High intensity laser deposition of diamond-like carbon thin films


Material Information

High intensity laser deposition of diamond-like carbon thin films
Physical Description:
xiii, 140 leaves : ill. ; 29 cm.
Qian, Fan
Publication Date:


Subjects / Keywords:
Materials Science and Engineering thesis, Ph. D
Dissertations, Academic -- Materials Science and Engineering -- UF
government publication (state, provincial, terriorial, dependent)   ( marcgt )
bibliography   ( marcgt )
non-fiction   ( marcgt )


Thesis (Ph. D.)--University of Florida, 1996.
Includes bibliographical references (leaves 133-139).
Statement of Responsibility:
by Fan Qian.
General Note:
General Note:

Record Information

Source Institution:
University of Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
aleph - 002118307
oclc - 35828393
notis - AKV8036
System ID:

Full Text








To Kak, for her love and support.


I wish to acknowledge the help from various people who made the completion of

this study possible. First, I would like to thank my advisors, Dr. Rajiv Singh at the

University of Florida and Dr. Peter Pronko at the University of Michigan, for giving me

this unique opportunity to carry out the experiments as part of a collaborative research

project between these two schools. Their constant encouragement and insightful guidance

through out this intriguing yet sometimes strenuous endeavor are much appreciated. I

owe my special thanks to Subrata Dutta, a fellow student whom I have been working with

during the past two years. His help in setting up the experiments and his knowledge in

laser optics have proven to be indispensable for this study. Together we have struggled

through the oftentimes frustrating start-up period, battled with the customarily

unpredictable laser system, and enjoyed the delight of a successful run. I would like to

acknowledge the assistance I have received from some of the other students at the

University of Michigan, including Paul, Marc, Detao, Jonathan, Rob, Anchun, Allen and

Xinbin. Thanks also to my friends Don, Dongu, John, Rajan, Jenfon and Jakie at the

University of Florida, who have kept me in touch with the school I am officially associated


During the course of this research project, I have received assistance from people

beyond the scope of my daily contact. I would like to express my appreciation to Dr.

Clark from the Applied Physics Department at UM, Mr. Mansfield and Mr. Rollins from

the EMAL facility of UM, Dr. Weber from the Ford Research Laboratory, Dr. Brown

from Oak Rridge National Laboratory, and Drs. Davis, Kowalchyk, and Pelletier from the

Kaiser Optical Systems, for their help in some of the sample analyses.

Last but not least, I extend my thanks to the administrative staff I have been

associated with at the Center for Ultrafast Optical Science, University of Michigan, whose

courtesy and corporation made the days and months here more memorable. Marni, Fayne,

Autumn, Serina, Tammy, Martha, Julie and Judy are among them.


A CK N O W LED G M EN T S ......................................................................... .................. iii

LIST OF TABLES .................................................................... vii

LIST O F FIG U R E S.......................................... ................................................... viii

ABSTRACT ................................... ......... ................. ...................... xii


1 INTRODUCTION AND OVERVIEW ..................... ..........................1...
B background and M otivation................................................................ .................... 1
D issertation O utline .................... ................................................................... 4

2 LITER A TU R E R EV IE W ..................................................................... ............... 6
Diam ond-like Carbon (DLC) Thin Film s .............................................. ................ 6
Structure and Property ....................................... ......................... ................ 6
A applications ............................ ........ ........................................ ............... 11
D position M echanism s........................................ ................... 15
Laser D position of D LC ................................................................... ................ 18
Laser-Solid-Plasm a Interactions....................................................... .................. 33
N anosecond Laser Plasm a........................................................... .................. 39
Fem tosecond Laser Plasm a............................................................ ................ 52

3 EXPERIMENTAL CONDITION AND CONFIGURATION ............................... 60
Laser System ........................... ............................................... ............... 60
Thin Film D position System ............................................... ..... ................. 67
D LC Film D position Conditions ...................................................... .................. 73
Time-of-Flight (TOF) Measurement Set-up................................ 74

