implies a direct laser-material interaction or differences in material deposition during
the ablation event.
In light of this discovery, the proposed mechanism contended that as the laser
wavelength approached that of the resonant line of molybdenum, the nonresonant laser
beam produced defect sites at the sample surface while indiscriminately removing
material, accounting for the background baseline. However, the resonant laser beam
may interact directly with the sample surface selectively removing molybdenum atoms.
The gaseous phase plume, composed mostly of molybdenum particles, can then absorb
resonant radiation from the trailing part of the same laser pulse, which Hou contends is
possibly due to radiation trapping. Hou also asserted that the RLA process was
relatively nonthermal in nature since very low laser energies were used.
This theory takes into account both a direct laser-material interaction, as well as
borrowing from the original mechanism proposed by Verdun [7]. However, as
discussed earlier in Chapter 3, the resonant wavelength response of the surface is
generally thought to be very broad band, with possibly minimal contributions from the
single-pulse ablation/resonant excitation mechanism, as postulated by Verdun.
104
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
4.4 RLA Mechanisms Based on Present Work
After complete redesign of the laser ablation apparatus, in order to eliminate
reproducibility issues as noted by Yang [63], significant effort was made to reproduce
the ablation wavelength spectra previously obtained in this laboratory. It was
determined, after considerable effort, that similar spectral scans were obtainable for the
determination o f molybdenum in stainless steel samples only by significantly reducing
the slit width of the monochromator to 0.15 pm. However, the same irreproducibility,
as seen in the earlier work by Hou [72], was once again observed. The widths of the
peaks were significantly narrower that those recorded by Yang and Hou. It was
concluded that this narrow width was more likely to be controlled by the response time
o f the detection electronics such as the temporal response of the preamplifier, and the
integration time o f the boxcar integrator, rather than the spectral linewidth o f the laser
previously suggested by Hou and Yang. An ablation wavelength scan using a 0.15-pm
slit width is shown in Figure 4.4, which is notably similar to that obtained by Hou,
shown in Figure 4.5.
105
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
/ —■s
relative emission intensity (a), 379.8 nm
d
1.2
0.8
Mo
0.6
0.4
0.2
312.8
313
313.2
313.4
313.6
ablation wavelength (nm)
Figure 4.4: RLA wavelength scans obtained using a 15-p.m slit width.
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
313.
Cr, V
Na
ao
1
‘55
1 0.2
I
I
<D
<L>
>
312.8
313
1--313.2
—
1—
313.4
—
1—
313.6
313.8
ablation wavelength (nm)
Figure 4.5: Unpublished RLA wavelength scans obtained by Hou [72].
107
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
At this narrow slit width, it was determined that the observed spikes were a
result o f emission of light from random clusters of material ejected from the surface,
that just happened to be emitting in the wavelength range of the monochromator
centered at 398.3 nm.
It wasn’t until the ablation wavelength scan range was extended from a 1-nm
range to a 3-nm range, and the slit width was changed to that used in the acquisition of
emission scans, that a reproducible ablation profile was observed (Chapter 3). Here, a
considerably broader absorption response of the surface, as a function of laser
wavelength, is clearly observed, although the maximum emission signal still appeared
to be centered on the resonant wavelength of molybdenum.
In addition to the wavelength ablation scans, rigorous crater studies performed
on pure copper as a function of wavelength and power density reveal a substantial lasermaterial interaction, previously unreported in the literature. Due to these significant
advances, the mechanism o f RLA must be further revised to incorporate these findings.
4.5 Proposed Mechanism Based on Current Work
As discussed in Chapter 2, the MGR, KF, and Antoniewicz models for
desorption of metal atoms, ions, and chemisorbed species are summarized in Figures
4.6, 4.7, and 4.8, respectively.
108
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
excitation to non- or
anti-bonding state
atom becomes repulsive
(Pauli Expulsion)
desorption o f atom
Figure 4.6: Flow diagram o f MGR mechanism resulting in the desorption of an atom.
109
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
ionization of atom
creating exciton
exciton decay
desorption
desorption of an atom
electron capture
desorption of an ion
Figure 4.7: Flow diagram ofK F mechanism resulting in the desorption o f an ion or
atom.
110
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
ionization of atom
surface attraction
electron neutralization
desorption o f an atom
re-ionization
desorption o f an ion
Figure 4.8: The Antoniewicz model for the desorption of ions and atoms.
Ill
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
A revised RLA mechanism, based on much of the aspects of the mechanisms
shown in Figures 4.6, 4.7, and 4.8 for resonant desorption, is shown in Figure 4.9 This
proposed theory takes into account both the laser-material interaction as well as a lasergaseous atom interaction. When the laser is directed onto the sample surface, the atoms
at the surface absorb this radiation, becoming excited. These energetic atoms begin to
spread out, forming a "bubble" of material on the surface, which has been demonstrated
using ps pulse-width ablation on a pure copper sample [73]. This phenomenon,
demonstrated in Figure 4.10, is not wavelength dependent. The atoms located near the
outer wall o f the bubble, which have partial gas-phase atom characteristics, may then
have sufficient energy to escape, resulting in non-resonant ablation, shown in Figure
4.11A. However, the analyte atoms with partial gas-phase characteristics may interact
with the laser pulse when it is tuned to the analyte's resonant gas phase transition,
resulting in Pauli expulsion and therefore an increase in the number of analyte atoms
escaping from the surface, shown in Figure 4.11B [50]. The resonant radiation can then
interact with these gas-phase atoms sending electrons into higher excited states. These
electrons then cascade down from the upper Rydberg states, which can account for the
longer time over which these atoms can be detected compared to normal ablation. In
addition to gas-phase absorption of radiation, the cyclic absorption/emission
phenomenon o f radiation trapping may also play a significant role, especially in pure
samples where a substantial amount o f analyte atoms are readily available for removal.
In the case o f the pure aluminum sample, the concentration of the resultant plasma is
probably significant enough to trap the resonant radiation. This may also account for the
outer ring of re-deposited material as seen in the pure aluminum samples. This is
112
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
substantiated by experiments by Burakov et al. for the sputtering of aluminum from
pure aluminum and aluminum oxide samples using a 308-nm excimer laser, compared
to a 1064-nm Nd:YAG laser, where preferential sputtering was observed by optical
emission spectroscopy [74],
In a pure sample, such as copper, aspects of these mechanisms may be
amplified. Since the bulk, rather than trace amounts, is being resonantly ablated, the
process is extremely efficient such that material is actually re-deposited back to the
surface, as evident in the crater photos. It should also be noted that when ablating a
pure sample, the laser-induced plasma on the sample surface appears visually brighter
as the resonant wavelength is approached. The resonant ablation phenomenon on a pure
sample may result in significantly enhanced radiation trapping, consequently producing
an increased plasma temperature. This may then lead to a back heating effect of the
sample surface, which would explain the significant melting as evident from the SEM
crater images.
113
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
non-resonant
ablation
K
)
n/
partially
desorbed K
atoms absorb
resonant n /
V
radiation
Pauli
expulsion
resonant
interaction
with atoms
condensation
radiation
trapping /
backheating
o
312.3 313.3 314.3
wavelength (nm)
Figure 4.9: Flow diagram o f RLA mechanism based on current work. See text for
details
114
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission.
Figure 4.10: A side view o f the lattice of the sample surface upon laser pulse impact,
115
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
8
(B)
(A)
Figure 4.11: (A) Non-resonant ablation results in fewer desorbed atoms, compared its
online counterpart (B). See text for details.
116
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
4.6 Conclusion
Although the linewidth of the laser or the resolution o f the monochromator may
not limit the resolution of RLA-MBP-AES, as once thought by Hou and Yang, RLA is
very effective for the detection of trace metals in solid samples. The proposed theory,
drawing from both LA and LID mechanisms in the chemistry and physics literature, as
well as the data presented in Chapter 3, provides a new conceptual framework for the
phenomenon of RLA. The proposed theory accounts for a distinct laser-material
interaction, which has been disregarded in the widely accepted mechanism proposed by
Verdun [7] nearly 15 years ago. This laser-material interaction was substantiated by the
ring of redeposited material in the optical crater images and changes in surface
morphology in the SEM images. However, according to Verdun's theory, no changes in
crater morphology as a function of wavelength would be expected.
In addition, Verdun's theory only accounts for the increase in number of ions,
attributing this to a RIS-like mechanism, as previously discussed in Chapter 2. The data
presented in Chapter 3 clearly illustrates an increase in number of atoms, as well. A
survey of the physics literature corroborates desorption of atoms, as well as ions, under
laser irradiation which provides the basis of some possible mechanisms. Although
these mechanisms are used to describe atom and ion desorption stemming from LID, a
much softer form o f LA, they can be applied to RLA, which is on the threshold between
LID and LA.
The data presented in this and preceding chapters illustrate the potential efficacy
of RLA for trace metal analysis. This technique may be applicable to a number of
117
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
analytical methods lending its inherent selectivity and sensitivity. Such techniques
include coupling RLA to ICP-MS, MALDI, surface analysis techniques, and LIBS. A
detailed look at planned experiments for RLA-ICP-MS is found in Appendix 1. The
outlined experiments may also lend useful Information for further evolution of the RLA
mechanisms.
118
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
5. A REVIEW OF RECENT APPLICATIONS OF NEAR
INFRARED SPECTROSCOPY.\ AND OF THE
CHARACTERISTICS OF A NOVEL PBS CCD ARRAYBASED NEAR-INFRARED SPECTROMETER
5.1 Introduction
The vibrational modes of irradiated molecules can be excited by light, which
produces absorption bands in the IR region, normal modes in the mid-IR (MIR) region,
and overlapping MIR vibrational modes, or overtones, in the near-IR (NIR) region. The
broad and less intense peaks in the NIR region, in contrast with the sharp absorption
peaks in the mid-IR region, makes assignments of individual vibrations of a molecule
virtually impossible.
Two basic laws govern vibrational spectroscopy: Hooke’s Law and the Franck
Codon Principle. Hooke’s Law, Equation 5.1, describes the relationship between the
frequency of vibration, v, and a simple two body harmonic oscillator of masses mi and
m2.
Equation 5.1
The fundamental vibrations can be calculated for simple diatomic molecules in the
MIR, however, the NIR region is comprised of combination bands and overtones of the
MIR region at 2, 3, 4, etc. times the fundamental frequency.
The Franck-Codon Principle states that the probability of finding a given atom at
a certain point in a vibrating molecule is inversely proportional to its velocity at this
point. Therefore, atoms spend most of their time in the configuration with the lowest
119
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
kinetic energy, where the potential energy is nearly identical to that of the total energy.
The electronic transition of an absorbed photon occurs faster than the nuclei can adjust
to their new equilibrium position, therefore, the most probable positions of the nuclei
remain in the ground state equilibrium position. Therefore transitions tend to occur
between vibrational levels in which the nuclear configurations are the same in both
states and arise when nuclear kinetic energies are small. It is these small variations,
which give rise to anharmonicity, which result in the combination and overtone bands
that appear at imprecise multiples of the fundamentals.
Aside from the inherent disadvantage of simple band assignment, NIR
spectroscopy can be a very beneficial tool with applications ranging from in vivo
imaging to quality and process control. Virtually all organic compounds have a direct or
indirect absorbance in this region, particularly those with functional groups like hydroxyl, -carboxyl, -amine and carbon-hydrogen. This makes NIR spectroscopy
(NIRS) suitable for the determination of many compounds that vary from volatiles
species of interest for air quality emissions, to plastic identification. A wealth of
information can be obtained from a NIR spectrum including both chemical and physical
properties of the sample. Analysis time tends to be rapid, and there are few sample
preparation requirements; a spectrum may be obtained directly from a solid sample. The
amount of radiation absorbed in the NIR region is small. Reflectance spectroscopy can
be employed for solid samples where a transmittance measurement is not possible.
Although NIR spectral data acquisition is extremely rapid, the data analysis phase
can be very time consuming. Since NIR spectra are generally characterized by a
multitude of superpositions of oscillations, as discussed earlier, a visual evaluation of
120
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
acquired spectra is nearly impossible. Spectral differences between similar substances
may be a very slight shift, or a change in intensity in the wide absorption bands. Due to
the complexity of the NIR spectra, mathematical, and statistical, interpretation
techniques, commonly referred to as chemometrics are usually applied.
5.2 Chemometrics
Chemometrics is the application of mathematical procedures for processing,
evaluating, and interpreting large amounts of data and is not limited to the evaluation of
NIR spectra. Chemometrics is applied to find statistical correlation between spectral
data and known properties of a sample. It also enables the use of multivariate data, in
which patterns can be simultaneously extrapolated from the combined effects of all
associated variables. A major subset of pattern recognition is cluster analysis in which
data are grouped into natural classifications. When dealing with only two or three
dimensions, the analysis is trivial, but when more dimensions are present, computers are
used to aid in the process. A brief outline of the some of the techniques used in typical
chemometric analyses is given below.
Before the data can be processed, pre-processing is usually performed to
determine any baseline drift or slope in a spectrum. Such drift tends to occur in diffusereflectance measurements due to significant light scattering by particulate matter. This
drift is usually alleviated by baseline subtraction that involves a first and/or a second
derivative transformation. The second derivative is most often used, and is
advantageous for several reasons. First, positive peaks in a raw spectrum are converted
to negative peaks; Second, enhanced resolution is achieved to allow separation of
121
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
overlapping peaks and emphasis of small peaks; while additive and multiplicative
baseline shifts in a raw spectrum are removed [75]. However, each successive
derivative that is performed decreases the signal-to-noise ratio because two false valleys
are generated in the positive ordinate scale for every band in a negative direction [76].
Once baseline drift is corrected, those data points that fall well outside the trend
of the main population must be removed pending statistical verification. These outliers
may be the result of laboratory measurement errors, samples from different categories,
and/or instrumental error. It is critical to verify and remove outliers in both the training,
or calibration, set, and in the set of unknowns on which the calibration is to be tested for
validation and prediction. There are several methods for statistical determination of
outliers. Most often seen in the NIR literature is outlier removal from the calculated
Mahalanobis distance (H statistic). Typically, an outlier may be removed if the H
statistic is greater than three standardization units from the mean spectrum.
One of the most critical steps in NIR analysis is building a good calibration
model via pattern recognition methods (PRMs) to identify similarities and regularities
in the data. This involves calculating the regression equation based on the NIR spectra
and known analyte information. This model is then used to predict the future unknowns,
often expressed in terms of correlation [77] or distance [78]. There are two types of
PRMs, depending on whether or not the objects are known to belong to specific classes,
called supervised and unsupervised methods [79]. The latter involves clustering in an Ndimensional space without knowing the class to which the sample belongs, of which
principle component analysis and unsupervised neural networks are the most common.
The former involves training the system using a set of objects belonging to specific,
122
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
previously known classes. These methods can be broken down into the discriminant or
the modeling type [80]. Using discriminant methods, a pattern is split into the
corresponding number of classes that are included in the training set, which creates
bounds that are shared by each pattern space. The most commonly used methods for
discriminant analyses are the ^-nearest neighbor (KNN) function and potential function
methods (PFMs). By use of modeling methods, volumes are created in the pattern
space, which possess different bounds for each class. These bounds can be established
in the form of correlation coefficients, distances, residual variance, or supervised
artificial neural networks (ANNs).
The commonly used linear calibration methods include partial least squares
(PLS) and principle component regression (PCR), while locally weighted regression
(LWR) is used for nonlinear models. Partial least squares regression can be considered
an extension of the multiple linear regression model. A linear model specifies the
relationship between a dependent, or response, denoted Y, and a set of predictor
variables, denoted X, as shown in Equation 5.2.
Y = bo + biXi + b2X2 + ... + bnXn
Equation 5.2
In this equation bo is the regression coefficient for the intercept and the bi values are the
regression coefficients (for variables 1 through ri) computed from the spectral data.
The multiple linear regression model includes extensions to address more
sophisticated data analysis problems. The model serves as the basis for a number of
multivariate methods such as discriminant analysis (i.e., the prediction of group
membership from the levels of continuous predictor variables), principal components
regression (i.e., the prediction of responses from the dependent variables caused by
123
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
factors that underlie the levels of the predictor variables), and canonical correlation (i.e.,
the prediction of factors underlying responses of the dependent variables from factors
underlying the levels of the predictor variables). These multivariate methods all have
two important properties in common. These methods impose restrictions such that (1)
factors underlying the Y and X variables are extracted from the
Y 'Y
and X'X matrices,
respectively, and never from cross-product matrices involving both the Y and X
variables, and (2) the number of prediction functions can never exceed the minimum of
the number of Y variables and X variables.
Partial least squares regression extends multiple linear regression without
imposing the restrictions employed by discriminant analysis, principal components
regression, and canonical correlation. In partial least squares regression, prediction
functions are represented by factors extracted from the Y'XX'Y matrix. The number of
such prediction functions that can be extracted typically will exceed the maximum of
the number of Y and X variables.
Partial least squares regression is probably the least restrictive of the various
multivariate extensions of the multiple linear regression model. This flexibility allows it
to be used in situations where the use of traditional multivariate methods is severely
limited, such as when there are fewer observations than predictor variables.
Furthermore, partial least squares regression can be used as an exploratory analysis tool
to select suitable predictor variables and to identify outliers before classical linear
regression. PLS attempts to explain as much of the observed variation in the dependent
variables as possible using the minimum of relevant factors contained in the spectral
data [81],
124
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
In developing a calibration method, several parameters are evaluated including
factors, loadings, and scores. When choosing the proper number of factors, it is
important to avoid under- and over-fitting. The former results in an unreliable
prediction while the latter may result in more uncertainty in the calibration set thus
resulting in prediction errors. In order to optimize the number of factors, the PRESS
(Prediction Residual Error Sum of Squares) value can be calculated for every possible
factor. This is accomplished by building a calibration model with a certain number of
factors, then testing some samples of known concentration against the model. The sum
of the squared difference between the predicted and known concentrations gives the
PRESS value for that model. Scores are used to check sample homogeneity and possible
clusters. Loadings are used to interpret how variables are weighted in principle
component space.
The next step in data processing is validation, whereby the calibration model
developed is tested to ensure its validity. This is usually accomplished by splitting the
whole sample into two sets: one for calibration and the other for validation. If a limited
number of samples are available, a “leave-one-out” cross validation can be performed.
This done by leaving one sample out and the rest are used to build a calibration mode.
The calibration model then predicts the value of the sample that was left out. The
advantage of cross validation is that, unlike calibration with a full data set, the sample
being predicted is not included in the calibration model. Thus the model can be tested
independently.
There are several statistical methods to test the performance of the calibration
model, however, the common methods include standard error of calibration (SEC) or
125
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
standard error of estimate (SEE), multiple correlation coefficient (R), and standard error
of prediction (SEP) or standard error of cross validation (SECY). SEC refers to the
standard deviation of the errors found for the samples in the calibration set. This can be
compared to the accuracy of the reference method, and if this value is lower than that of
the error of the reference method, the model can be considered overfitted. The root
mean square error of prediction (RMSEP) can be calculated to measure the accuracy of
the calibration, as a measure of the difference between the predicted and true values.
The multiple correlation coefficient is dimensionless, where if equal to zero
denotes no correlation between reference values and those found using the calibration
model. However, a maximum R-value of unity denotes exact correlation, although
highly unlikely. Typical R values for “good” calibration models lie between 0.90 and
0.95, while exceptional values may lie between 0.98 and 0.99.