4 RESULTS AN D D ISCU SSION ................................................. ....... ................ 78
Tim e-of-Flight (TOF) M easurem ents ................................................. ................ 78
D L C F ilm P properties ............................................................................. ................ 95
Scanning Probe M icroscopy (SPM )............................................. ................. 95
Variable Angle Spectroscopic Ellipsometry (VASE)..................................... 99
Ram an Spectroscopy............. ............................... ................ ................ 112
Electron Energy Loss Spectroscopy (EELS) ........................... 125


5 CON CLU SION S ....... ............................................. ................................ 130

LIST O F REFEREN CES...................................... .......................... ................. 133

BIOGRAPHICAL SKETCH .......................................................... ................. 140


Table page

1 Properties of diamond, CVD diamond, graphite and DLC................................... 12

2 Properties of DLC films deposited by laser beam .............................. ................. 26

3 Kinetic energy of carbon particles as a function of laser wavelength...................... 30

4 Experimental conditions for Ti:Sapphire laser deposition of DLC films................. 73

5 Experimental conditions for Nd:YAG laser deposition of DLC films.....................74

6 Thickness and Tauc bandgap of Ti:Sapphire laser deposited DLC films .............. 108

7 Thickness and Tauc bandgap of Nd:YAG laser deposited DLC films ..................... 108

8 Fitted Raman parameters of Ti:Sapphire laser deposited DLC films ....................... 121

9 Fitted Raman parameters of Nd:YAG laser deposited DLC films ........................... 124


Figure page

Fig. 2-1 Lattice structure of diam ond........................................................ ................7...

Fig. 2-2 Lattice structure of graphite......................................................... ................ 7

Fig. 2-3 Schematic of laser-solid-plasma interactions............................................... 41

Fig. 2-4 Schematic profile of plasma density, pressure, and velocity gradients ..................
perpendicular to the target surface.................................................... ................. 43

Fig. 2-5 Ion concentration as a function of distance from the graphite target, at 5 x 1010
W /cm 2 ............................................................. ....................................................... 5 1

Fig. 2-6 Fraction of various carbon ions characterized by "frozen ionization" as a function
o f laser inten sity ............................................... ...... .................... ................. 5 1

Fig. 2-7 Differences between femtosecond and nanosecond laser induced plasmas......... 56

Fig. 3-1 Schematic of the CPA process ................................................................... 62

Fig. 3-2 Schematic of the CPA Ti:Sapphire laser system ....................... .................. 64

Fig. 3-3 Schematic of the high vacuum thin film deposition system ................................ 66

Fig. 3-4 Deposition arrangement for the Nd:YAG laser beam ........................................ 68

Fig. 3-5 Deposition arrangement for the Ti:Sapphire laser beam .................................... 72

Fig. 3-6 Schematic of the TOF experiment .................................................................... 75

Fig. 3-7 Layout of the field-free TOF drift tube..................................... 76

Fig. 4-la TOF spectrum plotted on the time scale at laser intensity of 2 x 109 W/cm2 .... 79

Fig. 4-lb TOF spectrum plotted on the velocity scale at laser intensity of 2 x 109 W/cm2
... ........................................................................................................................... 7 9

Fig. 4-2a TOF spectrum plotted on the time scale at laser intensity of 2 x 1010 W/cm2. .. 80

Fig. 4-2b TOF spectrum plotted on the velocity scale at laser intensity of 2 x 1010 W/cm2
............................................................................................................................... 8 0

Fig. 4-3a TOF spectrum plotted on the time scale at laser intensity of 6 x 10" W/cm2 ..81

Fig. 4-3b TOF spectrum plotted on the velocity scale at laser intensity of 6 x l011 W/cm2
................................................ .............. ...................................................... 8 1

Fig. 4-4 Most probable and maximum carbon ion velocity as a function of laser intensity
............................................................................................................................... 8 2

Fig. 4-5 Most probable and maximum carbon ion kinetic energy as a function of laser
in te n sity ................................................................................................................... 8 2

Fig. 4-6a TOF spectrum plotted on the time scale at laser intensity of 3 x 1014 W/cm2...85

Fig. 4-6b TOF spectrum plotted on the velocity scale at laser intensity of 3 x 1014 W/cm2
............. ..... ........................................................................................................... 8 5

Fig. 4-7a TOF spectrum plotted on the time scale at laser intensity of 1 x 1015 W/cm2 .. 86

Fig. 4-7b TOF spectrum plotted on the velocity scale at laser intensity of 1 x 101" W/cm2
...... ........................................................................................................................ 8 68 6

Fig. 4-8a TOF spectrum plotted on the time scale at laser intensity of 6 x 1015 W/cm2... 87

Fig. 4-8b TOF spectrum plotted on the velocity scale at laser intensity of 6 x 10" W/cm2
............................................................................................................................... 8 7

Fig. 4-9 Most probable and suprathermal carbon ion velocity as a function of laser
in te n sity ................................................................. ............................................... 8 8

Fig. 4-10 Most probable and suprathermal carbon ion kinetic energy as a function of laser
in te n sity ................................................................................................................... 8 8

Fig. 4-11 AFM image of DLC film deposited by Nd:YAG laser at 3 x 1010 W/cm2........ 96

Fig. 4-12 AFM image of DLC film deposited by Nd:YAG laser at 1 x 101 W/cm2 ........ 96

Fig. 4-13 AFM image of DLC film deposited by Nd:YAG laser at 6 x 10" W/cm2 ........ 97

Fig. 4-14 AFM image of DLC film deposited by Ti:Sapphire laser at 3 x 1014 W/cm2 .... 97

Fig. 4-15 AFM image of DLC film deposited by Ti:Sapphire laser at 1 x 1015 W/cm2 .... 98

Fig. 4-16 AFM image of DLC film deposited by Ti:Sapphire laser at 6 x 1015 W/cm2 .... 98

Fig. 4-17 Transmittance of DLC film deposited by Ti:Sapphire laser at 3 x 1014 W/cm2
............................................................................................................................. 1 0 2

Fig. 4-18 Transmittance of DLC film deposited by Ti:Sapphire laser at 6 x 1015 W/cm2
............................................................................................................................. 1 0210 2

Fig. 4-19 Experimental and fitted y and A for DLC film deposited at 3 x 1014 W/cm2.. 103

Fig. 4-20 Experimental and fitted y and A for DLC film deposited at 6 x 10" W/cm2.. 103

Fig. 4-21 Refractive indices n of DLC films deposited by Ti:Sapphire laser, as a function
of w avelength .................................... ............................................................... 105

Fig. 4-22 Extinction coefficients k of DLC films deposited by Ti:Sapphire laser, as a
function of wavelength ....... ..................................... .............................. 105

Fig. 4-23 Tauc plot of DLC films deposited by Ti:Sapphire laser.............................. 106

Fig. 4-24 Refractive indices n of DLC films deposited by Nd:YAG laser, as a function of
w avelength ................ ........ .................................................... ........ ........... 109

Fig. 4-25 Extinction coefficients k of DLC films deposited by Nd:YAG laser, as function
of w avelength........................................... ............................... ...... ................ 109

Fig. 4-26 Tauc plot of DLC films deposited by Nd:YAG laser.................................. 110

Fig. 4-27 Schematic of the Raman spectrometer....... ........................................ 114

Fig. 4-28 Raman spectrum of DLC film deposited by Ti:Sapphire laser at 6 x 1014 W/cm2
.......................................... ....... .......................... ................................ ................ 1 15

Fig. 4-29 Raman spectrum of DLC film deposited by Ti:Sapphire laser at 1 x 10" W/cm2
................................................................................................ ........................... 1 15

Fig. 4-30 Raman spectrum of DLC film deposited by Ti:Sapphire laser at 3 x 10" W/cm2
............................................................................................................................. 1 1 6

Fig. 4-31 Raman spectrum of DLC film deposited by Ti:Sapphire laser at 6 x 10" W/cm2
............. ............................................................................................................... 1 1 6

Fig. 4-32 Raman spectra of the graphite target and a CVD diamond film.................. 117