The SEP or SECV values refer to the standard deviation of the errors found in
samples not in the calibration set. If SEC values are approximately the same as SECV
values, then the calibration model is probably a valid one.
Finally, the calibration is used to predict the concentrations of unknowns,
assuming that the unknowns are in the same sample population as the samples used in
the calibration set. Also, the unknown must be tested to determine whether or not it is
an outlier.
5.3 Instrumentation
Traditional wavelength dispersive spectrometers use a rotating grating in a
Czemy-Turner or Ebert configuration, which must be precisely aligned. These
126
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
spectrometers must be maintained in a controlled environment, free of dust and
vibration, and are generally very costly. However, recent advances in high sensitivity
detectors, both CCD's and diode arrays, improvements in holographic gratings, and the
availability of low-loss optical materials have allowed for a new generation of
spectrometers at a fraction of the cost and size of a traditional spectrometer. These
compact spectrometers are rugged and portable, because there are no moving parts.
There are several CCD-based spectrometers that are commercially available,
designed as portable field instmments that provide real-time results. These
spectrometers collect the light through a fiber optic cable where it is collimated onto a
diffraction grating. For low-light situations, or to increase the signal size, the fiber optic
cable can be replaced with a slit, which eliminates light transfer variations caused by the
bending of the fiber. The spectrum is then directed across two mirrors and onto an
array, where the light is converted into electrical output. The folded path length also
results in a smaller instmment. Most arrays are thermoelectrically cooled to at least 10°C to decrease dark current.
Various semiconductor-based array detectors are commercially available,
including indium gallium arsenide (InGaAs) and lead sulfide (PbS) detectors. While
InGaAs detectors offer a smaller wavelength range (0.8 - 1.7 fim) compared to their
PbS counterparts (1 .0 - 3.2 fim) InGaAs arrays have, to some extent, replaced PbS
detectors due to their faster temporal response, which is required for rapid scanning.
InGaAs arrays have the added advantage of higher signal-to-noise ratios and higher
sensitivity than PbS detectors. A significant advantage of a PbS detector is its cost
effectiveness, which is typically one-fifth of the cost of its InGaAs counterpart. This
127
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
creates a very attractive alternative when several spectrometers are required for on- or
in-line process control, in such situations as industrial emissions monitoring, or air
quality-monitoring situations.
Instrument response can vary between spectrometers, as well as between
samples, and therefore the instruments require standardization. Bouveresse et al.
describe two methodologies to standardize instrumental response between NIR
instruments [82]. The first is based on a univariate slope/bias correction of the predicted
values. The spectra of samples collected on both a “master instrument” and a “slave
instrument”, given by mT and ST, respectively, are multiplied by the regression
coefficients b obtained on the master instrument, as shown in Equations 5.3 and 5.4
y =( 'T)=mT x b
Equation 5.3
y =(<r)= T x b
Equation 5.4
where y(mT) are the concentration values of the Nt subset samples determined by use of
the spectra and the calibration model built on the master instrument, and y(sT) are the
concentration values of the same subset as determined using the slave instrument
without taking into account different instrumental responses of the two instruments. The
predicted values of y(mT) are then plotted against the ones of y(sT), and a univariate
linear model that fits those Nt points is computed using the orthogonal least squares
approach. This simple univariate method cannot be damaged by artifacts or influenced
by problems that occur in spectral space, such as data heterogeneity. However, this
assumes simple differences between the master and slave instruments and may be
unacceptable when complex differences occur.
128
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
The second approach is the more sophisticated, piecewise, direct standardization
based on a multivariate correction of spectra. This method relates the spectral intensities
obtained at the ith wavelength on the master instrument to those contained in a moving
spectral window containing a few neighboring wavelengths on the slave instrument. For
each wavelength of the master instrument, a multivariate model is then built between
the spectral intensity obtained at the z'th wavelength on the master instrument and the
corresponding spectral window on the slave instrument, and the regression coefficients
are placed in a banded diagonal transfer matrix. In order to transfer the spectra from the
complete data set, the slave subset-matrix is multiplied by the transfer matrix.
Multiplying the transferred spectra by the regression coefficients built on the master
instrument allows prediction. Although this method is preferred for complex differences
in spectra, the authors recommend the use of at least five standardization samples in
order to obtain reliable results.
5.4 Literature NIR Applications
In this section, applications are reviewed that illustrate the exploitation of the
advantages of the use of simple NIR instrumentation. Such advantages include the
inherent rapid analysis time, and minimal sample preparation that is possible, as well as
the inexpensive nature of the instrumentation.
5.4.1 Skin Applications
Skin cancer is the most common form of cancer today. Before a biopsy is
performed, skin abnormalities are diagnosed by visual inspection, during which
129
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
malignant lesions may be undetected. There is a need for rapid, non-invasive screening
detection of carcinoma for which the use of NIR spectroscopy offers several
advantages. Tissue is transparent to the NIR wavelength range, and therefore, NIR
probing is completely non-invasive. It provides the opportunity to eliminate human
error during visual diagnosis and it may eliminate the need for surface biopsy
techniques. Another attractive feature is that fluorescence, which often hinders
biological sample analyses, is not usually present.
Fendel et al. have employed NIR-FT-Raman spectroscopy to study skin lesions
[83]. By using this technique, biochemical differences and the secondary structures of
tissues can be investigated for early cancer diagnosis. They have shown that by fitting
this instrument with a fiber optic probe, in vivo measurements are easily obtainable. In
addition to normal skin, the authors obtained Raman spectra of eczema, psoriatic skin,
and malignant Kaposi sarcomas. The Raman spectrum of psoriatic skin closely
resembled that of normal skin. However, the spectrum of eczema clearly illustrated an
increase in lipid and water content in the outermost layer, the stratum comeum (SC).
The spectrum of Kaposi sarcoma indicated a decrease in collagen, typical of cancer,
while an increase in histones and DNA/RNA. Using cluster analysis, comparison
between normal, benign lesions, and malignant sarcomas were clearly differentiated.
Four frequency ranges were selected: standardization of baseline at 2500 cm'1; the
strongest band (CH2 deformation) at 1450 cm '5; amide III at 1260 cm'1; and the protein
backbone vibration at 940 cm'1. A considerable separation by cluster analysis between
normal and diseased tissue was achieved. The Raman spectrum of melanoma was
obscured by the fluorescence of the pigment melanine. Finer cluster analysis
130
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
classification of melanomas with respect to stages, thickness, and other factors may
yield better Raman results by decreasing the standard deviation for a particular disease,
thus resulting in a larger statistical significance between spectra
McIntosh et al. investigated the potential of NIR spectroscopy to characterize
and differentiate a variety of skin lesions based on scattering properties [84]. For each
subject, three NIR spectra were recorded for the lesion and the contralateral area of
apparently normal tissue, which acted as the control. Reflectance spectra were collected
using a 10 n m slit width and consisted of 32 co-added scans with 40 s acquisition time
per scan. The results were histologically verified by a surface biopsy on the lesion. The
spectra obtained were grouped into one of six lesion categories: actinic keratoses, basal
cell carcinoma, dyplastic melanocytic nevi, actinic lentigines, banal common acquired
nevi, and seborrheic keratoses. The authors reported significant noise in the 1850 2400 nm region of the spectra due to strong absorption of water and therefore truncated
the data to the 400 - 1840 nm region, leaving a total of 720 data points. The 32 scans of
each sample were then averaged and those that lay outside two standard deviations were
omitted and attributed to patient movement during data acquisition. Distinct differences
in the spectra between the lesion and control were not observed by visual inspection,
therefore univariate statistics were applied. For each lesion category, paired t tests were
applied to find significant differences between lesion and control skin spectra. When
these values were plotted versus wavelength, significant differences were revealed in
several areas that varied slightly between each category. To assess statistical differences
between categories, Fisher’s least significant differences (LSD) and Duncan’s multiple
range tests were applied. However, there were no differences in any region that allowed
131
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
differentiation between all classifications of lesions. This technique allowed for noninvasive differentiation between normal skin and lesions. However, it was thought that
the long acquisition time, over 20 minutes per sample, may cause patient discomfort.
Woo et al. have used a portable NIR reflectance system with a photodiode array
detector to determine the moisture content in skin [85]. The system included a tungsten
halogen lamp light source, an InGaAs photodiode-array microspectrometer, an internal
portable battery and it was connected to a laptop computer through an RS-232 cable.
The fiber optic probe consisted of nine fiber-optic bundles - eight surrounding bundles
for illumination and a center bundle for collecting light from the sample. A holder was
used to maintain a 0.3 mm gap to avoid probe contact with the skin and allow for a
reproducible spectrum to be obtained. The results from the portable instmment were
compared to that of a conventional scanning NIR spectrometer. The relationship
between water content and NIR absorbance was determined using a partial least squares
regression (PLS) for both hairless mouse skin and human skin. The hairless mouse skin
samples were soaked for 1 hour to absorb a maximum amount of water. The samples
were then weighed at 1-h intervals while drying in a desiccator, and measured using
both the portable NIR system and the benchtop scanning NIR instrument. Moisture
content in human skin samples was also measured by the capacitance method. Using the
portable NIR instrument, differences in the NIR spectra of the hairless mouse skin
samples were clear at 1450 nm (OH band stretch of water). However, a shift in the
baseline of the NIR spectra made the univariate calibration method difficult, therefore
partial least squares (PLS) regression was employed to elucidate the relationship
between NIR absorbance and water concentration. The authors concluded that the
132
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
performance of the portable NIR instrument was comparable to a scanning-type
spectrometer in the 1150 - 1650-nm range, allowing a more stable spectmm and a rapid
analysis time within two seconds.
Martin used NIR to measure water content in various skin layers in vivo by
employing NIR reflectance spectroscopy [86]. Although the water in the SC is
associated with lipid bilayers as well as the protein moiety, ambient humidity, also
plays a role. By varying the ambient humidity the author was able to assign peaks for
three types of water associated with the SC. The tentative assignments were for the lipid
bilayers, secondary and primary water on protein groups at 1875, 1909, and 1923 nm,
respectively. A fourth type, which appeared at 1890 nm, was assigned to the water
below the SC layer. An interesting finding was the misconception of the use of
moisturizers. When applied to the skin, scattering is reduced, which suggests that the
moisturizer is simply smoothing the skin rather than hydrating it.
5.4.2 Plastics Identification
Plastics identification presents a challenge in terms of efficiency, accuracy, and
cost effectiveness. Since most recyclable plastics are opaque, remote measurements
must be done in diffuse reflectance mode, where large light loses occur [87].
Rohe et al. have used transmission NIRS of a polymer extrusion process for the
determination of polyethylene/polypropylene (PE/PP) melt composition as a means of
real-time quality control [88]. Due to the harsh conditions of the extrusion process, such
as the high temperature and pressure requirements, as well as the high viscosity and
corrosiveness of the polymer melt, the author’s developed a robust fiber optic probe to
monitor the process. The probe design consisted of a spherical lens welded to a fiber,
133
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
both of which were made of quartz. The central optics were encased in a stainless steel
sleeve of which the empty space was filled with a high temperature resistant epoxy.
This entire setup was then enclosed by another protective sleeve with an opening for the
spherical lens, which had contact with the polymer melt. The opposite side of the
housing had a standard connector for an optical fiber. The advantages of this design
included the high temperature resistance of the quartz, which could withstand
temperatures up to 800°C, whereas the temperature of the extrusion process lay in the
200 - 350°C range. The use of the high temperature epoxy eliminated internal tensions
caused by differences in thermal coefficient of expansion of quartz and housing
material. By employing PLS analysis, the performance of the NIR system for the
determination of the PE/PP composition had deviations between predicted and actual
polymer compositions in the range of ±2.5%.
De Groot developed an automated sorting system, utilizing NIRS for the online
separation of wood, plastic, and stone [89]. In order to speed up the process, “mini­
spectra” were recorded rather than complete NIR spectra. Only six wavelength regions
were measured, which had been previously determined [90] to be suitable for
demolition waste separation. Diffuse near-infrared reflected radiation from reference
objects were collected perpendicularly from a conveyor belt to obtain classification
results. Performance and robustness were then tested under both laboratory and
industrial conditions. The authors demonstrated that the selection of wavelength
regions, the application of LDA and SNV preprocessing, and the use of the
Mahalanobis distance, provided acceptable classification results on the pilot plant.
However, classification performance dropped in the presence of wet objects, or during
134
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
measurements at the periphery of the conveyor belt. Several recommendations were
made including the use of a larger diode array and a greater wavelength range.
Van den Broek et al. have developed a remote sensing NIR system for real-time
plastic discrimination for use at waste facilities [91]. Although this had been previously
investigated [92], the authors coupled the spectroscopic imaging with a new
classification technique. The earlier work was based on classical classification
techniques, namely linear discriminant analysis (LDA) and partial least squares (PLS).
Although these techniques afford simplicity in principle and implementation, they
neglect the non-linear behavior of the spectral data in the NIR region. To circumvent
this issue, the author’s proposed the use of an artificial neural network (ANN).
The validation process of the ANNs consisted of three databases: classification,
precision, and robustness. The author’s showed that their method exceeded the
mandated 80% correct classification. However, this could be enhanced by improvement
of the experimental arrangement to reduce shadow contributions, together with further
stabilization of the sensor and/or the light source to improve reproducibility.
Feldhoff et al. have also developed an on-site post-consumer waste
identification system similar to that described by Van den Broek et a l [93]. The spectra
of various polymers (polyethylene, polyethyleneterephthalate, polypropylene,
polystyrene, and polyvinylchloride), as well as a cardboard/plastic compound, were
recorded between 900 and 1700 nm using an InGaAs array detector. The experimental
setup is shown in Figure 5.1. The conveyor belt moved at a speed of 1 m/s and with an
integration time of 6.3 ms, which allowed the recording of 158 spectra per second.
Using both principle component analyses plots and further classified with a
135
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
FuzzyARTMAP neural network, correct classification was achieved with 97%
accuracy. The authors attributed about 5% of error to those containers consisting of
polyethylene bottles with large polypropylene caps.
136
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
InGaAs spectrometer
computer
halogen lamp
collection optics
conveyor belt
1 m/s
reflector
Figure 5.1: An on-site waste identification system developed by Feldhoff et al. The
conveyor belt moved at a speed of 1 m/s and with an integration time of 6.3 ms which
allowed for recording o f 158 spectra per second.
137
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
5 .4 3
Hydrocarbon Samples
Choquette et al. have used Fourier transform NIR (FT-NIR) and Fourier
transform Raman (FT-Raman) for Identification and quantification of oxygenates in
SRM gasoline sample ampoules [94]. The current methods of analysis include the use
of gas chromatography coupled with various detectors. Although these methods may be
successfully employed, common to all is the necessity to open the SRM sample
ampoules, which is not required in the method described here. The ampoules were
attached to a spinning apparatus to homogenize the ampoule light scatter and
transmission characteristics. Since the variation of oxygenate concentration was
difficult to interpret visually, partial least squares (PLS) was be applied. By use of both
FT-NIR and FT-Raman, the authors were able to identify the oxygenate additives that
were present, as well as the oxygen concentrations, within 0.1% oxygen mass fraction.
This nondestructive technique had the additional advantage of acquisition times that
were less than 1 min.
Buerck et al. have developed a sensor for the determination of hydrocarbons in
groundwater [95]. This was accomplished by solid-phase extraction of analyte
molecules into the polymer cladding of a fiber. The fiber was attached to a conventional
spectrometer by which direct absorption measurements could be made. Typical
hydrocarbon concentrations ranged from 80 fig ml'1 to the limit of detection, which was
around 200 ng ml'1. One of the main advantages of this instrumentation was that it was
free of matrix interferences, such as background absorption of water, and stray light due
to turbidity in the sample. That authors indicated that the instrumentation is generally
useful for the monitoring of pollutants such as aromatic solvents, fuels, mineral oils or
138
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
chlorinated hydrocarbons with relatively low water saturation solubility, and if coupled
with a field-portable instrument would allow for on-site analyses. The results obtained
with the instrument were compared with those obtained with an HPLC-UV/VIS
instrument. However, for various reasons, the samples were stored for over a week
before HPLC analysis could be performed. This resulted in degradation of the aromatic
hydrocarbons, as demonstrated in a correlation plot of the two detection systems.
Oil leakage and other accidents pose serious environmental threats, therefore
rapid determination is essential in order to proceed with necessary action. Chung et al.
have demonstrated the usefulness of a robust, portable PbS NIR instrument using a fiber
optic interactance/reflectance probe for this application [96]. This probe consists of
concentric rings of 210 illuminating fibers that make up the inner core, and 210
receiving fibers that make up the outer ring. NIR spectra were collected in the 1100 2500-nm range, although useful specral information is located in the 1100 - 1650 n m
and 1800 - 2100 nm spectral ranges. Through the use of PCA and Mahalanobis
distance, six different typical petroleum products consisting of light straight-run (LSR),
naphtha, kerosene, gas oil, gasoline, and diesel were analyzed in 372 samples collected
over a 4-month period. Using this method, each product was identified with an
accuracy over 95%, while LSR, kerosine, gasoline, and diesel samples were predicted
with an accuracy of 99%. With total analysis time of less than one minute, NIR
analysis affords rapid and accurate analysis of petroleum products.
139
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
5.4.4 Agricultural Applications
5.4.4.1 F ru it samples
Water content in fruit can be used as a means of quality control [97]. In a quality
control application, several spectrometers are required, but wavelength calibration and
spectral response may vary within and between spectrometer models. Greensill et al.
[98] utilized various chemometric techniques to correct for differences between four
different NIR spectrometers for the determination of sucrose in a water-cellulose
matrix. The authors pointed out that the use of predictive models to transfer from a
master to a slave instrument must assume a stable, linear relationship between
instruments, which may not be the case. Rather, a “spectral response correction”
approach is more advantageous, due to the elimination of estimation errors. The authors
demonstrated that direct standardization of the wavelet coefficients of the first level
decomposition was most efficient. The wavelet transformation approach enables
analysis of signals at different levels of resolution in the time domain. The transform of
the data exhibits discrete steps in time on one axis, and discrete steps of resolution on
another resulting in a two-dimensional, time-scale domain analysis of the signal.
Wavelet coefficients are adjusted on the slave instrument to reconstruct the spectra in
the original domain [99].
A universal method for visualizing the sugar content in the flesh of melons has
been developed by Tsuta et al. [100] who used a CCD detector with band-pass filters.
Each filter created a spatial image of the melon sample for a specific spectral region.
This method had been previously employed for measurement of sugar distribution in
green-flesh melon and kiwifruit [101,102]. It was demonstrated that the chlorophyll
140
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
absorbance near 676 nm shows a strong inverse correlation with the sugar content.
However, it cannot be applied to a red-flesh melon because it is not solely dependent on
the sugar absorption band, but on some other color information at 676 nm. The authors
extracted a 25-mm-diameter cylindrical sample from the “equator” of a melon, and a
spectrum was obtained using a fiber optic probe. The wavelengths of 902 and 874 nm
were used to correlate sugar content, while the wavelengths 846 and 930 nm were used
to calculate the second-derivative absorbances. These two latter wavelengths were
chosen because they gave the highest correlation with sugar concentration using the
least number of band-pass filters. The raw spectra contained thermal noise; bias signals,
which is an electronic offset added to the signal to insure positive A/D counts;
variations in pixel-to-pixel sensitivity; and lighting variations on the sample surface.