Fig. 4-33 Raman spectrum ofgraphitized DLC film deposited by Ti:Sapphire laser at
l x 10 W /cm 2 .............................. .............. ................................. .................. 117

Fig. 4-34 Raman spectrum of DLC film deposited by Ti:Sapphire laser on Si substrate at
1 x 10 W /cm 2................................................ ............................................... 118

Fig. 4-35 Raman spectra of DLC films deposited by Nd:YAG laser at various intensities
............................................................................................................................. 1 1 8

Fig. 4-36 K-shell edge EELS spectra of DLC films deposited by Ti:Sapphire laser at 3 x
1014 and 5 x 10"1 W/cm2, along with that of a CVD diamond and a graphite film.... 128

Abstract of Dissertation Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Doctor of Philosophy




August, 1996

Chairman: Rajiv K. Singh
Major Department: Materials Science and Engineering

We have deposited hydrogen-free diamond-like carbon (DLC) films at laser

intensity in the 1014 1015 W/cm2 range, with 100 femtosecond (FWHM) Ti: Sapphire laser

beams. The films were studied with scanning probe microscopy (SPM), variable angle

spectroscopic ellipsometry (VASE), Raman spectrometry, and electron energy loss

spectroscopy (EELS). DLC films with good scratch resistance, excellent chemical

inertness and high optical transparency in the infrared (IR) range were made at room

temperature. As the laser intensity was increased from 3 x 1014 to 6 x 1015 W/cm2, the

films showed an increased surface particle density, a decreased optical transparency (85%

-+ 60%) and Tauc bandgap (1.4 -+ 0.8 eV), as well as a lower sp3 content (60% -- 50%).

The carbon ion kinetic energy was measured as a function of laser intensity with a Faraday

ion collector. The time-of-flight (TOF) spectra exhibited a double-peak distribution, with

a high energy suprathermal ion peak preceding a slower thermal component. The most

probable ion kinetic energy demonstrated an 10.5 dependence, when the laser intensity was

varied from 3 x 1014 to 6 x 1015 W/cm2. The ion kinetic energy increased from 300 to

2000 eV within this range. The kinetic energy of suprathermal ions increased from 3 keV

to over 20 keV and showed an 10.33 dependence. These ions are speculated to have

originated from an electrostatic acceleration field established by the suprathermal

electrons, due to resonant absorption of the intense laser beams.

In comparison, we have deposited DLC films with conventional Nd:YAG laser

beams (7 ns @ FWHM). These films also became more absorbing with decreased Tauc

bandgaps (2.4 -+ 0.9 eV) and a deteriorating surface topography, as a function of

increasing laser intensity in the 2 x 1010 6 x 10l W/cm2 range. TOF spectra of the

nanosecond pulse induced plasma exhibited only a single peak distribution with no

suprathermal ion peaks. The most probable ion kinetic energy increased with laser

intensity from 200 to 1900 eV and exhibited an 1040 dependence. However, no significant

difference in film quality was observed for the nanosecond Nd:YAG and femtosecond

Ti:Sapphire laser beams.


Background and Motivation

The interaction between laser beams and various forms of materials has been the

subject of immense research efforts, ever since the invention of the lasers. Much attention

has been directed at understanding the fundamentals of the laser-solid-plasma phenomena.

The advances in laser plasma physics also led to various practical applications in this area.

In the field of materials science, the use of laser beams to deposit thin films has shown

great promise in the synthesis of many advanced materials.

As a member of the physical vapor deposition (PVD) family, pulsed laser

deposition, or PLD, known for well over two decades, has gained prominence in the

deposition of a wide variety of thin film materials. They include superconductors,

semiconductors, dielectrics, metals, biomaterials, among others. In the PLD technique, a

high-powered laser beam is directed onto a solid target, creating a high pressure, high

temperature plasma with various excited target species (ions, neutrals, electronically

excited atoms, clustered microparticles, etc.). The ensuing hydrodynamic expansion of

this plasma in vacuum/gas medium leads to the deposition and growth of films with unique

physical and chemical properties. This technique can deposit not only materials which

reproduce stoichiometry and properties of the bulk target, but also films with desired

properties very different from that of the starting material. The advantage of laser

deposition lies primarily in the simplicity of system set-up, good deposition parameter

control, low contamination level, and possible low temperature epitaxial growth.