The authors then corrected for these various factors by employing Equation 5.5 [100].
raw image - dark frame
processed image = ---------------xM
flat field - dark frame of flat field
^
.
Equation 5.5
In order to compensate for the above mentioned sources of thermal noise and bias
signals, a dark frame image was recorded under the same conditions as the raw image,
with the exception of lighting, and subtracted from the raw image. A flat field image
was taken by recording a uniformly lit flat field and its dark frame counterpart was also
recorded. The ratios of these sets of images, which were used to compensate for the
effects of variations in pixel-to-pixel sensitivity and lighting variations, were then
multiplied by M to restore the ratio of the images to the original image intensity level,
where M is the intensity value averaged over all pixels of the flat field after dark frame
141
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
subtraction. The average intensities of the images were converted into an average
absorbance and each spectrum was then converted into a second-derivative spectrum
according to the truncated Taylor expansion, shown in Equations 5.6 and 5.7.
d2A902 - A9 3 0 —2 A902 - Ag74
Equation 5.6
d“A.87.4-A 9 0 2 - 2Ag74 - A846
Equation 5.7
Then, MLR analysis was carried out to acquire the calibration curve for the sugar
content on the imaging system. The sugar distribution map of the half-cut red-flesh
melon was constructed using a linear color scale. Color images published by the authors
showed that the sugar content increases from the rind to near the seeds. These results
indicated that this method was useful for determining sugar content in melons and may
be applicable to other constituents in other agricultural products.
Steuer et al. have used NIR spectroscopy for the classification of various types
of essence oils, and for the determination of individual components extracted from a
variety of citrus fruits, which has application for in-line quality control measurements
[103], The samples were scanned in the 1100 - 250 nm range with a resolution of 2 nm.
Principle component analysis (PCA) was performed on the spectral data only, and a
modified PLS was used for the calibration development. The SECV associated with the
technique, which was below approximately 1%, was within the range of that obtained
using gas chromatography, with a flame ionization detector, for the main components
(e.g., limonene, y-terpinene, sabinene) and general chemical-physical properties (optical
rotation value, aldehyde content).
In the past, the major components in orange juice have been determined using
conventional wet chemistry methods, which is generally very laborious and time
142
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
consuming. Li et al. have used a combination of NIRS and multivariate calibration for
determination of glucose, fructose, sucrose, citric and malic acids in dry extracts of
orange juices [104]. NIRS of aqueous solutions, such as fruit juices, results in overlap
of vibrational bands of water with solutes, and usually cannot be deconvoluted [105];
therefore the water must be removed. Li et al. utilized a technique developed by
Meurens et al. [106], whereby the aqueous sample was suspended on a fiberglass
support and was dried to produce a dry extract. Using this method, 218 orange juice
samples were recorded in transmission mode, of which 150 randomly selected samples
were used to establish the calibration, and the remaining 68 were used to validate the
model. The authors developed comparison data using both stepwise multiple linear
regression (SMLR) and partial least squares regression algorithms. While SMLR
resulted in a better prediction, because of elimination of X-variables that were not
relevant, these X-variables actually carry unique information about Y. Therefore, the
PLS method is expected to give more reliable predictions, although results obtained
from both methods were very close. Unfortunately, minor components, such as malic
acid, could not be determined with acceptable precision. Therefore, future work will
involve the investigation of various methods of spectral standardization to improve the
sensitivity and precision of the NIRS analysis.
Dambergs et al. have quantified methanol, which is an undesirable contaminant
in grape-derived ethanol that is used in the production of fortified wines [107]. This
wine-fortifying spirit, known as SVR (from the Latin spiritus vini recticatissimus),
generally contains 97% v/v ethanol with low concentrations of methanol and other
volatile compounds. Typically, gas chromatography is used to quantify methanol in
143
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
wine. However, the feasibility of NIRS for this analysis was explored. In addition to a
wide range of SVR samples from four producers and three vintages, two other major
distillation process fractions were analyzed: “feints” and “heads.” Feints are primary
fractions from a first pass through distillation columns and are the starting material for
the final SVR fraction. Heads are low-boiling point waste fractions collected from the
top of the distillation columns and represent the upper limit of methanol concentration.
Various combinations of wavelength ranges and mathematical treatments were tested of
which the wavelength range 1200 - 2450 nm and PLS regression on first derivative
spectra were optimal. The number of factors used in the SVR methanol calibration was
relatively high. However, this was to compensate for hydrogen bonding in a mixed
alcohol system that causes absorbance nonlinearities. To avoid over fitting, the number
of factors was optimized using the PRESS value. The authors contended that increasing
errors in NIR calibrations may be due to seasonal variations, which may not pose a
problem for those producers in the same geographical location. Other areas of which
matrix effects posed a problem are in the heads. Here, matrix variation is evident and,
although similar to the fortifying spirit, they are relatively high in methanol,
acetaldehyde, and ethyl acetate levels. The authors suggested the use of ANN
calibration models to improve the accuracy in situations of nonlinearity. A multiple
linear regression calibration model was also developed using fixed wavelengths derived
from continuous scans. The authors believed that this presents a viable, cost effective
option for routine analyses. The wavelengths were chosen at points of maximum
correlation with the methanol reference value, and provided a calibration with an R2
value of 0.98, compared with 0.998 using continuous scans. Although not part of the
144
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
initial investigation, the authors were able to develop a robust model for the
determination of total alcohol content in the fortifying spirit, based on the assumption
that the major components are ethanol and water.
5 A 4.2 M eat samples
Ding et al. have employed NIRS as a means of adulteration detection in
hamburger meat [108]. Reflectance spectra of raw, cooked, and minced hamburgers
were recorded between 400 and 2500 nm and plotted verses log (1/R), where R
represents reflected energy. Spectral data of the samples were compressed into principal
components (PC) by principle component analysis before discriminant analysis was
performed using canonical discriminant analysis (CDA) and K-nearest -neighbor
(KNN) methods. A modified partial least squares method was used as a regression
method to predict the adulteration level, therefore calibration equations were developed
for each adulterant. The established calibration equations for prediction of adulteration
levels with mutton, pork, skim milk powder, and wheat flour had standard errors of
cross-validation of 3.33, 2.99, 0.92, and 0.57%, and coefficients of variance of 0.87,
0.89, 0.99, and 1.00 respectively. Adulterants of 5 - 25% were detected with accuracy
up to 92.7%. As the adulteration level increased, the accuracy increased.
Gonzalez-Martm quantified protein and infiltrated fat in the pork loin muscle of
Iberian swine [109]. The amount and composition of the fat of swine are determinant
factors in the quality of the meat and derived meat products. By coupling NIR analysis
with a fiber optic probe, instantaneous analysis at the production level was possible.
Reference chemical measurements were performed with ground and homogenized
samples from which the fat was extracted with petroleum ether and evaporated, while
145
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
protein determination was accomplished by the Kjeldahl method. NIR spectra of 56
samples of pork loin samples were recorded by applying the fiber optic probe to intact
pork loin samples and also after the samples had been cut and homogenized. The
Mahalanobis distance was calculated to remove outliers ±3.0 standardized units from
the mean spectrum. Calibration was then performed by applying an MPLS, and its
accuracy was optimized using several methods, to give the highest R value and lowest
standard error of calibration (SEC), of which multiplicative scatter correction on first
derivative spectra was best. The content of infiltrated fat and protein was found to vary
from 3 to 19% and 21 to 31%, respectively. Utilizing the fiber optic probe, analysis of
the ground meat samples resulted in SEP values of 0.53% and 0.44% for fat and protein,
respectively, while intact analysis resulted in SEP values of 0.74% and 0.81%
. respectively.
5.4.43 Milk samples
Determination of principal milk constituents is of great importance in the dairy
industry. As noted by Laporte et al. [110], the dairy industry requires true protein and
casein determination, rather than just protein. There is a need for an inexpensive and
rapid method of analysis. The authors performed transmission NIRS analysis from 1100
to 2500 nm, on both unhomogenized and pasteurized milk samples, to evaluate the
feasibility of this technique. Two types of calibrations were prepared: an overall
calibration containing 96 samples of homogenized and unhomogenized, and also
unpasteurized and unhomogenized, raw milks; and a partial calibration containing 76
unhomogenized samples. The Mahalanobis distance (H statistic) was calculated in order
to determine outliers, based on a sample having an H statistic > 3.0 standardization units
146
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
from the mean spectrum. Using this procedure, eight outliers were eliminated from the
partial calibration and five were deleted from the overall set. Calibrations were
performed by modified PLS, and they were optimized using the scattering correction
standard normal variate (SNV). Standard errors of calibration were 0.12%, 0.06%,
0.04%, and 0.05% for fat, crude protein, true protein and casein, respectively.
Validation of the overall calibration with an independent set of samples gave standard
errors of prediction of 0.07% for fat, 0.06% for crude protein and casein, and 0.05% for
true protein. All statistical parameters were better using overall calibration, rather than
partial calibration, with the exception of fat, which indicates that homogenization, has
an effect on NIR fat determination. Although a relatively small number of samples were
included in the calibration model, this approach to analysis of fat and nitrogenous
constituents in milk was found to be reliable.
Sasic et al. have also studied raw milk samples, but by exploration of the
potential of using the near NIR region, from 800 - 1100 nm [111], This region is
particularly important for nondestructive or noninvasive analyses of biological samples
and is useful because of the availability of sufficiently sensitive detectors for this region
[112]. To compensate for the shifting baseline, a multiplicative scatter correction
method was applied. The authors note that baseline instability is caused mainly by light
scattering due to fat globules, which may be useful for the determination of fat content;
however, it diminishes the importance of chemically dependent variability especially
when other milk components are analyzed. In the determination of fat content, the
application of various PLS loading weights were used to illustrate spectral variances in
NIR spectra associated with known NIR frequencies of fat, with a band at 928 nm being
147
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
most important. The prediction of protein content was good, with better correlation
coefficients and RMSEP when the wavelength range was narrowed to 900 - 1100 nm.
This was true even though bands representative of protein had a much lower intensity
than those of fat. Lactose, however, was much more difficult to quantify due to noisy
regression coefficients.
5A.4.4 Miscellaneous agricultural samples
Qui et al. have evaluated the feasibility of using NIRS for the determination of
fructose, glucose, sucrose, maltose, free acid, lactone, and hydroxymethylfurfural
(HMF) in commercial bee honey samples [113]. Both transmission and reflective NIR
spectra were recorded, while the optical path length of the quartz cuvette was varied to
optimize the calibration. From a total of 74 brands produced in 11 countries, 50 were
selected to form a calibration set, while the remaining 24 samples formed the validation
set. For the two scanning modes and different path lengths, the transmittance mode
showed sharper peaks and better resolution, while a 1 mm-path length cell had the
greatest energy transmission. These were both verified by having the lowest SECV. A
modified PLS regression model outperformed MLR and PCR, and was therefore used
for all honey constituents except moisture content, of which optimal calibration was
developed with a PLS regression model. The chemical composition of the samples were
determined by using a refractometer for moisture, a titrimetric method for free acid and
lactone, a spectrophotometric method for HMF, and HPLC for fructose, glucose,
sucrose, and maltose. The calibration was validated by accurately determining the
contents of moisture, fructose, glucose, sucrose, and maltose with squared correlation
148
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
coefficients of 1.0, 0.97, 0.91, 0.86, and 0.93 between the predicted values and the
reference values, respectively.
Buning-Pfaue et al. have shown that utilizing a fiber optic-fitted FT-NIRS
instrument for the determination of the degradation of deep-frying fats can dramatically
reduce the time and labor involved in typical method of analyses [114]. Although there
have been other NIRS investigations on frying fats, there is a lack of statistical data that
illustrate possible applications [115,116 ,117,118, 119,120,121]. Here, a total of 50
samples were taken from canteens, restaurants, and snack bars and analyzed for acid
values (AVs), total polar material (TMT), dimeric and polymeric triglycerides (DPTG)
and food oil sensor (FOS) values, which were used to estimate the degradation of frying
fats. By applying a first derivative to the spectra, baseline shifts, peak overlaps, and
other spectral defects were removed, which increased the spectral information. The
standard error of calibration (SEC) and standard error of prediction (SEP) statistical
criteria were used to determine if the developed method was acceptable for prediction
of chemical parameters of the unknown samples. The bias, which is the average
difference between measured and predicted values, was calculated to determine validity
of the method, while the standard error of cross validation (SECV) was used to check
the multivariate model. The authors illustrated that the statistical criteria, after crossvalidation for TMT and DPTG, were poor which indicated that this method is adequate
for screening only.
149
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
5.4.5 Pharmaceutical Applications
NIR diffuse reflectance spectroscopy has proved to be a useful tool in the
quality control aspect of pharmaceutical manufacturing where only minimal sample
manipulation is required. Blending is a combination of three basic processes: (1)
diffusion, the random motion of individual particles relative to one another; (2)
convection, the motion of a group of particles from one position to another; and (3)
shearing, a change in configuration of ingredient particle locations through the
formation of slip places in the mixture [122]. Ideally, the result is a homogenous
distribution throughout the blender, which has usually been confirmed by sampling at
predetermined time intervals while a sampling probe is inserted at various depths and
analyzed using high performance liquid chromatography (HPLC) or UV/Visible
spectroscopy [123], However, depending on the flow properties of the powder, this
sampling approach may introduce inhomogeneity in the powder bed or preferential
adherence of powder onto the probe as it passes through the bed. The use of NIRS as an
on-line technique for powder blending in pharmaceutical preparations has been
explored, as homogeneity ultimately affects proper dosages of solids.
Sekulic et al. have coupled a NIR spectrometer fitted with a fiber optic probe
that was mounted to a blender, at the axis of rotation, so that the components being
blended were in contact with the probe to determine if the point of sample homogeneity
was reproducible [124], The spectrometer was configured to acquire and average 32
scans per sample, in transmission mode, over the 1100 - 2500 nm wavelength range.
Measurements were made in 2 nm increments, and at 1-minute intervals for a total
blending time of 25 minutes. The blend consisted of 10% sodium benzoate, which
150
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
provided aromatic content typical of pharmaceutical compounds, 39% microcrystalline
cellulose, 50% lactose which was all blended for 20 minutes, after which 1%
magnesium stearate was added and blended for another 5 minutes. As expected, the
spectra changed significantly at the beginning of the blending, but, as blending
continued, spectral variability decreased to an indistinguishable level. Homogeneity was
difficult to determine even upon plotting first and second derivatives, therefore a
moving block standard deviation on the collected spectra was calculated. From this, it
was expected that as blend homogeneity increased, the standard deviation would
approach zero. Since the objective was to perform these measurements in real time, the
standard deviation of the moving block standard deviation was calculated to determine
the time at which homogeneity occurred. This demonstrated an optimal blending time
of less than 10 minutes. However, the standard deviation increased with the addition of
the magnesium stearate. The frequency of the spectra collection needed to be increased.
This non-invasive method offers several advantages, including real-time analysis, and
elimination of time-consuming traditional analysis techniques; multicomponent
homogeneity analysis, rather than only active ingredient homogeneity analysis;
decreased worker exposure to the blend, and optimized instrument usage.
Blanco et al. have also used this technique, using a fiber optic to monitor the
active ingredient, gemfibrozil, in Trialmin 600®, in three different physical forms of
tablets production: the blended, or granulate, product is a powder, the cores are
compacted samples, and coated tablets are the cores containing a lacquer [125]. The
ultimate goal of this work was to develop a single calibration model for analyses during
the different production steps, without the need to run an individual calibration for each
151
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
step, and to simplify quality control by using a single measuring system for all sample
types. Synthetic samples were prepared to encompass a wide concentration range that
spanned the tolerated limits of the nominal value of the active ingredients. The
pharmaceutical product contains the active ingredient gemfibrozil at a nominal level of
755 mg g '1 in the blended product and the cores, which decreases to 729 mg g'1 in the
coated tablets due to the application of the lacquer coating. Other ingredients include
pregelatinized starch and microcrystalline cellulose in concentrations of 130 mg g'1 and
75 mg g~\ respectively. Minor constituents include colloidal silica, carboxymethyl
starch, magnesium stearate and polysporbate 80. A NIR spectrum was recorded for a
total of 26 laboratory samples prepared with varying amounts of all constituents with
the exception of the polysorbate, which was at a concentration below 0.8%. Other
samples were produced from various lots of raw materials, particle size distribution,
compactness, and granulation. Those samples in the calibration set were chosen by
PCA. The relative standard deviation of the prediction error was less than 1% using
PLS on first-derivative spectra.
Scarfi et al. have investigated the use of NIRS for the identification of
counterfeit drugs, which is becoming increasingly important [126]. Black market
counterparts of prescription drugs contain altered compositions of both active and
inactive ingredients, which can pose a severe risk to the health of a patient. The use of
NIRS offers several advantages over conventional techniques, such as HPLC, which
can be a long, costly, and destructive procedure. NIRS offers on-site, direct, non­
destructive, and rapid analysis. Here, identification is based on the comparison of the
NIR spectrum of a black market drug with typical spectra of the authentic drug, by use
152
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
of multivariate modeling and classification algorithms (PCA/SIMCA). The authors have
shown that a first derivative can minimize spectral changes associated with several
parameters including tablet geometry, physical differences in tablet faces, and position
relative to the probe beam. Using a multivariate model, they were able to correctly
classify all counterfeit drugs, thus unequivocally demonstrating the potential for NIRS.
Rager et al. have demonstrated the benefits of applying NIR spectroscopy for
the standardization of Hypericum perforatum, which is the active ingredient in St.
John’s wort [127]. This work is significant in that that the use of NIR for the
determination of plant extracts has been demonstrated to have a relatively poor
detection limit (about 1%). Two components of Hypericum perforatum include
hyperforin and 13,118-biapgenin, which are present at 1 5% and 0.1 - 1.7%,
respectively. Reference measurements were performed by reversed-phased HPLC on 35
extracts of different origins. Spectral variability from scattering was eliminated by
applying the second derivative of the plots, which also decreased the signal-to-noise
ratio by a factor of four. Using PLS regression, a multivariate calibration was done
separately for the two ingredients. Satisfactory calibration statistics were obtained for
hyperforin with a root mean square error of calibration of 0.17 and a root mean square
error of prediction of 0.22 at a concentration range from 1 to 6% in the dry extracts.
However, due to the significantly lower concentration of 13,118-biapgenin, the
concentration range was limited from 0.55% to 0.35% and a multivariate analysis was
recalculated for the lower concentration range, resulting in a root mean square error of
prediction of 0.007%.
153
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
Laasonen et al. used NIR reflectance spectroscopy for the identification of
Echinacea purpurea, dried milled roots [128], In addition to mislabeling errors, which
often occurs between different species of Echinacea, the roots of Parthenium
integrifolium L. commonly adulterate commercial roots of Echinacea purpurea [129];
therefore a rapid identification alternative to HPLC and TLC is required. In this study,
dried roots of E. purpurea from the Astercea family were used as the true samples.
False samples were E. pallida, E. angustifolia (two other Echinacea species), and P.
integrifolium, another plant from the same Asteracea family. Samples not received in
powder form, either entire roots or cut, dried roots, were milled to a powder with a
broad distribution size of 90 - 1000 pm. To remove humidity as a variable, samples
were dried using an infrared dryer at 110°C until the loss in weight was less than 0.1%
during 50 s. Nine adulterated samples were prepared containing E. purpurea and 5%,
10%, or 20% of the milled adulterant. PCA was performed on second-derivative spectra
to investigate qualitative differences between the samples in principle component space.
A total of 10% of the E. purpurea and 0% of the false samples from that validation set
were misidentified. The author’s attributed this misidentification due to differences in
particle size distribution of one E. purpurea batch compared to that of another. They
showed that adulterated E. purpurea samples can be detected at a minimum of 10%
adulteration, exemplifying it’s potential for routine use.
Although diffuse reflectance NIR is useful for the determination of sample
homogeneity by examining a small cross-sectional area, it becomes ineffective when
acquiring other data of other pharmaceutical importance, such as tablet degradation,
tablet hardness, and quantification of active ingredients in intact tablets. Here,
154
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
transmission NIR measurements are used which provide analysis of greater area, in
addition to adding depth, which can give volume measurements.
Gottfries et al. have compared both transmission and diffuse reflectance NIRS
measurements in the determination of metoprolol succinate in intact tablets [130]. In
addition to varying concentrations, tablets also varied in size and lots of metoprolol
pellets, and microcrystalline cellulose was used to study packing density and batch-tobatch differences, respectively. This resulted in 40 different batches. Although a
narrower wavelength range was available in the transmission mode, due to insufficient
penetration of light, prediction for models based on transmission spectra were much
more accurate compared to its diffuse reflectance spectra. This is attributed to the
sensitivity o f the diffuse reflectance measurement to changes in homogeneity.
Sampling geometry plays a critical role in acquisition of transmission of NIR
spectra, particularly in terms of light leakage, which can pose a severe threat resulting in
diminished spectral quality. Sparen et al. have investigated the effects of light leakage
obtained with different sample holder geometries on content calibrations for
transmission NIR on tablets [131], The authors built PLS models of both a composite
and compact tablet using sample wells of varying heights and diameters according to a
multivariate design, using LC as a reference method. They have verified that a small a
disparity in tablet-to-well diameter diminished the spectral reproducibility, although
surprisingly gave the best predictions and more robust models. The authors have
concluded that rather than minimizing light leakage at calibration, it should be included
as a factor in the multivariate model to predict variation in light leakage.
155
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
5.4.6 Biological Applications
When applying pattern recognition techniques to spectral data, problems may
arise that result in varying success rates of classification by typical pattern recognition
methods despite a linearly separable training set [132], For this reason, genetic
algorithms (GAs) are used for pattern recognition analysis of NIR spectral data. GAs
are flexible numerical optimization techniques based on the concepts of genetics and
natural selection [133,134,135,136,137,138]. A set of features, in this case wavelengths,
are identified to optimize the separation of the classes in a plot of the two or three
largest principle components of the data. The bulk of the information encoded by the
selection features is based on differences between classes in the data set, and the GA
focuses on those classes that are difficult to classify during training. The GA evolves
over time as it minimizes convergence to achieve the correct classification in a fashion
similar to neural network approaches [132].
Small et al. utilized this technique for the analysis of glucose in biological
matrices [139]. The authors demonstrated that, prior to building the model, the use of
digital filtering as a spectral preprocessing technique could improve the results of PLS
regression by extracting only those frequencies associated with the spectral bands of
interest, while removing spectral artifacts such as noise and baseline variation.
Optimization of five important variables were important for the successful application
of this strategy: the position and width of the bandpass filter, the starting and ending
points of the spectral range submitted to the PLS regression, and the number of terms
employed in the calibration model. However, this method failed to account for
interaction effects among the five variables involved. Since both bandpass digital
156
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
filtering and PLS regression both attempt to extract analyte information from the
spectra, the authors indicated that the five variables must be carefully optimized in order
to obtain synergistic effects between the digital filtering and the PLS regression.
In order to study the above interaction effects, Shaffer used FT-NIR and GAs for
the measurement of glucose in three different matrices, namely triacetin and bovine
serum albumin (GTB), human serum, and bovine blood [140]. The authors developed a
GA configuration that could optimize dissimilar variables and designed an optimal
fitness function to monitor the progress of the optimization. Three different fitness
functions were compared to determine optimum performance.
Torella et al. have monitored blood loss in man by studying the relationship
between tissue oxygenation and NIR spectra [141]. Hemoglobin saturation in the
cerebral (CSO2 , left frontal cerebral cortex) and peripheral (pSC>2 , left calf) were
monitored in blood donors both ten minutes prior and ten minutes after the collection of
470 mL of blood. The oxygenation index, HbD was then derived by subtraction of the
concentration of deoxygenated hemoglobin, HHb, from the concentration of the
oxygenated hemoglobin of the right calf, C^Hb. In order to compensate for changes in
pathlength, a two wavelength light source and two detectors were used 3 and 4 cm from
and inline with the light source. The “near” detector collected scattered light from more
superficial tissues, while the “far” detector also collected scattered light from deeper
tissue. By subtracting the “near” from the “far” sensor data, the effect of the pathlength
was essentially cancelled to produce a continuous output that consisted of only regional
hemoglobin saturation from deeper tissues. Maximum values for the decrease in
oxygenation index occurred 10 minutes after blood collection for CSO2 , with a mean fall
157
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
of 2.5, and at the end of blood collection for pSC>2 with a mean fall of 3. There was also
a maximum after 8% of blood volume loss for HbD with a mean fall of 7.2. This is a
potentially viable technique for blood loss in humans due to the good correlation
between NIRS parameters and blood loss.
Kusaka et al. performed similar analyses for the quantitation of cerebral
oxygenation, but reported more details of their statistical analyses [142]. NIRS has also
been used to monitor hemoglobin in blood as a means of quantifying tissue
oxygenation. This required an understanding of the path length of light, which can
change significantly as scattering and absorption can contribute to optical losses.
[143,144,145]. Time resolved spectroscopy is used to correct for this flattened spectrum
in which the transit time of each photon through the tissue of interest is measured. By
correlating the relationship between chromophore concentration and actual absorption
in scattering materials assessed by time-resolved spectroscopy, the path length
differences were alleviated. Multicomponent least squares analysis was then used to
quantify each component of the spectrum. The results were still influenced by variations
in the mean light pathlength. Therefore, concentration ratios of the respective
components were normalized with respect to the light path. The concentrations of total
cerebral oxygenation, T-Hb, were determined from the concentrations of each
chromophore. T-Hb and the mean cerebral Hb oxygen saturation, SbC>2 , were accurately
determined both in vitro and in a newborn piglet brain. The mean cerebral Hb oxygen
saturation of newborn piglets at an oxygen / nitrogen ratio of 0.21 was determined to be
63 ± 4%. Umbilical arterial, and left internal jugular venous Hb oxygen saturation, were
simultaneously determined to be 95 ± 2% and 52 ± 11%, respectively.
158
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
5.5 Characteristics o f a novel PbS CCD array-based near infrared
spectrometer
Several real samples were used to characterize a prototype PbS CCD arraybased portable spectrometer [146]. Spectral differences were studied in natured and
denatured protein samples, as well as moisture content in skin samples, water content in
melons, plastics identification, and an approach to the determination of the thickness of
plastics was developed. In these preliminary studies chemometric methods were
deferred in favor of a study of basic characteristics of the spectrometer.
5.5.1 Experimental Design
The SM-301 miniature prototype PbS array spectrometer from CVI Laser
Corporation (Putnam, CT) used a 1000-micron fiber optic cable to collect light from the
sample. A spherical collimating mirror directed light onto a diffraction grating (1600
grooves/mm 120-run blaze angle), and then to an aspherical mirror and finally to a flat
mirror, which redirected the spectrum at a right angle to the PbS array, as shown in
Figure 5.2. The array was comprised of 256 pixels calibrated for the 1568 - 2008 nm
wavelength range and was thermoelectrically cooled to -10°C to reduce background
noise. Data were transferred via a 12-bit PCMCIA connection to a computer to collect
the data in real-time [147].
The Windows-based software provided for both data collection and data
manipulation. Data collection features included CCD temperature monitoring, repetitive
or single scanning, variable time averaging, binning, and integration time, as well as
dark current and reference scanning and a balancing function, to correct for both signal
159
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
bias and pixel response. Data could be collected using wavelength, wavenumbers, or
pixels on the x-axis and absorbance, scope mode (time), irradiance, transmittance, or
reflectance on the y-axis. Post-collection data analysis options included zoom and peak
pick, fast Fourier transform (FFT) smoothing, and export to ASCII files.
160
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
camera mirror
collimating
m ir ro r
entrance
siit n
mirror
PbS array
Figure 5.2: The portable near-infrared spectrometer. The light is directed into the
spectrometer through a fiber optic cable where it is collimated onto a diffraction grating,
directed across two mirrors onto the PbS array where 256 pixels convert the light into
electrical output.
161
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
The experimental setup was designed to be vertical in order to accommodate
horizontal sample orientation, as shown in Figure 5.3. A 500-W tungsten source was
used to illuminate the sample from beneath. The light was directed through a 0.5-cm
diameter hole in the source housing, focused by a Suprasil lens, passed through the
sample position, and focused onto the end of the fiber optic cable. The sample cell was
housed within a custom-designed slide holder, and mounted on a horizontal
translational stage. For each of the samples mentioned in this paper, the sample cell was
a standard pre-cleaned silica microscope slide. Silica was chosen for its negligible
absorbance across the entire detection region.
A custom-machined slide holder, which allowed standard silica slides to be
reproducibly positioned, was fastened to the translational stage with Teflon screws. The
stage had a 1300-micron range of movement and provided for surface scans across a
sample cross-section. The translational stage was fixed to an optical bench with a
magnetic mount, and adjustable brackets secured it. The fiber optic cable was a 1000micron diameter cable. While the cross-section of the fiber allowed the maximum
amount of light to enter the spectrometer, the size severely limited the instrument’s
spectral resolution, which is shown here later. A 50-mm focal length,/, Suprasil lens
was employed because it fully transmits light across the spectrometer’s range.
162
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
spectrometer
fiber optic
translation
7,
f9, Suprasil lens
light
source
Figure 5.3: The experimental arrangement consisted o f a 500-W tungsten bulb directed
through a 0.5 cm slit in the source housing, focused by a Suprasil lens, passed through a
sample, and focussed onto the end of a fiber optic cable.
163
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
A brief study was conducted to establish the optimum source position relative to the
spectrometer by imaging the highest intensity on the end of the fiber optic. Source
positions investigated were 2 / and beyond 2/, w here/ is the focal length of the Suprasil
lens. With the source located at twice the focal length (10 cm), an image the size of the
source (filament) was produced on the end of the fiber optic. At distances greater than
twice the lens’s focal length, a diminished image was produced on the fiber optic. These
smaller images were easier to completely focus into the fiber optic. Optimization of the
source locations was conducted with the equipment shown in Figure 5.3, with 40-Vdc
to drive the lamp. Figure 5.4 illustrates that by placing the source beyond twice the
lens’s focal length, a greater intensity of light was produced at the fiber optic cable.
While 2 / was suitable in intensity, increasing the source intensity decreases noise;
therefore, the source was placed beyond 2f. The source was placed about 15 cm from
the lens. All further studies were conducted with voltage at 120V with the source 15 cm
from the lens. This combination gave the maximum source intensity across the entire
range.
164
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
5000
S3
c3
4000
w
53
3000
.S
(U
2000
3o
1000
• ■■source at f (10 cm)
— source beyond f (15 cm)
<D
>
«-l
0
1568
1668
1768
1868
1968
wavelength (nm)
Figure 5.4: When the source is placed beyond twice the lens’s focal length, increased
intensity was produced at the fiber optic cable. The source was placed about 15 cm
from the lens.
165
Reproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission.
5.5.2 Samples
5.5.2.1 Moisture content in skin
Determination of water content is easily attainable since water absorbs across
the entire NIR region, therefore the determination of water appeared to be a good
candidate to test the spectrometer’s capabilities. Water content in skin is important in
that it affects the elastic properties of the outermost layer of skin, the stratum comeum
(SC) [86]. Several published studies have identified and quantified various types of
water associated with the different layers of skin, as discussed earlier in the review.
Moisture contents of skin samples obtained from the bicep area were investigated here.
Skin samples were obtained in a similar manner to a surface biopsy technique
[148,149]. A thin layer of cyanoacrylate glue (SuperGlue) was applied to a cooled glass
slide, pressed onto the bicep area for about 30 s and then slowly removed, as previously
employed by Sun et al. [150]. The slides were then placed on the translational stage for
analysis. The skin layer was examined for water content as a function of successive
layers of skin removed from the bicep area.
For skin removed from the bicep area, the water concentration decreased as
subsequent layers were removed. It can be concluded from Figure 5.5 that the most
water was inherent in the top layer of skin, where elasticity is most important, while
dryer skin was below. The water spectrum is superimposed in the figure to illustrate that
water, in fact, was the analyte responding in the skin samples. It can be seen from the
data shown here that the spectral resolution of the instrument was fairly poor.
166
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
Moisture content in calf skin samples were also investigated, however, the
spectra did not reveal the expected trend as seen in bicep skin. Upon examination of
calf-skin samples under a microscope, slides were found to contain multiple layers of
skin, varying from 3 to 25 layers. However, bicep skin was more reproducibly removed,
varying only between one and two layers.
167
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
1.4
1.2
layer 1
layer 2
layer 3
water
0.2
1568
1668
1768
1868
1968
wavelength (rnn)
Figure 5.5: NIR spectrum of skin layers in the bicep area. A water spectrum is
superimposed to illustrate that water was the analyte responding in the skin samples.
168
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
5.5.2.2 Water content in melons
Water content in fruit is important to the grocery industry and can be used in the
determination of freshness of fruits and vegetables. Excessive water content can lead to
fruit degradation. Since water absorbs across the entire range of the PbS array
spectrometer, moisture content as a function of thickness or time was investigated.
Honeydew and muskmelons were analyzed for water content. The melons were sliced
evenly using a custom-built slicer that allowed for adjustable thickness. Generally, 1
mm slices were used.
First, honeydew melons were held in a fixed position on the translational stage
to determine the effect of time on the water content in one area of the melon. As the
water evaporated at room temperature both absorbance and change in mass were
monitored. The honeydew melon slices were sliced into circles with diameter 1 mm (the
size of the fiber optic cable), so that the absorbance could be correlated to water loss.
An absorbance scan of these slices were recorded over a period of 70 minutes as shown
in Figure 5.6. Each slice was weighed on an analytical balance immediately before
recording the spectrum and these mass losses were associated with the absorbance
readings at three known water wavelengths: 1698, 1718, and 1748 nm. For the analysis
of melons it was difficult to cut thin slices of reproducible thickness as regions of high
fiber content were more rigid which resulted in thicker slices in those regions. Further,
melons are not uniform in composition between melons. Hence, absorbance
measurements were not directly comparable. Calibration curves were constructed to
correlate water mass with absorbance measurements. The mass of water was determined
169
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
by subtracting the dry-fiber mass value obtained at the end of the 70-min analysis.
While these calibration curves, shown in Figure 5.7, do not have the same origin they
have similar slopes, suggesting that interferences were present that affected the signal at
the water wavelength. This was not explored further.
170
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
0.25
1698 nm
1718 nm
1748 nm
i.20
8
S3
0.15
I
8 o.io
1.05
0
10
20
40
30
50
60
70
tim e (min.)
Figure 5.6: Absorbance for honeydew melon slices recorded for a 71-minute drying
period at each known water wavelength.
171
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
0.16
♦ 1698 ran
* 1718 nm
• 1748 nm
y = 3.2408x + 0.095
RM l.9626
•P 0.08
y = 4.1232x + 0.0564
R2=0.9716
.9083x +
R2=0.9609
0.002
0.004
0.006
0.008
0.01
mass of water (g)
Figure 5.7: Water content as a function of time in honeydew melons. Calibration
at three known water wavelengths: 1698,1718, and 1748 ran.
172
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
Muskmelons were used to compare the differences in water content across the
melon surface, beginning at the peel and moving inward towards the center of the
melon as a function of micrometer position. Here, Figure 5.8, 0 microns denotes the
position closest to the fruit’s skin, while 500 microns denotes the position nearest the
center. The data presents a “moving average” of absorbances since the fiber optic width
is 1000 microns, while the translational stage was only moved 100 microns per
measurement. Water content appeared to increase as the center of the fruit was
approached, as shown in Figure 5.8.
Diminutive changes in position changed spectral output significantly, thus interfering
with surface scans. No consistent change in spectral intensity as a function of specific
mass loss could be determined; this inconsistency was thought to be a function of melon
fiber content variation from melon to melon. The skin region tended to dry more
quickly than the inner region. In order to compensate for changes in regional drying
time, it was found that a sample should be extracted that was only large enough to be
illuminated to just match the fiber optic cable (1000 micron diameter) and not
significantly bigger. Also it was possible that sugars responded in the same spectral
region, and may have been interfering with the water spectrum.
173
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
0.9
1698 nm
0.6
<u
1718 nm
1748 nm
-
0 .6 *
I 04
'
|
Kj
0.3-
I
0.2 4
0
100
200
300
400
500
600
micrometer position (pm)
Figure 5.8: Muskmelon water content as a function o f sample position at the three
wavelengths associated with water: 1698,1718, and 1748 nm. Here, 0 microns denotes
the position closest to the fruit’s skin, while 500 microns denotes the position nearest
the center.
174
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
5.5*23 Plastics identification
The use of plastics is ever increasing and due to their lack of biodegradability,
recycling has become a crucial concern. A more efficient and effective method of
plastics separation is necessary to aid in subsequent plastics reuse. One consideration is
the cost effectiveness of the optical recognition system compared to the value of the
material being handled; therefore, there is a need for a relatively low-cost remote sensor
[87], Plastics can be identified as a function of structure in the NIR region. This study of
plastics attempted to identify characteristic peaks for several types of widely available
plastics. Plastics could be discriminated using NIR spectroscopy as the spectra are
structurally based. Each structure yielded a characteristic peak within the analytical
region. In general, the PbS-array spectrometer could distinguish among the various
polymers, although polymers that were similar in structure (CH bonds only) could not
be distinguished. The prototype spectrometer, which was calibrated for the small region
1.5 - 2.0 microns, was not well-suited for plastic determination, as many of the
characteristic bands occur below the spectrometer’s detection region. Specifically,
while the PbS detector can distinguish amongst plastics with some unique functional
group that has absorption in the region of 1.5 - 2.0 microns, a PbS detector cannot
distinguish among plastics with no functional group present on the carbon chain. A
better-suited detector for plastics separation would be a standard InGaAs detector of
wavelengths 0.8 - 1.7 microns, in which the characteristic peaks of the carbon chain can
be seen.