The deposition characteristics of PLD differ significantly from conventional

thermal evaporation methods. The PLD technique is first distinguished by the formation

of a high temperature (104 105 K) plasma, and ejected target species with high charged

states and high kinetic energies (-10 1000 eV, which is approximately one to three

orders of magnitude higher than the energies expected from conventional evaporation

methods), followed by the forward-directed nature of the ablated materials. The energetic

nature of laser induced plasma is believed to have played an important role in synthesizing

certain thermodynamically metastable materials, such as diamond-like carbon (DLC),

cubic silicon carbide (c-SiC), and cubic boron nitride (c-BN), etc.

DLC film, a unique form of carbon, has drawn great attention over the years,

primarily due to its prospect as an alternative to diamond films. While DLC films may be

inferior to diamond in terms of thermal stability, their microhardness, chemical inertness,

and optical transparency in the infrared (IR) and partially in the visible spectrum are quite

similar. In addition, a smoother surface topography, better adhesion to the substrates and

relative ease in fabrication have made DLC films a more desirable choice in certain


Most successful PLD depositions of DLC thin films conducted so far have

employed either excimer lasers (KrF, ArF, XeCl, etc.), or Q-switched solid state lasers

such as various frequency modified Nd:YAG beams. These lasers all have pulse duration

in the nanosecond range. Depending on the laser energy and beam diameter focused on

the target, laser intensities in the 108 10 W/cm2 range are most commonly achieved.

Past experiments have indicated that at this intensity range, the diamond-like component,

i.e., the fraction of sp3 bonded carbon atoms increased as a function of increasing laser

intensity. The plasma formed under these conditions was found to consist of ionized

carbon particles with moderately high kinetic energies, increasing from several

electronvolts up to several hundreds of electronvolts as a function of increasing laser

intensity. On the other hand, as a kinetic condensation process, formation of the

metastable diamond-like sp bonding structure is expected to be destroyed under excessive

high energy ion bombardment. In other words, the more stable sp2 graphitic structure will

again be favored under such conditions. For the nanosecond lasers, however, this

reversion phenomenon has yet to be clearly demonstrated. Because substantially higher

laser energy is often of very limited availability, while submicron beam spot required to

achieve higher intensity is limited by diffraction and coherence effects.

This motivated us to look into new laser systems which can produce higher

intensity levels. Instead of nanosecond lasers, we have deposited DLC films using laser

beams with a 100 femtosecond (FWHM) pulse duration. The laser system employed in

our study was a solid state Ti:Sapphire laser system developed at the Center for Ultrafast

Optical Science, University of Michigan. It is based on the latest chirped pulse

amplification (CPA) techniques. By using this system, we were able to push the intensity

up to the 1014 10 15 W/cm2 range with moderate laser energies. The laser plasma created

under these conditions was studied through time-of-flight (TOF) experiments. This study

was intended to establish a correlation between the carbon ion kinetic energy and the DLC

film properties, as a function of laser intensity. Similar experiments were also conducted

using a conventional Q-switched Nd:YAG laser (7 ns @ FWHM) beam, which served as a

reference system to that of the femtosecond study.

Femtosecond laser deposition of thin film materials remains a largely unexplored

territory in the field of materials science. These ultrashort laser pulses enable the

generation of extremely high laser intensities, which may have significant application for

the synthesis and processing of advanced materials such as DLC. Though certain

similarities will be present, the laser-matter-plasma interactions associated with the

ultrashort femtosecond pulses are expected to be distinctively different from that in the

nanosecond regime. Most research efforts so far are limited to the phenomena associated

with high field plasma physics. These include high-harmonic generation, high-brightness,

short pulse X-rays formation and wakefield plasma acceleration. With the advance of

ultrashort laser techniques, scientists are just beginning to explore their potential value in

the areas of thin film deposition, micromachining, as well as medical-surgical procedures.