Thickness studies were conducted using poly-ethylene terephthalate (PET). PET
plastic bottles were cut into 1” by 1” squares. Thickness variance was achieved by
175
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
stacking squares upon each other and thickness was measured using a caliper. The
stacked plastic pieces were held tightly together using a clamp-spring holder, which
allowed the light to pass through the middle. An FTIR spectrum (Jasco FTIR 4 IOC) of
PET samples revealed a distinct spectral feature at 1660 nm, while a peak at 1618 nm
was relatively small by comparison, as seen in Figure 5.9A. This spectrum is
representative of a 1.4 mm thick sample of PET plastic. Both peaks increased in
intensity as the plastic thickness increased. The array provided little response between
1666 - 1670 nm, due to three known dead pixels in the array, where the FTIR analysis
provided a large response, as seen in Figure 5.9B. Additionally, the same analysis using
the PbS array provided response at 1614 nm rather than the 1618 nm peak evident in the
FTIR spectrum. This discrepancy indicates a 4-nm inaccuracy in the wavelength
calibration. Therefore, due to the dead pixels and slight wavelength discrepancy, the
large peak at 1660 was not observed. Also, it seems that the pixel response was more
sensitive for longer wavelengths. Calibration curves were constructed from data
collected on both the FTIR and the PbS spectrometer for the peak at 1618, as shown in
Figure 5.10. Calibration curves were corrected for an estimated 10% loss (due to
reflection) at each layer (5% per surface). For the PbS-array, when corrected for
reflection, the calibration curve slope approached zero, which suggested that there was
no absorbance by the plastic. For the FTIR spectrometer a linear calibration was
possible after correction for reflection losses. Rather than using layers of plastic, a piece
of uniform thickness would have been desirable to calibrate for plastics thickness,
which would have eliminated the reflection occurring between sheets of plastic. The
discrepancy between the two spectrometers was probably due to the superior quality of
176
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
the light gathering optics, and higher sensitivity, of the FTIR spectrometer but this
not at all clear, and the issue was not studied further.
177
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
1660 nm
(A)
<L>
1
,p
0.6
■S
0.4
1618
0.2
1600
1700
1800
1900
2000
wavelength (nm)
6.0
5.0
O
4.0
■s
3.0
2.0
1614 nm
0
1600
1668
1768
1868
1968
wavelength (nm)
Figure 5.9: (A) An FTIR spectrum of poly-ethylene terephthalate (PET) samples
revealed a spectral feature at 1660 nm, while a peak at 1618 nm was relatively small by
comparison. This is a representative spectrum of a 1.4 cm thick sample of PET. (B) A
NIR spectrum, with the PbS array spectrometer, of PET samples shows a slight
wavelength shift at the characteristic 1618-nm peak and lack of response near 1660 nm.
178
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
0.6
y = 0.201x
R2 = 0.9972
• 1618 Absorbance
■ Corrected 1618 Absorbance
y = 0.1741x
R2 = 0.9967
0
0.5
• Absorbance
1.4
Q
1.2
g
0-8
•§
2
1.5
1
thickness (mm)
2.5
y = 0.282x2 - 0.2165x
R2 = 0.9757
■ Corrected Absorbance
0.6
y = 0.0714x
0.4
R2 = 0.7307
0.2
0.5
1
1.5
2
2.5
thickness (mm)
Figure 5.10: (A) Calibration curves, corrected and uncorrected, are presented for data
collected from a benchtop Jasco FTIR for the same plastic samples at 1618nm. It was
assumed that there was a 5% loss at each surface, or 10% for each layer that was been
added for analysis. (B) Calibration curves, corrected and uncorrected, for data collected
from the PbS spectrometer at 1618 nm. The corrected absorbance curve accounted for
an estimated 5% reflection loss per surface in the sandwiched plastic. When corrected
for reflection, the calibration curve slope approached zero, which suggested that there
was no absorbance by the plastic.
179
Reproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission.
5.5.3 Discussion
The PbS-array spectrometer allows for a relatively economical instrumental
arrangement. Since silica transmits near 100 percent in the NIR region, standard
microscope slides allow for a very cost effective sample cell. An ordinary 120V 500-W
tungsten lamp provides high intensity IR light across the entire analysis region, thus
eliminating the need for an expensive light source.
The translational stage provided both horizontal and vertical sample positioning,
which provided for a large selection of samples and applications. The instrument could
also be adapted to analyze any physical state: solids, liquids, or gases. The software for
the instrument was flexible in that it measured absorbance, reflectance, transmittance,
and intensity. Another advantage, inherent in all CCD-based spectrometers, was near
real-time spectral output. The instrument had a time response in the millisecond time
scale, which may prove valuable in kinetics studies and other time-dependent
investigations.
Experimentally, variability in the spectra was found resulting from slight sample
movements, which prevented quantitative comparisons of sample spectra. For example,
the peaks associated with water in melon samples did not denote absolute water content
with respect to another melon sample. Absorbance values on two or more spectra could
not be directly compared; hence, the instrument was more suited for qualitative
analysis, rather than quantitative analysis.
Another disadvantage was the limited wavelength range of the spectrometer of
which water had broad absorption bands throughout. Therefore, the desired analyte
180
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
must either be water or samples must be sufficiently dried. This could be achieved by
using droplets on a slide surface, or solution drops between two glass slides, or by use
of short-path cells. Solids must be thinly sliced to minimize water content and to
maximize transmission of light.
Measurement of the change in signal as a function of mass of samples was
extremely sensitive to sample position when the sample was not uniform in
composition. It was our experience that the slightest vibration required recalibration of
the detector during water content analysis in melons. In this regard, the prototype
spectrometer is not well suited for field deployment where environmental factors play a
role. Plastic was more uniform and accordingly, analyses were not as sensitive to
detector vibration.
In general, spectra obtained on the PbS-array spectrometer were verified by use
of an FTIR spectrophotometer that covered a wider wavelength range in the NIR region.
Such instruments have much better resolution, and reproducibility than the portable
spectrometer.
An additional concern was the calibration of the individual pixels on the PbS
array. Pixel calibration determines where peaks will appear in the spectrum; hence,
calibration affects reliability of spectral output.
The spectrometer’s resolution was limited by the diameter of the fiber optic
cable. While larger diameter fibers allow more light to pass, the resultant spectral
resolution was poor. A smaller diameter fiber would have provided better resolution,
but the signal would have been attenuated accordingly.
181
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
The low resolution of the spectrometer also impeded identification of the exact
wavelength where a peak may have been present. For example, in the plastic thickness
studies, the FTIR benchtop instrument indicated a PET peak at 1614, while the CVI
spectrometer displayed the peak at 1618. This may have been due partially to resolution
issues, as well as calibration issues.
5.5.4
Conclusion
Through the use of computers and new spectrometer technology, low cost
instrumental NIRS is rapidly developing into a viable analytical technique, as indicated
in this review. For some applications such instrumentation might replace timeconsuming, sample preparation-intensive techniques such as HPLC, GC, and other wet
chemistry methods, and the approach has the potential to be applicable to a wide variety
of samples as shown here. New cost-efficient spectrometers now available will allow
for both on- and inline process and quality control. With regard to the characterization
of the Spectrometer described here, the spectral resolution was limited by slit width or
fiber diameter. Using an InGaAs array, which is becoming more widely available and
more affordable, would increase the sensitivity, signal-to-noise ratio, and sampling
rates. Enhanced spectral resolution through better alignment procedures is currently in
development.
182
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
APPENDIX A. RESONANT LASER ABLATION WITH
INDUCTIVELY COUPLED MASS SPECTROMETER
DETECTION USING A NOVEL CELL DESIGN
A .l Introduction
The major drawback of commercially available laser ablation inductively
coupled plasma mass spectrometer (LA-ICP-MS) instruments is the non-stoichiometric
vaporization phenomenon known as fractionation, which ultimately affects calibration
and accuracy, as previously discussed in Chapter 3. Several approaches have been
developed to diminish the effects of fractionation including increasing the transportation
efficiency and varying laser parameters, such as wavelength, spot size, and power
density.
It has been previously demonstrated that UV lasers provide more efficient
material removal from a sample compared to visible lasers - a frequency doubled (355
nm) or tripled (266 nm) Nd:YAG laser both cause less fractionation effects [4]
compared to their 1064-nm, fundamental wavelength, Nd:YAG counterpart. However,
distinct differences in fractionation using various UV laser wavelengths have been
shown to be matrix as well as power density dependent [151].
Also, transport efficiency has become an integral part of diminishing the effects
of fractionation. Optimization of transfer tube length, diameter, and tubing material as
well as gas mixture, has all been demonstrated to be critical factors [152], Primarily,
there has been one basic design of ablation chamber used for LA-ICP-MS, shown in
Figure A .l, with minor variations of flow path of carrier gas flow.
183
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
Figure A .l: Traditional cell designs used in LA-ICP-MS experiments. Typical cell
volume ranges from 0.2 to 12 cm3.
184
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
The inherent shape of the cell lends to turbulence, resulting in increased
collisions whereby possibly increased fractionation. This time-dependent fractionation
effect has been demonstrated, although cell geometry has been overlooked [6].
Therefore, by decreasing the volume of the cell, the flow efficiency has been shown to
improve, as measured by the width of the transient signals which become narrower as a
result of increased efficiency [152]. However, decreasing the cell volume also presents
two major limitations: sample size and range of ablation angles.
Laser ablation is inherently useful for localized microanalysis, however, the
minuscule amount of ablated material may not be truly representative of the bulk
sample, which may be inhomogeneous. Therefore, a small sample size severely limits
methods of bulk analysis commonly employed when using LA sampling, such as
rastering of the laser or sample in order to obtain an “average” of the analyte present on
the sample surface.
Sample angle and polarization have also been shown to affect the emission
signal, as discussed in Chapter 3. The traditional ablation cell designs restrict the
capability to use a wide range of angles to exploit the increased laser-material
interaction. Although the laser angle may be changed, the beam may be attenuated
since it may pass through a greater thickness of quartz window, which may ultimately
pose a problem, since very low laser powers are required for RLA, compared to
arbitrary-wavelength normal ablation. This problem may be overcome by angling the
sample, although the flow pattern and cell volume can be dramatically changed. In
order to address these issues, and increase the analytical efficacy of LA-ICP-MS,
resonant laser ablation (RLA) can be coupled with a novel cell design, discussed here.
185
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
As discussed in Chapter 1, the selective enhancement of the RLA technique has
been demonstrated using time of flight mass spectrometry (TOF-MS), microwave
induced plasma atomic emission spectroscopy (MIP-AES), and laser induced
breakdown spectroscopy (LIBS). However, RLA has not yet been coupled with ICPMS. By selectively inducing fractionation, RLA may offer several advantages for LAICP-MS including a decrease in matrix interferences resulting from a simplification of
mass spectra and decrease in limits of detection (LODs), as seen using TOF-MS
detection.
In addition to RLA’s intrinsic selectivity, the novel cell described here may
provide greater transport efficiency over previous designs due to the shape of the cell,
which might provide for a non-turbulent carrier gas flow profile through it. Also, other
parameters of the RLA phenomenon can also be easily exploited with this design, such
as sample angle and laser polarization.
A.2 Novel Resonant Laser Ablation ICP-MS Cell Design
The body of the cell, shown in Figure A.2, is made of Deiron® and fitted with a
half-round, 1.5-mm thick quartz window directly above the sample holder. A
photograph of the cell is shown in Figure A.3. Argon is introduced into the cell through
a quarter-inch Swagelok fitting, passes through a series of baffles, then through a halfround tube array of half-inch, 18-gauge, 316 stainless steel hypodermic tubing before
entraining the ablated material. A detailed look at the inside of the cell body is shown
in Figure A.4.
186
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
The tube array offers several advantages including a laminar flow of argon
across the sample surface and also prevention of window “fogging” with ablated
material. The entrained ablated material is then focused through a half funnel before
entering the transport tube, shown in Figure A.5. The total volume of the cell including
the focusing funnel is 5 cm3, which is in the range of cells previously used for LA-ICPMS studies, which is 0.07 - 10 cm3.
187
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
18 gauge
hypodermic tubing
half round quartz window
sweep gas inlet
sample holder
Figure A.2: Novel cell design for use with RLA-ICP-MS.
188
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
V ” Swagelok argon inlet
,
quartz window
sample holder
angle graduations
Figure A.3: The cell body is calibrated in 3° increments and the cell holder allows for
complete rotation of the cell body. The base of the cell holder can be mounted directly
to an optical bench.
189
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
stainless steel
tube array
argon inlet baffles
O-ring seal
(A)
(B)
Figure A.4: (A) The argon sweep gas enters the body o f the cell through a
Swagelok
fitting before passing through a series of baffles. The insert is mounted with a series of
set screws sealed with an O-ring fitting into (B), where the argon then passes through
the tube array, which is composed 316 stainless steel hypodermic tubing lA ” in length.
190
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
transport tube
ICP-MS
argon sweep gas
Figure A.5: Experimental arrangement for RLA-ICP-MS incorporating the novel cell
design.
191
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
The sample cell holder consists of a sample cup holder which can accommodate
5/8”-diameter samples, as used in the RLA-MIP-AES setup, so that samples can be
used interchangeably between experimental arrangements. The shaft of the holder is
equipped with friction-fit O-ring seal to allow for ease of sample replacement.
The body of the cell has been mounted on a stand, which can be affixed to an
optical bench. Also, the body has been calibrated in 3°-increments so that simple
rotation of the entire cell body results in sample angle rotation. It has been previously
demonstrated in Chapter 3 that the laser-material interaction can be increased by cell
rotation; however, previous sample designs could not accommodate wide range of
angles. Also, the half-round window provides uniform window thickness as the cell is
rotated, thereby minimizing laser attenuation as a function of sample angle.
Other parameters, which have been demonstrated to affect the mass spectra peak
shape, are transport tube length and diameter, as well as argon flow rate through the
cell, which need to be addressed to fully optimize the transport efficiency.
A.3 Efficiency Test of Novel RLA-ICP-MS Cell Design
In order to test the efficiency of this new cell design, a traditional LA-ICP-MS
cell must be employed for comparison. Zhou had designed a cell of this type for the
determination of lead in bovine liver using a laser diode ablation-deposition system
[153]. However, in order to critically compare the efficiency of the two cell types, the
cell volumes must be equivalent. A teflon insert was placed in a second lab-constructed
cell to reduce the volume to match that of the new cell design, 5 cm3. The underside of
192
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
the insert was machined so that a sample may be inserted to allow the argon gas to flow
directly over the surface of the sample sweeping the ablated material to the outlet port.
A 1.5-mm thick quartz window was glued to the top of the cell, which was the same
thickness of the half-round quartz window used in the new cell design, so that laser
attenuation when ablating normal to the sample surface is comparable between cells.
The in-house built traditional cell design is shown in Figure A.6.
A comparison of transport efficiency can be conducted while ablating a pure
metal sample using the OPO laser in single-shot mode at an arbitrary UV wavelength
while maintaining other transport parameters, such as transport tube length and diameter
and argon flow rate. The transient signal produced can provide evidence of efficiency
of each cell by measuring peak height, area, and FWHM.
193
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
quartz window
argon outlet
teflon insert
glass cell
argon inlet
inset sample holder
Figure A.6: In-house built traditional cell for efficiency comparison.
194
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
A.4 Resonant Effect in LA-ICP-MS
It has been demonstrated in Chapter 3 that RLA results in an increased signal,
possibly due to an increased number of atoms, which is evident when employing MIPAES detection. Coupling this technique to ICP-MS may lend further insight into the
mechanism by which this occurs. Since the ICP torch ionizes the introduced material
indiscriminately, much like the MIP, the intensity of the signal produced is directly
proportional to the amount of material that has been ablated. The difference here,
however, is that there is little perturbation of the plasma upon sample introduction.
Therefore if a strong laser-material interaction were present, the intensity of the
resultant mass spectrum would reflect it. Conversely, if the resonant enhancement
phenomenon occurs in the gas phase, no distinct differences would be seen between the
resonant and non-resonant ablation. In order to substantiate this theory, a transient
signal may be measured while scanning the wavelength of the ablation laser through the
resonant gas phase transition of the analyte in a pure metal sample.
Typically, non-resonant LA relies on high-powered UV lasers ranging in power
density up to 10 GW c m 2, yet RLA-ICP-MS will be a much “softer” form of sample
vaporization since it has been previously shown that lower laser energies are required to
yield significant ablation, in the range of 10 MW cm'2 [8], The ion yield will be
observed based on a function of wavelength and power density to determine optimum
conditions.
195
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
A.5 RLA-ICP-MS Calibration
Sample introduction into the ICP-MS is normally accomplished by dissolving
the solid sample, typically through acid digestion methods, or by use of slurries. The
sample is then aspirated through a nebulizer into the spray chamber before entering the
plasma torch. The basic setup of an ICP spray chamber is shown in Figure A.7.
When employing a solid sampling technique, such as LA, two types of
calibration may be employed: matrix-matched solid samples and aqueous calibration.
Matrix-matched standards are most commonly employed in LA-ICP-MS since ablation
yield varies with the sample matrix [154], In instances when these standards have not
been available, fused-glass beads [155] or pressed pellets [156,157,158] of standard and
matrix materials have been prepared. In-house available NIST stainless steel samples
of varying trace metal content may be used as matrix-matched standards, as well as
commercially available standards for solid sampling analysis techniques including those
that have been developed for LA, electrothermal vaporization (ETV), and glow
discharge (GD). The argon-entrained dry aerosol produced by these methods can be
introduced directly into the plasma torch, by passing an argon sweep gas over the
sample surface directly into the torch, as shown in Figure A.8.
Aqueous calibration provides for an attractive alternative where standard
addition can be accomplished by introducing standard solutions [159]. This is achieved
by mixing the dry aerosol from the ablation chamber with the nebulized aqueous
solution using a T-joint, where the standard and dry aerosol are mixed. This calibration
scheme is shown in Figure A.9. Although this provides a viable alternative, its efficacy
196
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
is hindered by the response of the calibration using a mixture of a wet and dry aerosol
resulting in the formation of oxides. Several methods have been employed to desolvate
the wet aerosol including the use of a desolvation unit in-line between ICP and the Tjoint [158].
197
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
plasma torch
spray chamber
aux gas
cooling gas
sample
uptake
Figure A.7: ICP-MS glassware.
198
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
Ar + ablated
7 material
cooling gas
aux gas
Figure A.8: ICP-MS arrangement for the introduction of vaporized solid sample.
199
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
Ar + ablated
material
aqueous
standard
cooling gas
Figure A.9: ICP-MS arrangement for aqueous standard addition calibration.
200
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
A. 6 Quantitative Multi-Element Analysis
Since the ICP-MS is capable of multi-element detection, a two-component
material, such as brass, or multi-component material, such as stainless steel, may be
used. Ablation at the analyte’s resonant wavelength, while monitoring several other
constituent’s masses may lend useful information to better account for the mechanism.
However, because the lifetime of the transient signal is only on the order of 1 - 3 s,
methods of transient extension may be applied.
A single bead string reactor (SBSR) has been used to extend the transient signal
produced by LA up to 5 s [160], The SBSR consists of a short length of tubing (35 cm
long x 1 cm inner diameter) packed with 4 mm glass beads. The principle behind the
SBSR is that a turbulent flow is created which transforms the normally skewed transient
to a more Gaussian-shaped profile.
Another design consists of the use of a 100-mL round bottom flask in-line
between an electrothermal vaporization furnace (ETV) and the ICP torch. The transient
extension (Tex) chamber has demonstrated its abilities to extend the transient signal to
nearly 20 s, however the signal still remains skewed.
The disadvantage of these two approaches is the constantly changing transient
signal intensity, which makes quantitative mass scans difficult to obtain. Balasanek et
al. have developed a piston-like chamber, in which the dry aerosol is collected from the
ETV in the piston chamber then pumped into the ICP, resulting in a square wave [161].
The length of the transient square wave is dependent on the flow rate of the carrier gas,
albeit affecting the intensity of the transient. Through the use of this square wave
201
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
generator, full quantitative mass scans of 248 masses were acquired with precision of
better than 11% across the mass range.
A. 7 Conclusion
The research outlined here represents the future direction of the use of RLA for
practical quantitative analysis, none of which has been previously reported in the
literature. These experiments not only will lend insight into the mechanism of RLA, but
also provide a more efficient means of sample introduction, since much lower energies
are required for RLA compared to non-RLA. However, the instrumentation described
here can be beneficially applied universally to non-RLA techniques for the reasons
outlined above.
202
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
APPENDIX B. RESONANT LASER ABLATION
APPARATUS ALIGNMENT
B .l Introduction
One o f the difficulties with the first generation o f RLA experimental
arrangement in the RGM lab was the lack of flexibility and mechanical limitations
affecting alignment issues, and ultimately the reproducibility of both ablation and
emission scans. One of the initial objectives for this author was to design and develop
an experimental arrangement that allowed (1) ease of sample replacement; (2) ability to
rotate sample without disturbing the vacuum; (3) precise alignment o f the ellipsoidal
mirror for MIP image control; and (4) focal point adjustment of the ablation laser beam.
Although the final design met each of these key requirements, the complexity of the
initial alignment was an arduous task, albeit much better controlled than the original
design. This appendix has been written to outline alignment procedures in order to
mitigate future alignment difficulties.
The basic layout o f the RLA arrangement is shown in Figure B. 1. The arrows in
the figure represent the movable parts and the axes upon which the movement occurs.
It can be seen from this figure that the prism which directs the beam normal to the
sample surface, and the ellipsoidal mirror, are both fully translatable, which allows for
great flexibility, but tends to make alignment more difficult. However, if the alignment
is performed left to right, that is, the laser followed by detection, then the efficiency of
the alignment process can be greatly increased.
203
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
beam focusing lens
laser beam
90° ellipsoidal mirror
optical rail
monochromator
emission
—f
o
Figure B.l: Overall RLA experimental arrangement.
204
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
B.2 Laser Alignment
The alignment of the laser beam is critical to ensure that it passes through the
iris, ellipsoidal mirror, and the ablation chamber with sufficient clearance so that the
beam is not clipped. Before one can begin alignment of the beam at the ablation
chamber, care must be taken to ensure that the beam is at a 90° angle with respect to the
first prism. To do this, significantly reduce the radius of the iris and place a pinhole
along the beam axis on the optical bench after the iris but before the beam passes
through the focusing lens. This is shown in Figure B.2. If the beam cannot be directed
by adjusting the right angle prism, then the Pellin-Broca crystal in the OPO laser must
be aligned. This procedure, as well as other basic laser operating parameters, is
discussed in Appendix C.
205
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
right angle
prism
iris
pinhole
laser
Figure B.2: Alignment o f the laser beam through the focusing lens.
206
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
Once the laser beam is properly aligned through the focusing lens, it must be
directed onto the second right angle prism to direct the beam through the ellipsoidal
mirror and into the ablation chamber. The difficulty here arises since this right angle
prism is mounted on a tilt/rotation stage, which allows rotation in every direction. Care
must be taken to ensure that the beam is directed at a right angle so that it may pass
through the ellipsoidal mirror. The comer edge of a business card may be used to
monitor the position of the beam to ensure that it passes completely through hole in the
ellipsoidal mirror. Refer to Figure B.3 for the action of the individual adjustment knobs
on the mount. Proceed with the alignment as follows: place a level perpendicularly to
the ground on the face normal to the incident beam and adjust (B) until level, then place
the level perpendicularly to the ground on the adjacent face, while adjusting (A). Knob
(C) may be adjusted while monitoring the laser beam using the business card, and later
fine-tuned.
207
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
(A)
(A)
(B)
(B)
(Q
(Q
Figure B.3: Tilt/rotation prism stage knob adjustments for directing the beam through
the ellipsoidal mirror into the ablation chamber.
208
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
Once the laser beam is properly directed at a right angle downward, it should
pass directly through the ellipsoidal mirror. Since the ellipsoidal mirror is mounted on a
separate, much smaller optical bench, it is subject to “human interaction” and any subtle
bump may affect the beam alignment. Depending on the severity of the misalignment,
adjustment of tilt/rotation stage may be required to correct the alignment. However, if it
is grossly misaligned, the optical bench may need to be slightly shifted.
Now that beam is directed through the ellipsoidal mirror, the final step in the
beam alignment is directing it into the ablation chamber and onto the sample surface.
The entire ablation apparatus is fixed to the aluminum baseplate held with four set­
screws. Although complete removal of the apparatus isn’t suggested for routine
alignment, gross adjustment of the setscrews is a viable option in situations where the
apparatus has been removed for some reason. The LIP may be difficult to see on the
surface of the metal sample, so it is hard to determine if clipping is occurring. A
drawing of what clipping looks like is shown in Figure B.4. The glass cell must be
replaced with the three-quarter round glass cell, so that a card may be inserted to check
for clipping. This is shown in Figure B.5. Slight manipulation of the tilt/rotation stage
is required to correct this problem.
209
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
(A)
(B)
Figure B.4: A business card is used to ensure that the laser beam is passing through
unscathed (A) and is not being clipped as it passes through the ellipsoidal mirror and
ablation chamber (B).
210
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
%-round glass cell
■M
, s
Figure B.5: The glass cell is replaced with a 3A round glass cell and a business card is
placed over the sample in order to check for beam clipping.
211
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
B.3 MIP Emission Alignment
After completing the laser beam alignment, the MIP emission must be properly
aligned. Since the focal points of the ellipsoidal mirrors are fixed, the monochromator
and the MIP must be moved in order to ensure focusing. However, since the
monochromator is bulky, it can be replaced by a business card in order to do this. The
ellipsoidal mirror is mounted on a kinematic mount precisely angled above the ablation
chamber. A pictorial description of the mount and the focal points of the mirror are
shown in Figure B.6. The focal points for the mirror are 140 mm and 260 mm,
corresponding to the vertical and horizontal focal points, respectively. The focal points
can be roughly adjusted by measuring the distances from the center of the mirror to 45
mm above the ablation cell base, which is roughly the height of the center of the MIP
cavity, and to the position where the monochromator is to be placed.
At this point, the glass cell and MIP should be removed and the MIP alignment
tool should be placed though the Cajon fitting on the underside of the ablation cell base,
The alignment tool is a penlight powered with a variable DC power supply that has
crosshairs to indicate the center position of the MIP cavity. Once turned on, tuning the
height of the ablation chamber along the rails can focus the crosshairs onto the slit of
the monochromator. Care must be taken to insure that both axes of the crosshairs are
best focused. This setup is shown in Figure B.7.
212
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
\
/
\
/
/
/
Figure B.6: The kinematic-mounted ellipsoidal mirror knob adjusts.
213
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
Figure B.7: Crosshair image adjust on monochromator slit. The inset demonstrates
possible crosshair images. Images resembling (A) and (B) demonstrate that only one
axis is focused, while image (C) shows proper focusing o f both axes.
214
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
Once the image of the crosshairs is properly focused, the set-screws may
be tightened to secure the height of the ablation chamber platform
215
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
APPENDIX C. BASIC OPO OPERATION
C.l Introduction
This appendix is to serve as a quick reference guide to basic laser operations for
the Spectra Physics MOPO-730 Optical Parametric Oscillator Laser. New users should
become familiar with fundamentals of the parametric process, which can be found in
References [162,163] A more detailed discussion of the techniques described here can
be found in the operating manual [164].
The basic layout of the MOPO controller is shown in Figure C .l. This is the
screen that the operator would observe after initial startup of the controller. The largefont number in the upper right represents the current wavelength, which is the
wavelength of the “signal” beam, while the small-font number below this represents the
wavelength of the “idler” beam. The series of three vertical buttons on the left side of
the controller are the operation mode keys; the series of five horizontal buttons are the
control keys; and the arrow keys are the adjust keys.
216
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
H
OPERATE
OPERATE
■PH
o U o .Z U U
SETUP
—
H I
| |
SCAN
SETUP
MONITOR
MONITOR
GOTO
308.200
u
DT
START
SCAN
MOVE
■
■
□
616.400nm
mm
RECALL
SAVE
0
0
■
y
■
Figure C. 1: Configuration o f the control panel.
C.2 Writing Tables
When a new wavelength range is desired, a table must be written to optimize the
laser energy in that given wavelength range. On the MOPO controller, select the SCAN
SETUP operational mode key. The menu above the control keys will then allow you to
select the wavelength range. For typical RLA scans, a 3-nm minimum range is
required, centered on the resonant transition of the analtye. For example, as discussed
in Chapter 3, the resonant gas phase transition for molybdenum is 313.259 nm;
therefore, this wavelength will be roughly the midpoint of the 3-nm wavelength scan
range. To enter this range in the MOPO controller, as shown in Figure C.2, depress the
buttons below the BEGIN until a beep is audible. The hundreds place of the wavelength
217
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
will then be underlined. Use the arrow adjust keys to toggle to the correct number, then
select the button below BEGIN again to toggle to the tens place. Repeat this process
until the starting wavelength of your scan range has been selected, then hold the button
below BEGIN until a beep is audible. In the case o f the molybdenum scan, “311.800
nm” should be entered. Repeat this procedure to enter the ending wavelength of your
wavelength scan, which would be 314.800 nm for molybdenum.
H
OPERATE
OPERATE
o U o .Z U U
SETUP
n |
■
SCAN
SETUP
SCANS
MONITOR
DT
MONITOR
1
■
□
616.400nm
BEGIN
282.000
END
284.000
CONT.
0.010
SHOTS
0
■
■
■
■
m
□
Figure C.2: Wavelength range selection screen.
Once the wavelength range has been selected, press the OPERATE mode key twice.
The display will once again change to that shown in Figure C.3.
218
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
OPERATE
OPERATE
SETUP
SCAN
SETUP
MONITOR
MONITOR
DEVICE
36C R Y S
MOD =
WVL =
SET =
LPT
=
METHOD
LAGRNG
IDL 0
282000
9826
18144
ADJUST
CRY =
PRS =
CONT.
2112
18156
ABORT
□
□
Figure C.3: Table writing display.
Two crystals must be tuned in order to fully optimize the energy in the given
wavelength range: the OPO crystal and either the 56° or the 36° crystal based on the
chosen wavelength range. The 36° crystal is used for the 274.000 nm - 345.000 nm
range, while the 56° crystal is used for 220.000 nm - 270.000 nm wavelength range.
Select the button below the DEVICE and use the arrow keys to toggle between devices
then hold the button below the appropriate device until a beep is audible. The
METHOD window will then be highlighted. The method chosen is based on the device
that is selected, however, for OPO, 36°, and 56° crystals, the Lagrangian method of
calibration, LAGRNG, will be employed. Using the arrow keys to select the
appropriate method, depress the key below METHOD until a beep is audible. It should
be noted that the minimum wavelength range for adjusting the OPO crystal is 6 nm,
while the minimum wavelength range for the 36° and 56° crystal is 1 nm, therefore the
219
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
wavelength range may need to be adjusted accordingly. The ADJUST window will
then be highlighted. Press the key below ADJUST and energy bars will be displayed,
as indicated in Figure C.4.
OPERATE
OPERATE
SETUP
SCAN
SETUP
MONITOR
MONITOR
DEVICE
36 CRYS
MOD
WVL
SET
LPT
METHOD
LAGRNG
IDLO
282000
9826
18144
ADJUST
I
CONT.
40
ABORT
□
□
Figure C.4: Optimization of laser energy during table writing procedures.
As indicated in Figure C.4, the arrow keys are to be used in order to maximize the laser
energy in the energy bar. The down arrow key must first be depressed for several
seconds to alleviate any mechanical hysterisis; the meter will decrease consequently.
The up arrow is then depressed to maximize the laser energy. After sufficient energy is
acquired, press CONT until a beep is audible. The wavelength will then automatically
220
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
advance several hundred picometers and this process must be continued until the energy
has been optimized across the selected wavelength range.
Once the laser energy has been optimized for the OPO crystal, the device may
then be changed to optimize the laser energy using the 36° or 56° frequency doubling
crystals, denoted as 36_CRYS and 56_CRYS, respectively, in the exact same manner as
outlined for the OPO crystal.
C.3 Beam Steering
As seen throughout this thesis, optical alignment is critical and it begins with
proper alignment to ensure the laser beam is correctly pointed as it comes out of the
OPO laser at the exit port. This is accomplished by steering the Pelin-Brocca crystal.
Assuming alignment of beam steering optics to the ablation chamber is accurate, a
business card with crosshairs may be placed below the ellipsoidal mirror. Select the
appropriate wavelength range, however, a 6-nm minimum wavelength range is required,
and select PB_CRYS from the DEVICE menu. Here, the calibration method for tuning
the Pellin Broaca will be the LSQ_MRQ method. Again, the down arrow key must be
depressed for several seconds to alleviate any mechanical hysterisis, after which the
appropriate arrow key may be depressed based on the direction of steering desired,
while observing the beam spot on the business card.
221
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
APPENDIX D. COMPUTER-CONTROLLED STEPPER
MOTOR
D .l Introduction
A stepper motor control board, Aerotech 1401/4003 stepping motor translator
board was modified so that it may be interfaced to a computer through an RS-232 cable
to allow precise motion control of the sample position in the ablation chamber, as well
as computerized laser attenuation. The specific application for attenuation and sample
control will be discussed later.
A four-phase stepper motor (Eastern Air Devices, Dover, NH) was used to drive
the appropriate device. The switching sequence for the clockwise rotation of the
stepper motor is shown in Table D .l, accompanied by the motor wiring diagram in
Figure D.l.
A control board (RMV Electronics, Vancouver, BC, Canada) was used an
interface between the stepper motor translator board and the computer via a DB-9 RS232 cable. The control board was connected directly to the four power transistors on
the translator board used to drive the motor. The control board provided pulses to the
appropriate transistor as indicated in Table D .l to control direction and speed of the
motor. A schematic o f the board layout is shown in Figure D.2.
222
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
Table D .l: Switching sequence for four-phase stepper motor.
Step
SW1
SW2
SW3
SW4
1
on
off
on
off
2
on
off
off
on
3
off
on
off
on
4
off
on
on
off
SW 1
SW 2
SW 3
SW 4
Figure D.l: Four-phase stepper motor wiring diagram.
223
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
power transistors
output to motor
RS-232
control board
GND
AC
{
stepper motor
translator board
Figure D.2: Stepper motor translator with control board modification for computer
interface. The color codes are the same as for those shown in Figure D.l.
D.2 Software
RMV Electronics also provided 32-bit software, Sport232 vl.O, for the stepper
motor control board so that speed and direction can be easily controlled. The COMM
tab, shown in Figure D.3, allows one to select which serial port is being occupied by the
control board as well as the baud rate.
224
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
«
s l'H I M
>
i iin tm i liy .in l
m
PRA
Figure D.3: Port and baud rate selection screen.
The analog-to-digital conversion (ADC) and PULSES tabs were not used for this
application, although this board is capable of these functions. The STEPPERS tab
contains the stepper motor functionality of the board, shown in Figure D.4.
MJiS
CQiM
r
.. .........
Step tape
Spswet
faM Speed:
poo
• Motjofl CofTStVidnds . .....- ..
SWIPfl#*
r .. ....
|400
po““
pH” — ’
r Matte------- ------ — -
M#¥h8irirtian
■ J s J ...a ..1
L ia e J
■ rnm ................... ........... --
f? HaifRsp
C Monophase
C IWwte
I
PRA
Figure D.4: Stepper motor control screen.
225
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
The speed control is very flexible, allowing for even an increasing speed as the stepping
progresses, although this functionality has not been used for the author’s purpose. Once
the motion parameters have been entered, the stepping is initiated by choosing the
appropriate direction of the motion.
D.3 Attenuator Control
A broadband variable optical attenuator (Newport, Model 935-3, Fountain Valley, CA)
was used to control the laser energy by manually adjusting the attenuation. This
attenuator has been characterized elsewhere [165]. In order to automate the attenuator,
the adjustment knob was replaced with a shaft fitted to the stepper motor, as shown in
Figure D.5.
Figure D.5: Computer controlled broadband attenuator. The beam path through the
attenuator is shown in the insert.
226
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
A glass slide was used to direct 10% of the laser beam into a power meter. The
output o f the power meter was then directed into a boxcar integrator so the energy of
each individual pulse may be monitored as a function o f the attenuator. The number of
steps required to achieve the attenuator’s full range of motion was determined to be 750
steps. This was programmed into the stepper motor software, and the attenuator was
scanned at a rate o f 250 steps per minute. A plot of attenuation with respect to stepper
motor position is shown in Figure D.6.
0.25
0.2
V)
I
.S3
<5
2ra
-
0.15 0.1
-
OJ
>
'S 0.05 -
a
0
200
400
600
800
stepper motor position
Figure D.6: Laser attenuation as a function of stepper motor position.
From this plot, precise stepper motor precision can be determined for desired laser
energy.
227
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
D.4 Sample Holder Rotation
A stepper motor from Eastern Air Devices, model LA23ACK-2 (Dover, NH),
was mounted to the shaft of a sample holder in order to obtain a fresh sample surface,
which was particularly useful when obtaining craters for image studies. This is
demonstrated in Figure D.7.
sample cup
shaft
stepper motor
Figure D.7: Stepper motor mounted to the sample holder.
When performing these studies, the stepper motor was advanced 100 steps, using the
stepper motor control software, after 1000 shots at a particular location.
228
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
REFERENCES
[1] X.L. Mao, A. C. Ciocan, R.E. Russo, “Preferential Vaporization During Laser
Ablation Inductively Coupled Plasma Atomic Emission Spectroscopy.” Appl. Spec. 52
(7), 1998,913-918.
[2] E.F. Cromwell, P. Arrowsmith, “Fractionation Effects in Laser Ablation Inductively
Coupled Plasma Mass Spectrometry.” Appl. Spec. 49 (11), 1995, 1652 - 1660.
[3] S.M. Eggins, L. P. I. Kinsley, J.M.G. Shelley, “Deposition and Element
Fractionation Processes During Atomospheric Pressure Laser Sampling Analysis by
ICP-MS.” Appl. Surf. Sci. 127-129, 1998, 278 - 286.
[4] D. Figg, M. S. Kahr, “Elemental Fractionation of Glass Using Laser Ablation
Inductively Coupled Plasma Mass Spectrometry.” Appl. Spec. 51(8), 1997, 1185 1192.
[5] D. Gunther, R. Frischknecht, C.A. Heinrich, H-J. Kahlert, “Capabilities of an Argon
Fluoride 193 nm Excimer Laser for Laser Ablation Inductively Coupled Mass
Spectrometry of Geological Materials.” J. Anal. At. Spectrom. 12 (9), 1997, 939 - 944.
[6] D. Gunther, C.A. Heinrich, “Comparison of the ablation behavior of 266 nm
Nd:YAG and 193 nm ArF excimer lasers for LA-ICP-MS analysis.” J. Anal. At.
Spectrom. 1999, 14 (9), 1369 - 1374.
[7] F.R. Verdun, G. Krier, J.F. Muller, “Increased Sensitivity in Laser Microprobe Mass
Analysis by Using Resonant Two-Photon Ionization Processes.” Anal. Chem. 59 (10)
1987 1383- 1387.
229
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
[8] I.S. Borthwick, K.W.D. Ledingham, R.P. Singhal, “Resonant laser ablation - a
novel surface analytic technique.” Spectrochim. Acta B 47B (11), 1992, 1259 - 1265.
[9] C.G. Gill, T.M. Allen, J.E. Anderson, T.N. Taylor, P.B. Kelly, N.S. Nogar, “Lowpower resonant laser ablation of copper.” Appl. Optics 35 (12), 1996, 2069 - 2082.