Due to the lack of comparable information, on the subject of femtosecond laser

deposition of thin films, we have limited our investigation to the study of DLC film

deposition characteristics. This approach is merited by the findings that DLC film quality

is a strong function of the laser intensity for nanosecond laser pulses. Our intention in this

study was to look beyond the relatively matured nanosecond laser deposition techniques,

into an area where new phenomena associated with the ultrashort laser-matter interactions

may be beneficial to the synthesis of advanced materials such as DLC, c-BN, and P3-C3N4.

Dissertation Outline

The primary objective of this study was to investigate the effects which a high

intensity, femtosecond laser generated plasma would have on thin film materials.

Diamond-like carbon (DLC), because of its monoatomic nature and relatively well

established correlation between the film properties and laser intensity in the nanosecond

range, was chosen as the model material.

Chapter 2 first gives a general review on the subject of diamond-like carbon thin

film depositions. This includes an introduction to the DLC concept, its atomic structure

and properties, accompanied by a discussion of the state of the art deposition techniques

and associated film formation mechanisms. Aspects of the laser-solid-plasma interactions

within the intensity regime relevant to those used in our study are discussed.

Chapter 3 accounts for the experimental aspect of this study. A detailed
description of the unique femtosecond Ti:Sapphire laser system is given. Deposition

system set up and DLC thin film deposition conditions are outlined, followed by a

discussion of the time-of-flight (TOF) experiments.


Chapter 4 presents the results of each experiment described in Chapter 3, including

the carbon ion kinetic energy measured as a function of laser intensity through TOF, and

DLC film properties characterized by various techniques. Topics relevant to these results

are discussed.

The conclusions from this dissertation are summarized in Chapter 5.


Diamond-like Carbon (DLC) Thin Films

Structure and Property

Carbon atoms, when bonded with each other, have three basic types of electron

hybridization: one-dimensional sp' bonding in an acetylene like structure, trigonal sp2

bonding in a layered, threefold graphite structure, and tetrahedral sp3 bonding found in a

fourfold-coordination diamond structure.

In a tetrahedral diamond structure, the four L shell carbon electrons in the 2s and

2p orbitals are reconfigured: one of the two 2s electrons is elevated to the 2p orbital, the

three electrons in the 2p orbital then recombine with the remaining 2s electron, giving rise

to four new sp' orbitals. These hybrid atomic orbitals are labeled sp3 since each one is

now a combination of one s and three p orbitals. The sp3 orbital configuration allows the

development of an extremely strong bond when combined with a sp' orbital from another

carbon atom. The concentration of these bonding electrons optimizes the repulsive and

attractive forces between themselves and both carbon nuclei. By convention, a directional

sp3 orbital is called a sigma (a) orbital, and the bond a a bond. Every tetrahedron of these

hybridized sp3 atoms combines in space with four other carbon atoms at an angle of -1090

to form the three dimensional diamond lattice structure, as shown in Figure 2-1 [1]. Each

carbon atom in a diamond structure is therefore said to be sp3 bonded and has four a

bonds. The sp' C-C bond in diamond has a bond length of 1.54 A and a bond energy of

370 kJ/mol.

Fig. 2-1 Lattice structure of diamond.



B To
6.70 A
3.35 A, [

1.42 A

1.42 A 2 a

Fig. 2-2 Lattice structure of graphite.

In a graphitic carbon structure, one of the two ground level 2s electrons is excited

and combined with the two 2p electrons to form three hybrid sp2 orbitals, plus a fourth

unhybridized free 2p orbital electron. The three sp2 orbitals form a 1200 angle from each

other in the same plane. Each sp2 orbital is called a a orbital and the bond a C bond. The

fourth non-hybridized p electron forms a non-symmetrical orbital perpendicular to this

plane, and it is available to form a subsidiary 7t bond with other carbon atoms. This

configuration is responsible for the layered hexagonal lattice structure of graphite in space,

as illustrated in Figure 2-2 [1]. Each carbon atom in a graphite is said to be sp2 bonded

and has three a bonds and one n bond. The a and 7t bond in graphite have a bond length

of 1.42 A and 3.35 A, respectively. The a bond in graphite is thus shorter and with a

bond strength of 570 kJ/mol, is even stronger than that of diamond, due to the more

overlapped sp2 electron density configuration in graphite. The nt bonds, however, are

much weaker with a bond energy of only 7 kJ/mol. These frail van del Waals nt bonds are

responsible for the softness of graphitic materials.