[10] G.C. Eiden, J.E. Anderson, N.S. Nogar. “Resonant laser ablation: semiquantitative
aspects and threshold effects.” Microchem. J. 50 (3), 1994, 289 - 300.
[11] M. Campbell, R-E. Zheng, K.W.D. Ledingham, ‘The detection of trace
manganese in various matrices using 45° resonant laser ablation-time-of-flight
spectrometer.” Section 8: Industrial Applications of Sensors. Paper presented at
Sensors and their Applications XI, Sept. 2001 237 - 243.
[12] H-M. Pang, E.S. Yeung, “Laser-Enhanced Ionization as a Diagnostic Tool in Laser
Generated Plumes.” Anal. Chem. 61 (22), 1989, 2546 - 2551.
[13] C.J. McLean, J.H. Marsh, A.P. Land, A. Clark, R. Jennings, K.W.D. Ledingham,
P.T. McCombes, A. Marshall, R.P. Singhal, M. Towrie. “Resonant Laser Ablation.” Int.
J. Mass Spectrom. Ion Proc. 96 (1), 1990, R1 - R7.
[14] L. Wang, I.S. Borthwick, R. Jennings, P.T. McCombes, K.W.D. Ledingham, R.P.
Singhal, C.J. McLean, “Observations and Analysis of Resonant Laser Ablation of
GaAs.” Appl. Phys. B 53 (1), 1991, 34 - 38.
[15] M. Capitelli, F. Capitelli, A.V. Eletskii, “Photoresonant Plasma as a New Tool for
LIBS.” Laser Phys. 10 (6), 2000, 1244 - 1250.
230
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
[16] D.Y. Tsipenyuk, M.A. Davydov, “Resonant laser-induced breakdown as an
efficient tool for spectrochemical analysis of a substrate.” L a s e r Phys.
6
(4) 1996 806 -
810.
[17] C.G. Gill, A.W. Garrett, P.H. Hemberger, N.S. Nogar, “Resonant Laser Ablation as
a selective metal ion source for gas-phase ion molecule reactions.” J. Am. Soc. Mass
Spectrom. 7 (7), 1996, 664 - 667.
[18] R.F. Haglund, “Microscopic and mesoscopic aspects of laser-induced desorption
and ablation.” Appl.
Surf.
Sci. 96-98, 1996, 1 - 13.
[19] R.F. Haglund, “Mechanisms of Laser-Induced Desorption and Ablation” from
Laser Ablation and Desorption. J.C. Miller and R.F. Haglund (Eds.) Experimental
Methods in the Physical Sciences v 30. Academic Press, Boston, 1998.
[20] V. Philipsen, J. Bastiaansen, P. Lievens, E. Vandeweert, R.E. Silverans, “Resonant
electron transfer in the emission of ion-beam sputtered metal atoms studied by double
resonance laser ionization.” Vacuum 56 (4), 2000, 269 - 274.
[21] R. Srinivasan, B. Braren, “Ultraviolet laser ablation of organic polymers.” Chem.
Rev. 89 (6 ), 1989 1303- 1316.
[22] M. Hemmerlin, J-M. Mermet, “Determination of elements in polymers by laser
ablation inductively coupled plasma atomic emission spectrometry: effect of the laser
beam wavelength, energy and masking on the ablation threshold and efficiency.”
Spectrochim. Acta, 5 IB (6 ), 1996, 579 - 589.
231
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
[23] G. Koren, A. Gupta, R.J. Baseman, M.I. Lutwyche, R.B. Laibowitz, “Laser
wavelength dependent properties of yttrium barium copper oxide (YBa2 Cu3 0 7 -8 ) thin
films deposited by laser ablation.” Appl. Phys. Lett. 55 (23), 1989, 2450 - 2452.
[24] S.A. Reid, W. Ho, F.J. Lamelas, “Pulsed Laser Ablation of Sn and SnC>2 Targets:
Neutral Composition, Energetics, and Wavelength Dependence.” J. Phys. Chem. B 104
(22), 2000, 5324-5330.
[25] H. Schittenhelm, G. Callies, A. Straub, P Berger, H. Hugel, “Measurements of
wavelength-dependent transmission in excimer laser-induced plasma plumes and their
interpretation.”
J. Phys. D 31(4), 1998, 4 1 8 - 427.
[26] G. Krier, F. R. Verdun, J. F. Muller, “Coupling a tunable dye laser to a LAMMA
500 microprobe.” Fres. Z. Anal. Chem. 322 (4), 1985, 379 - 382.
[27] J.P. Young, G.S. Hurst, S.D. Kramer, and M.G. Payne “Resonance Ionization
Spectroscopy.” Anal. Chem. 51 (11), 1979, 1050A - 1058A.
[28] C. J. McLean, J. H. Marsh, A. P. Land, A. Clark, R. Jennings, K. W. D.
Ledingham, P. T. McCombes, A. Marshall, R. P. Singhal, M. Towrie, “Resonant laser
ablation (RLA)” Int. J. Mass Spectrom. Ion. Proc. 96 (1), 1996, R1 - R7.
[29] L. Wang, K. W. D. Ledingham, C. J. McLean, R. P. Singhal, “Laser-induced
collisional processes in resonant laser ablation of gallium arsenide.” Appl. Phys. B 54
(1), 1992,71-75.
[30] J. K. Gibson, “Resonant Laser Ablation of Lanthanides: Eu and Lu Resonances in
the 450 - 470 nm Region.” Anal. Chem. 69 (2), 1997, 1 1 1 - 117.
232
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
[31] G. C. Eiden, N. S. Nogar, ‘The two-photon spectrum of iron and silicon detected
by resonant laser ablation.” Chem.
P h ys.
Lett. 226 (5 - 6), 1994, 509 - 516.
[32] P. Wei-xian, R. Deas, K.W.D. Ledingham, R.P. Singhal, “Trace element analysis
in Compound Semiconductor Material by Resonant Laser Ablation/Ionization Mass
Spectrometery.” Chin. Res. Chin. Uni. 12 (3), 1996, 240 - 244.
[33] G. Rothschopf, J. Zoller, R. Lewis, C. Grant, R. Schur, R. Estler “Electronic state
detection/partitioning of atomic nickel during resonant laser ablation.” Inter. J. Mass
Spectrom. Ion Proc. 151 (2/3), 1995, 167 - 174.
[34] S.T. Dai, J. Lin, W. Lu, R.T. An, L.L. Tai, D.Y. Chen. “Resonant laser ablation of
copper and it’s microanalysis.” Appl. Phys. A 69 (Suppl.), 1999, S167 - S169.
[35] J.K. Gibson, “Resonant Laser Ablation Mass Spectrometry of Barium Compounds:
Enhanced Ionization at the 455.4 nm Ba+ Absorption.” Rap. Comm. Mass Spectrom. 10
(3), 1996, 256 - 262.
[36] K. Watanabe, T. Iguchi “Modeling of vaporization processes of resonant laser
ablation.” Appl. Phys. A 69 (Suppl.), 1999, S845 - S848.
[37] C.G. Gill, A.W. Garrett, P.H. Hemberger, N.S. Nogar, “Selective laser
ablation/ionization for ion trap mass spectrometry: resonant laser ablation.”
Spectrochim. Acta B 51B (8), 1996, 851 - 862.
[38] T Gotz, M. Bergt, W. Hoheisel, F. Trager, M. Stuke, “Laser ablation of metals: the
transition from non-thermal processes to thermal evaporation.” Appl. Surf. Sci. 96 - 98,
1996, 2 80-286.
233
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
[39] W.P. Hess, A.G. Joly, D.P. Gerrity, K.M. Beck, P.V. Sushko, A.L. Shluger,
“Selective laser desorption of ionic surfaces: Resonant surface excitation of KBr.” J.
Chem. Phys. 115 (20), 2001, 9463 - 9472.
[40] R. H. Ritchie, “Surface plasmons in solids.” Surf.
ScL,
34 (1) 1973 1 - 19.
[41] T. Gotz, M. Vollmer, F. Trager, “Desorption of metal atoms with laser light of
different polarization.” Appl. Phys. A, A57 (1), 1993, 101 — 104.
[42] W. Hoheisel, K. Jungmann, M. Vollmer, R. Weidenauer, F. Trager, “Desorption
Stimulated by Laser-Induced Surface-Plasmon Excitation” P h y s. Rev. Lett. 60 (16),
1988, 1649 - 1652.
[43] M.J. Shea, R. N. Compton “Laser Ejection of Ag+ Ions from a Roughened Silver
Surface: Role of the Surface Plasmon.” Laser Ablation —Mechanisms and
Applications, Oak Ridge, Tennessee, USA, Springer-Verlag. 1991.
[44] X. Y. Zhu, “Surface photochemistry.” Ann. Rev. Phys.
C h e m .,
45, 1994, 113 - 144.
[45] D. Menzel, R. Gomer, “Desorption from metal surfaces by low-energy electrons.”
J. Chem. Phys. 41 (11), 1964, 3311 - 3328.
[46] P.A. Redhead, “Interaction of slow electrons with chemisorbed oxygen.” Can. J.
Phys. 42, 1965, 88 6 -9 0 5 .
[47] M.L. Knotek and P.J. Feibelman, “Ion desorption by core-hole Auger decay.”
Phys. Rev. Lett. 40 (14), 1978, 964 - 967.
[48] J.W. Gadzuk. from Desorption Induced by Electronic Transitions: DIET IV. G.
Betz, P. Varga (Eds.) Springer Series in Surface Sciences 19. Springer Verlag.
234
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
Proceedings of the Fourth Intematinal Workshop, Gloggnitz, Austria, October 2 - 4 ,
1989. New York. 2 - 1 1 .
[49] P.R. Antoniewicz, “Model for electron- and photon-stimulated desorption.” P h ys.
Rev. B 21 (9), 1980, 3811 - 3815.
[50] P. Wurz, J. Samthein, W. Husinsky, G. Betz, “Different Processes for Desorption
of Ground- and Excited-State Atoms Under Electron Bombardment of Alkali-Halides.”
Desorption Induced by Electronic Transitions (DIET IV) —Proceedings of the Fourth
International Workshop, Gloggnitz, Austria, Springer-Verlag, 1989.
[51] L. J. Radziemski, T. R. Loree, D. A. Cremers, N. M. Hoffman, ‘Time-resolved
laser-induced breakdown spectrometry of aerosols.” Anal. Chem., 55 (8), 1983 1246 1252.
[52] V. Majidi, “Laser-induced plasmas. A versatile tool for chemical analysis.”
Spectrosc. 8 (3), 1993, 16 - 17.
[53] I. B. Gomushkin, J. E. Kim, B. W. Smith, S. A. Baker, J. D. Winefordner,
“Determination of cobalt in soil, steel, and graphite using excited-state laser
fluorescence induced in a laser spark.” Appl. Spectrosc. 51 (7), 1997, 1055 - 1059.
[54] .A. F. Molisch, B. P. Oehry, Radiation Trapping in Atomic Vapours, Clarendon
Press, Oxford, 1998.
[55] T.B. Lucatorto, T.J. Mcllrath, “Efficient Laser Production of a Na+ Ground-State
Plasma Column: Absorption Spectroscopy and Photoionization Measurement of Na+.”
Phys. Rev. Lett. 37 (7), 1976, 428 - 431.
235
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
[56] T.J. Mcllrath, T.M. Lucatorto “Laser Excitation and Ionization in a Dense Li
Vapor: Observation of the Even Parity, Core-Excited Autoionizing States.” Phys.
R e v.
Lett. 38 (24), 1977, 1390-1393.
[57] D. Riley, L. Doyle, R. Al-Wazzan, ‘Trapping of radiation in a laser ablated
YBa2Cu30 7 plume” J. Appl.
P h y s.
79 (9), 1996, 7223 - 7226.
[58] J. Chorazy, T. Stacewicz, “Investigation of radiation trapping in atomic vapour
excited by weak resonant laser pulses” Acta Phys. Pol. A 96 (3-4), 1999, 383 - 390.
[59] IS . Borthwick, K.W.D. Ledingham, R.P. Singhal, “Resonant laser ablation - a
novel surface analysis technique.” Spectrochim. Acta B 47B (11), 1992, 1259 - 1265.
[60] T.M. Allen, C.H. Smith, P.B. Kelly, J.E. Anderson, G.C. Eiden, N.S. Nogar,
“Analysis of Solids, Films and Multilayers by Resonant Laser Ablation.” Proc SPIE.
2385, 1995, 39 - 50.
[61] J. Uebbing, A. Ciocan, K. Niemax, “Optical emission spectrometry of a microwave
induced plasma used with laser ablation of solid samples.” Spectrochim. Acta B 47B
(5), 1992, 601 - 610.
[62] C. I. M. Beenakker, “A cavity for microwave-induced plasmas operated in helium
and argon at atmospheric pressure.” Spectrochim. Acta
B
3 IB (8-9), 1976, 483 - 486.
[63] K.X Yang, Ph.D. Dissertation, University of Connecticut, May 1998.
[64] X. Hou, Ph.D. Dissertation, University of Connecticut, May 1999.
[65] J. X. Zhou, X. Hou, S. J. Tsai, K. X. Yang, R. G. Michel, “Characterization of a
Tunable Optical Parametric Oscillator Laser System for Multielement Flame Laser
236
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
Excited Atomic Fluorescence Spectrometry of Cobalt, Copper, Lead, Manganese, and
Thallium in Buffalo River Sediment.” Anal. Chem. 69 (3),1997,490 - 499.
[66] M.D. Seltzer, Ph.D.
D is s e r ta tio n ,
University of Connecticut, 1986. 46 - 51.
[67] Experimental Transition Probabilities for Spectral Lines of Seventy Elements, U.S.
Department of Commerce, National Bureau of Standards, NBS Monograph 53, C.H.
Corliss and W.R. Bozman, July 20, 1962.
[68] L. Wang, I.S. Borthwick, R. Jennings, P.T. McCombes, K.W.D. Ledingham, R.P.
Singhal, C.J. McLean, “Observations and Analysis of Resonant Laser Ablation of
GaAs.” Appl. Phys. B 53 (1), 1991, 34 - 38.
[69] K.X. Yang, J.X. Zhou, X. Hou, and R.G. Michel, “Excimer Laser Ablation for
Microwave-Induced Plasma Emission Spectrometry.” J. Kor. Phys. Soc. 35 (3), 1999,
133 - 142.
[70] J. Heidberg, W. Claas, H.C. Langowski, M. Suhren, M. Warskulat, “Resonant
Desorption from Clean and Carbon-Covered Alkali Halide Surfaces: State-Specific
Detection by REMPI.” in “Desorption Induced by Electronic Transitions, DIET IV”,
Edited by G. Betz and P. Varga, Springer-Verlag, Berlin Heidelberg 1990.
[71] J. Heidberg, C. Langowski, G. Neubauer, “Laser-induced resonant desorption and
ablation on ultra-thin carbon films.” Surf. Sci., 189 - 190, 1987, 946 - 953.
[72] X. Hou, unpublished data collected in this laboratory
[73] S.S. Mao, X. Mao, R. Greif, R.E. Russo, “Initiation of an early-stage plasma during
picosecond laser ablation.” Appl. Phys. Lett. 11 (16), 2000, 2464 - 2466.
237
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
[74] V.S. Burakov, A.F. Bokhonov, M.I. Nedel’ko, N.V. Tarasenko, “Near-threshold
laser-induced sputtering of aluminum surface by UV and IR radiation.” Appl. Surf. Sci.
1 3 8 - 139, 1999, 3 50-353.
[75] Katsumoto, Y.; Jiang, J.; Berry, R.J.; Ozaki, Y. Modem Pretreatment Methods in
NIR Spectroscopy. Near Infrared Anal. 2001, 2, 29 - 36.
[76] Hruschka, W.R. Spectral Reconstruction. In Handbook o f near- infrared analysis;
Burns, D.A., Ciurczak, E.W., Eds.; Marcel Dekker, Inc.: New York, 1992.
[77] M. Blanco, J. Coello, H. Iturriaga, S. Maspoch, C. de la Pezuela, C. Russo,
“Control analysis of a pharmaceutical preparation by near-infrared reflectance
spectroscopy. A comparative study of a spinning module and fiber optic probe.” Anal.
Chim. Acta 298 (2), 1994, 183 - 191.
[78] Mark, H. in Handbook o f Near Infrared Analysis, ed. D.A. Bumsand, E. Ciurczac,
Marcel Dekker, New York, 1992, Ch. 13.
[79] M. Blanco, M.A. Romero, “Near-infrared libraries in the pharmaceutical industry:
a solution for identity confirmation.” Analyst, 126 (12), 2001, 2212 - 2217.
[80] M.P. Derde, D.L. Massart, “Supervised pattern recognition: the ideal method?”
Anal. Chim. Acta 191 (1), 1986, 1 - 16.
[81] M.A. Sharaf, D.L. Illman, B.R. Kowalski, Chemometrics, Wiley, New York, 1986.
[82] E. Bouveresse, C. Hartmann, D.L. Massart, I.R. Last, K.A. Prebble,
“Standardization of Near-Infrared Spectrometric Instruments.” Anal. Chem. 68 (6),
1996, 982 - 990.
238
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
[83] S. Fendel, B. Schrader, “Investigation of Skin and Skin Lesions by NIR-FT-Raman
Spectroscopy.” Fres. J. Anal Chem. 360 (5), 1998, 609 - 613.
[84] L.M. McIntosh, R. Summers, M. Jackson, H.H. Mantsch, J.R. Mansfield, M.
Howlett, A.N. Crowson, J.W.P. Toole, ‘Towards Non-Invasive Screening of Skin
Lesions by Near-Infrared Spectroscopy.” J. Inves. Derm. 116 (1), 2001, 175 - 181.
[85] Y-A. Woo, J-W. Ahn, I-K. Chun, H-J. Kum, “Development of a Method for the
Determination of Human Skin Moisture using a Portable Near-Infrared System.” Anal.
Chem. 73 (20), 2001, 4964 - 4971.
[86] K. Martin, “In Vivo Measurements of Water in Skin by Near-Infrared
Reflectance.” Appl. Spec. 52 (7), 1998, 1001 - 1007.
[87] T. Huth-Fehre, R. Feldhoff, F. Kowol, H. Freitage, S. Kuttler, B. Lohwasser, M.
Oleimeulen, “Remote Sensor Systems for the Automated Identification of Plastics.” J.
Near Infrared Spectrosc. 6 (1-4), 1998, A7 - A ll.
[88] T. Rohe, W. Becker, S. Kolle, N. Eisenreich, P. Eyerer, “Near Infrared (NIR)
Spectroscopy for In-line Monitoring of Polymer Extrusion Processes.” Talanta 50 (2),
1999, 283 - 290.
[89] P.J. de Groot, G.J. Postma, W.J. Meissen, L.M.C. Buydens, “Validation of Remote,
On-line, Near-Infrared Measurements for the Classification of Demolition Waste.”
Anal. Chim. Acta 453 (1), 2002, 117 - 124.
[90] P.J. de Groot, G.J. Postma, W.J. Meissen, L.M.C. Buydens, “Influence of
Wavelength Selection and Data Preprocessing on Near-Infrared-Based Classification of
Demolition Waste.” Appl. Spec. 55 (2), 2001, 173 - 181.