Similarly, in an acetylene like structure, one 2p electron combines with an excited

2s electron to form two sp' hybridized orbitals. These sp orbitals form an angle of 180,

making the sp molecule linear in space. Each carbon atom is said to be sp' bonded and

has two a bonds and two n bonds.

As a result of these different bonding structures, carbon has several allotropes.

They include graphite, diamond, and fullerene molecules. Another form of carbon, most

commonly known as diamond-like carbon, or DLC, can be considered as its fourth

allotrope. The name diamond-like is arguably originated from the work of Aisenberg and

Chabot [2], who first deposited hard carbon films with properties determined to be similar

to that of diamond. Ions extracted from a carbon arc in argon were used in their

experiment. Subsequent work by other researchers led to an explosive growth of interest

in this unique material.

DLC can be considered as a metastable form of carbon covering the spectrum

between diamond and graphite. But unlike crystalline graphite or diamond, which

respectively consists of 100% sp2 hybridized carbon atoms and 100% sp3 hybridized

carbon atoms, DLC films have a mixture of sp3 and sp2 hybridized bonds and are generally

amorphous. This is due to the lack of a long range order of the sp3 and sp2 bonds in DLC

films. On the other hand, DLC films possess properties similar to those of diamond rather

than graphite, including high mechanical hardness, high electrical resistivity, excellent

chemical inertness and optical transparency. For these same reasons, they have been one

of the most actively researched thin film materials over the past two decades, as an

alternative to diamond coatings.

Diamond-like carbon films can be broadly categorized into two forms:

1. Hydrogenated DLC (a:C-H, 'a' as amorphous) prepared using hydrocarbon gas

mixtures, mainly through chemical vapor deposition (CVD) techniques.

2. Hydrogen-free DLC (a:C) films fabricated through physical vapor deposition (PVD)

processes, using a solid graphite target.

Hydrogenated DLC films (a:C-H) are often deposited at elevated temperatures

with a mixture of hydrogen and various kinds of hydrocarbon agents, such as methane,

ethylene, or acetylene [3-5]. Most commonly used techniques in making these films

include hotfilament discharge, DC glow discharge, RF plasma discharge and other

variations of similar CVD processes. The deposition mechanism is thought to be similar

to that of CVD diamond. The presence of dissociated atomic hydrogen is believed to

preferentially remove sp2 bonds, thus in effect promote the growth of diamond-like sp3

bonds. Hydrogenated DLC films usually contain planar sp2 hybridized carbon clusters that

are interconnected by tetrahedral sp3 bonds in a random fashion. These films often contain

a large amount of hydrogen varying from about 10% up to 60%, and they are considerably

softer than diamond [6]. The ratio of the sp3/sp2 bonds was found to scale with the

percentage of hydrogen in the films [7,8]. A reduction of hydrogen concentration causes

degradation toward graphite [9].

Hydrogen-free DLC films (a:C) are most commonly deposited through ion beam

sputtering [10], direct ion beam deposition [11], magnetron sputtering [12], cathodic arc

discharge [13], and pulsed laser depositions (PLD) [14]. These films are hard, chemically

inert, and optically transparent in the infrared (IR) and near IR spectrum. They usually

have a somewhat higher film density but lower optical bandgap when compared with

hydrogenated DLC films.