239
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
[91] W.H.A.M. van den Broek, D. Wienke, W.J. Meissen, L.M.C. Buydens, “Plastic
material identification with spectroscopic near infrared imaging and artificial neural
networks.” Anal. Chim. Acta 361, 1998, 161 - 176.
[92] W.H.A.M. van den Broek, D. Wienke, W.J. Meissen, R. Feldhoff, T. Huth-Fehre,
T. Kantimm, L.M.C. Buydens, “Application of a Spectroscopic Infrared Focal Plane
Array Sensor for On-Line Identification of Plastic Waste.” Appl. Spec. 51 (6), 1997, 856
-8 6 5 .
[93] R. Feldhoff, D. Wienke, K. Cammann, H. Fuchs, “On-Line Post Consumer
Package Identification by NIR Spectroscopy Combined with a FuzzyARTMAP
Classifier in an Industrial Environment.” Appl. Spec. 51 (3), 1997, 362 - 368.
[94] S.J. Choquette, S.N. Chelser, D.L. Duewer, “Identification and Quantitation of
Oxygenates in Gasoline Ampules Using Fourier Transform Near-Infrared and Fourier
Transform Raman Spectroscopy.” Anal. Chem. 68 (20), 1996, 3525 - 3533.
[95] J. Buerck, S. Roth, K. Kraemer, M. Scholz, N. Klaas, “Application of a Fiber-Optic
NIR-EFA Sensor System for In Situ Monitoring of Aromatic Hydrocarbons in
Contaminated Groundwater.” J. Haz. Mat. 83 (1-2), 2001, 1 1 -2 8 .
[96] H. Chung, H-J. Choi, M-S. Ku, “Rapid Identification of Petroleum Products by
Near-Infrared Spectroscopy.” Bull. Korean Chem. Soc. 20 (9), 1999, 1021 - 1025.
[97] S. Schmidt, “Determination of Moisture Content by Pulsed Nuclear Magnetic
Resonance Spectroscopy.” J. Adv. Exp. Med. Biol. 302, 1991, 599 - 613.
240
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
[98] C.V. Greensil, P.J. Wolfs, C.H. Spiegelman, K.B. Walsh, “Calibration Transfer
Between PDA-Based NIR Spectrometers in the NIR Assessment of Melon Soluble
Solids Content.” Appl. Spec. 55 (5), 2001, 647 - 653.
[99] B. Walczak, E. Bouversse, D.E. Massart, “Standardization of Near-Infrared
Spectra in the Wavelet Domain.” Chemom. Intell. Lab. Syst. 36 (1), 1997, 41 - 55.
[100] M. Tsuta, J. Sugiyama, Y. S agar a, “Near-Infrared Imaging Spectroscopy Based
on Sugar Absorption Bands for Melons.” J. Agric. Food Chem 50 (1), 2002, 48 - 52.
[101] J. Sugiyama, “Visualization of Sugar Content in the Flesh of a Melon by NearInfrared Imaging.” J. Agric. Food Chem. 47 (7), 1999, 2715 - 2718.
[102] P. Martinsen, P. Schaare, “Measuring Soluble Solid Distribution in Kiwifruit
Using Near-Infrared Imaging Spectroscopy.” Post-harvest Biol. Technol. 14 (3), 1998,
271 -2 8 1 .
[103] B. Steuer, H. Schulz, E. Lager, “Classification and Analysis of Citrus Oils by NIR
Spectroscopy.” Food Chem. 72 (1), 2001, 113 - 117.
[104] W. Li, P. Goovaerts, M. Meurens, “Quantitative Analysis of Individual Sugars
and Acids in Orange Juices by Near-Infrared Spectroscopy of Dry Extract.” J. Agric.
Food Chem. 44 (8), 1996, 2252 - 2259.
[105] W.B. Fischer, H.H. Eysel, O.F. Nielsen, J.E. Betie, “Corrections to the Baseline
Distortions in the OH-stretch Region of Aqueous Solutions.” Appl. Spec. 48 (1), 1994,
107-112.
241
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
[106] M. Meurens, “Analysis of Aqueous Liquids by Near Infrared Reflectance
Spectroscopy.” In International M e e tin g o f Chemical Engineering; Achema 82:
Frankfurt, Germany, 1982.
[107] R.G. Dambergs, A. Kambouris, I.L. Francis, M. Gishen, “Rapid Analysis of
Methanol in Grape-Derived Distillation Products using Near-Infrared Transmission
Spectroscopy.” J. Agric. Food Chem. 50 (11), 2002, 3079 - 3084.
[108] H.B. Ding, R.J. Xu, “Near-Infrared Spectroscopic Technique for Detection of
Beef Hamburger Adulteration.” J. Agric. Food Chem. 48 (6), 2000, 2193 - 2198.
[109] I. Gonzalez-Martm, C. Gonzalez-Perez, J. Hemandez-Mendez, N. AlvarezGarcfa, J-L.H. Andaluz, “On-line Non-Destructive Determination of Proteins and
Infiltrated Fat in Iberian Pork Loin by Near Infared Spectrometry with a Remote
Reflectance Fiber Optic Probe.” Anal. Chim. Acta 453 (2), 2002, 281 - 288.
[110] M-F. Laporte, P. Paquin, “Near-Infrared Analysis of Fat, Protein, and Casein in
Cow’s Milk.” J. Agric. Food Chem 47 (7), 1999, 2600 - 2605.
[111] S. Sasic, Y. Ozaki, “Short-Wave Near-Infrared Spectroscopy of Biological Fluids.
1. Quantitative Analysis of Fat, Protein, and Lactose in Raw Milk by Partial LeastSquares Regression and Band Assignment.” Anal. Chem. 73 (1), 2001, 6 4 -7 1 .
[112] H. Sato, S. Wada, M. Ling, H. Tashiro, “Noninvasive Measurement of
Oxygenation of Hemoglobin by Direct Transmission of Near-Infrared Energy (7001000 nm) from an Electronically Tuned Ti: Sapphire Laser Driven by a Dual RadioFrequency Driving Method.” Appl. Spec. 54 (8), 2000, 1163 - 1167.
242
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
[113] P.Y. Qiu, H.B. Ding, Y.K. Tang, R.J. Xu, “Determination of Chemical
Composition of Commercial Honey by Near-Infrared Spectroscopy.” J. Agric. Food
Chem. 47 (7), 1999, 2760 - 2765.
[114] H. Buning-Pfaue, S. Kehraus, “Application of Near Infrared Spectroscopy (NIRS)
to the Analysis of Frying Fats.” Eur. J. Lipid Sci. Technol. 103 (8-9), 2001, 793 - 797.
[115] S.B. Engelsen, “Exporative Spectrometric Evaluations of Frying Oil
Deterioration.” J. Am. Oil Chem. Soc. 74 (12), 1997, 1495 - 1508.
[116] H. Li, F.R. van der Voort, J. Sedman, A. A. Ismail, “Rapid Determination of cis
and trans Content, Iodine Value, and Saponification Number of Edible Oils by Fourier
Transform Near Infrared Spectroscopy.” J. Am. Oil Chem. Soc. 76 (12), 1999, 491 497.
[117] A.J. Boot, A.J. Speek, “Determination of the Sum of Dimer and Polymer
Triglycerides and of Acid Value of Used Frying Fats and Oils by Near Infrared
Reflectance Spectroscopy.” J. AO AC Int. 11 (5), 1994, 1184 - 1189.
[118] A.A. Ismail, F.R. van de Voort, G. Emo, I. Sedman, “Rapid Quantitative
Determination of Free Fatty Acids in Fats and Oils by Fourier Transform Infrared
Spectroscopy.” J. AM. Oil Chem. Soc. 70 (4), 1993, 335 - 341.
[119] H.H.A. El-Rafey, J. Peredy, K. Kaffka, B. Naday, A. Balogh, “Possibilities of
Application of NIR Technique in Analysis of Oilseeds and Their Derivaties Examination of Used Frying Oils.” Fat Sci. Technol. 90 (5), 1988, 175 - 179.
243
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
[120] I. Dong, K. Ma, F.R. van den Voort, A. A. Ismail, “Stoichiometric Determination
of Hydroperoxides in Oils by Fourier Transform Near-Infrared Spectroscopy.” J. AOAC
Int. 80 (2), 1997, 345 - 352.
[121] H. Biining-Pfaue, R. Hartmann, J. Harder, S. Kehraus, C. Urban, “NIRSpectrometric Analysis of Food, Methodical Development and Achievable Performance
Values.” Fres. J. Anal. Chem. 360 (7-8), 1998, 832 - 835.
[122] K. Van Scoik, Theory o f Mixing Solids, Abbott Laboratories, Arden House
Conference, 1992.
[123] R.E. Schirmer, Modem Methods of Pharmaceutical Analysis, 2nd ed.; CRC
[124] S.S. Sekulik, S.S.; H.W. Ward, D.R. Brannegan, E.D. Stanley, C.L. Evans, S.T.
Sciavolino, P. A. Hailey, P.K. Aldridge, “On-Line Monitoring of Powder Blend
Homogeneity by Near-Infrared Spectroscopy.” Anal. Chem. 68 (3), 1996, 509 - 513.
[125] M. Blanco, J. Coello, A. Eustaquio, H. Iturriaga, S. Maspoch, “Analytical Control
of Pharmaceutical Production Steps by Near Infrared Reflectance Spectroscopy.” Anal.
Chim. Acta 392 (2-3), 1999, 237 - 246.
[126] S.H.F. Scafi, C. Pasquini, “Identification of Counterfeit Drugs using NearInfrared Spectroscopy.” Analyst 126 (12), 2001, 2218 - 2224.
[127] I. Rager, G. Roos, P.C. Schmidt, K-A. Kovar, “Rapid Quantification of
Constituents in St. John’s Wort Extracts by NIR Spectroscopy.” J. Pharm. Biomed.
Anal. 28 (3-4), 2002, 439 - 446.
244
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
[128] M. Laasonen, T. Harmia-Pulkkinen, C.L. Simard, E. Michiels, M. Rasanen, H.
Vuorela, “Fast Identification of Echinacea purpurea Dried Roots Using Near-Infrared
Spectroscopy.” Anal. Chem. 74 (11), 2002, 2493 - 2499.
[129] R. Bauer, H. Wagner, “New results in the analysis of Echinacea roots.” Sci.
Pharm. 55, 1987, 159-161.
[130] J. Gottfries, H. Depui, M. Fransson, M. Jongeneelen, M. Josefson, F.W.
Langkilde, D.T. Witte, “Vibrational Spectrometry for the Assessment of Active
Substance in Metoprolol Tablets: A Comparison Between Transmission and Diffuse
Reflectance Near-Infrared Spectrometry.” J. Pharm. Biomed. Anal. 14 (11), 1996, 1495
- 1503.
[131] A. Sparen, M. Malm, M. Josefson, S. Folestad, J. Johansson, “Light Leakage
Effects with Different Sample Holder Geometries in Qualitative Near-Infrared
Transission Spectroscopy of Pharmaceutical Tablets.” Appl. Spec. 56 (5), 2002, 586 592.
[132] B.K. Lavine, C.E. Davidson, A.J. Moores, “Genetic Algorithms for Spectral
Pattern Recognition.” Vibr. Spectrosc. 28 (1), 2002, 83 - 95.
[133] C.B. Lucasius, G. Kateman, “Understanding and Using Genetic Algorithms. Part
1. Concepts, Properties and Context.” Chemom. Intell. Lab. Syst. 19 (3), 1993, 1 - 33.
[134] C.B. Lucasius, G. Kateman. “Understanding and using genetic algorithms Part 2.
Representation, configuration and hybridization.” Chemom. Intell. Lab. Syst. 25 (2),
1994, 99 - 145.
245
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
[135] D.B. Hibbert, “Genetic Algorithms in Chemistry.” Chemom. Intell. Lab. Syst. 19
(3), 1993,277-293.
[136] D.E. Goldberg, Genetic Algorithms in Search, Optimization, and Machine
Learning, Addison-Wesley: Reading, MA, 1989.
[137] C.B. Lucasius, L.M. Beckers, G. Kateman, “Genetic Algorithms in Wavelength
Selection: A Comparative Study.” Anal. Chim. Acta 286 (2), 1994, 135 - 153.
[138] P. Vankeerberghen, J. Smeyers-Verbeke, R. Leardi, C.L. Karr, D.L. Massart,
“Robust Regression and Outlier Detection for Non-Linear Models using Genetic
Algorithms.” Chemom. Intell. Lab. Syst. 28 (1), 1995, 73 - 87.
[139] G.W. Small, M.A. Arnold, L.A. Marquardt, “Strategies for Coupling Digital
Filtering with Partial Least-Squares Regression: Application to the Determination of
Glucose in Plasma by Fourier Transform Near-Infrared Spectroscopy.” Anal. Chem. 65
(22), 1993, 3279 - 3289.
[140] R.E. Shaffer, G.W. Small, M.A. Arnold, “Genetic Algorithm-Based Protocol for
Coupling Digital Filtering and Partial Least-Squares Regression: Application to the
Near-Infrared Analysis of Glucose in Biological Matrices.” Anal. Chem. 68 (15), 1996,
2663 - 2675.
[141] F. Torella, R. Cowley, M.S. Thmiley, C.N. McCollum, “Monitoring Blood Loss
with Near Infrared Spectroscopy.” Comp. Biochem. Phys. A. 132 (1), 2002, 199 - 203.
[142] T. Kusaka, K. Isobe, K. Nagano, K. Okubo, S. Yasuda, M. Kondo, S. Itoh, K.
Hirao, S. Onishi, “Quantification of Cerebral Oxygenation by Full-Spectrum Near-
246
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
Infrared Spectroscopy using a Two-Point Method.” Comp. Biochem. Phys. A. 132 (1),
2002, 121 - 132.
[143] P. van der Zee, M. Cope, S.R. Arridge, M. Essenpreis, L.A. Potter, A.D. Edwards,
J.S. Wyatt, D.C. McCormick, S.C. Roth, E.O. Reynolds, “Experimentally Measured
Optical Pathlengths for the Adult Head, Calf and Forearm and the Head of the Newborn
Infant as a Function of Inter Optode Spacing.” Adv. Exp. Med. Biol. 316, 1992, 143 153.
[144] J.S. Wyatt, M. Cope, D.T. Delpy, P. van der Zee, S. Arridge, A.D. Edwards, E.O.
Reynolds, “Measurement of Optical Path Length for Cerebral Near-Infrared
Spectroscopy in Newborn Infants.” D e v . Neurosci. 12 (2), 1990, 140 - 144.
[145] C.E. Cooper, C. E.; Elwell, J.H. Meek, S.J. Matcher, J.S. Wyatt, M. Cope, D.T.
Delpy, ‘The Noninvasive Measurement of Absolute Cerebral Deoxyhemoglobin
Concentration and Mean Optical Path Length in the Neonatal Brain by Second
Derivative Near Infrared Spectroscopy.” Pediatr. Res. 39 (1), 1996, 32 - 38.
[146] Laboratory-constructed prototype PbS array-based spectrometer. Special thanks
to CVI Laser Corporation (Putnam, CT) for its loan of the SM-301 spectrometer.
[147] SM32Pro v2.1 True 32-bit software provided by CVI Laser Corporation (Putnam,
CT).
[148] R. Marks, “Histochemical Applications of Skin Surface Biopsy.” Brit. J. Dermat.
86(1), 1972,20-26.
[149] H. Zhai, HI. Maibach, “Effect of barrier creams: human skin in vivo.” Contact
Dermat. 35 (2), 1996, 92 - 96.
247
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
[150] Q. Sun, M. Tran, B.W. Smith, J.D. Winefordner, “Zinc analysis in human skin by
laser induced-breakdown spectroscopy.” Talanta 52 (2), 2000, 293 - 300.
[151] R.E. Russo, X.L. Mao, O.V. Borisov, H. Liu, “Influence of wavelength on
fractionation in laser ablation ICP-MS” J. Anal. At. Spectrom. 15 (9), 2000, 1115 1120.
[152] A.M. Leach, G.M. Hieftje, “Factors Affecting the Production of Fast Transient
Signals in Single Shot Laser Ablation Inductively Coupled Plasma Mass Spectrometry.”
Appl. Spec. 56 (1), 2002, 62 - 69.
[153] J.X. Zhou, Ph.D. Dissertation, University of Connecticut, May 1998.
[154] P. Arrowsmith, “Laser ablation of solids for elemental analysis by inductively
coupled plasma mass spectrometry.” Anal. Chem. 59 (10) 1987, 1437 - 1444.
[155] A. A. van Heuzen, “Analysis of solids by laser ablation-inductively coupled
plasma-mass spectrometry (LA-ICP-MS). I. Matching with a glass matrix.”
Spectrochim. Acta B 46B (14), 1991, 1803 - 1817.
[156] W.T. Perkins, R. Fuge, N.J. Pearce, “Quantitative determination of trace elements
in carbonates using laser ablation inductively coupled plasma mass spectrometry.” J.
Anal. At. Spectrom. 6 (6), 1991, 445 - 449.
[157] A.A. van Heuzen, J.B. Morsink, “Analysis of solids by laser ablation-inductively
coupled plasma-mass spectrometry (LA-ICP-MS). II. Matching with a pressed pellet.”
Spectrochim. Acta, B 46B (14), 1991, 1819 - 1828.
248
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
[158] D. Gunther, H. Cousin, B. Magyar, I. Leopold, “Calibration Studies on Dried
Aerosols for Laser Ablation-inductively Coupled Plasma Mass Spectrometry.” J. Anal.
At. Spectrom. 12 (2), 1997, 165 - 170.
[159] L. Moenke-Blankenburg, T. Schumann, D. Gunther, H-M. Kuss, M. Paul,
“Quantitative analysis of glass using inductively coupled plasma atomic emission and
mass spectrometry, laser micro-analysis inductively coupled plasma atomic emission
spectrometry and laser ablation inductively coupled plasma mass spectrometry.” J.
Anal. At. Spectrom, 7 (2), 1992, 251 - 254.
[160] L. Moenke-B lankenburg, M. Gackle, D. Gunther, J. Kammel, Processes o f Laser
Ablation and Vapour Transport to the IC P . ed. K.E. Jarvis, Royal Society of Chemistry:
University of Surrey, 1990, pp. 1 - 17.
[161] W.J. Balsanek, J.D. Venable, J.A. Holcombe, “Generation of a square wave
inductively coupled plasma scanning mass spectrometry using electrothermal
vaporization sample introduction.” J. Anal. At. Spectrom. 18 (1), 2003, 59 - 64.
[162] J.X. Zhou, X. Hou, K.X. Yang, S-J. J. Tsai, R.G. Michel, “Lasers Based on
Optical Parametric Devices: Wavelength Tunability Empowers Laser-Based
Techniques in the UV, Visible, and Near-IR”, Appl. Spec. 52 (5), 1998 176A-189A.
[163] C.L. Tang, L.K. Cheng, Fundamentals of optical parametric processes and
oscillators. Amsterdam : Harwood Academic Publishers, 1995
[164] Quanta-Ray MOPO-730 Optical Parametric Oscillator Instruction Manual.
Spectra Physics Lasers. Draft 12-12-94.
249
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
[165] K. Bennett, R.L. Byer, “Computer-controllable wedged-plate optical variable
attenuator.” Appl. Opt. 19 (14), 1980, 2408 -2412.
250
R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission.
Документ
Категория
Без категории
Просмотров
0
Размер файла
16 420 Кб
Теги
sdewsdweddes
1/--страниц
Пожаловаться на содержимое документа