A common feature amongst various PVD deposition processes is the presence of

ionized carbon species in the plasma. The bombardment of the exposed substrates with

these ions, which often have medium kinetic energies ranging from tens to hundreds of

electronvolts, has proven to be instrumental in attaining a condensed structure with a high

volume fraction of sp3 bonded carbon atoms. One inherent advantage to these non-

equilibrium techniques, where the plasma kinetic energy is much higher than that achieved

by purely thermal methods, is that substantially lower substrate temperatures are required

for these processes. This makes it possible to deposit high quality DLC films on heat

sensitive materials. Also, virtually free of C-H absorption bands, these hydrogen-free

DLC films offer some unique advantages when used in certain optical applications because

the scattering losses are greatly reduced. The growth of these a:C films is self-seeding at

room temperature, and substrates require little or no preliminary preparation. Wear and

tribological applications also benefit from the low surface friction generally displayed by

these films. In this case, they are better suited than CVD diamond films which have rough

facets and routinely require high growth temperatures.

The concept of DLC is very inclusive. It comprises a tremendous amount of

carbon-based thin film materials. Their structures often depend upon the preparation

methods. As a result, the properties of these DLC films vary drastically from one another.

However, two parameters are considered to have the most impact on DLC film quality,

regardless of the deposition technique: kinetic energy and elemental composition of the

primary plasma. For hydrogenated DLC films, as the impact energy increases, one obtains

material from graphitic polymer to dense hydrocarbon to dense a:C-H. For a:C films, sp2

based carbon films are expected at low energies and diamond-like sp3 based films at higher

energies. An ion beam or plasma generated from a solid carbon target generally contain

various types of polyatomic ions, neutrals, and electronically excited carbon microclusters

(Cn"m, C,, etc.). In certain sputtering processes such as ion beam assisted deposition or

laser deposition, the carbon energy usually has a Maxwellian-like distribution. The

polyatomic species at the low energy tail sometimes do not have enough energy, upon

impact to overcome their binding energies and break up into monoatomic particles at the

substrate surface. This could contribute to the development of undesirable surface

particulates and also suppress the sp3 bonding formation that is believed to be favored by

energetic/monoatomic ions. Similar notions hold when hydrocarbon sources used in the

CVD processes are considered. Species of the general form CxHy, (CxHy),', etc. can be

present in the plasma. Though the structure of these hydrocarbon precursors are often not

well known, the deposited a:C-H films are rarely found to depend on the type of

hydrocarbon used as the carbon source.

Table 1 summarizes some of the DLC film properties as well as those of diamond,

CVD diamond, and graphite.


Due to their unique microstructure with high percentage of sp3 bonded atoms,

DLC films possess many properties comparable to that of diamond. Some of their most

notable assets include high hardness, low friction, high electrical resistivity, excellent

chemical inertness and good optical transparency in the IR and near IR wavelength. They

are beginning to find usage in a wide range of areas. The following sections briefly discuss

some applications of the DLC films.

Table 1. Properties of diamond, CVD diamond, graphite and DLC.

Diamond CVD Graphite a:C a:C-H


Density (g/cm3) 3.51 3.3 -3.5 2.2 1.6-3.1 1.5-2.0

Hardness 10000 6000- 1200- 1200-

(kg/mm2) 10000 8500 5000

Bandgap (eV) 5.45 5.0-5.5 0.0 0.0-2.6 1.2-2.5

Resistivity 1016- 1018 1012- 1017 10" 17 0'6 0 10 16 10 1016


Refractive index 2.42 2.2-2.4 -3.0 1.6-2.4 1.8-2.2

( @ 632 nm)

Microstructure crystalline crystalline crystalline/ amorphous amorphous


Thermal metastable metastable stable metastable metastable


Bonding 100% sp3 -_100% sp3 100% sp2 sp3/sp2 sp3/sp2/sp1


Wear and tribological. Because of their superior hardness and extremely low friction

coefficients (< 0.04 at low humidity), DLC films can be used as wear reducing coatings

[15]. A DLC coated silica optical fiber can inhibit the permeation of moisture, thus reduce

the risk of brittle fracture [16]. Ion beam deposited DLC coatings have found commercial

use as protective/decorative layers on polycarbonate sunglasses, where they were found to

be more scratch resistant than most other coatings [17]. Overcoat for magnetic thin film

recording disks is yet another possibility for DLC films [18,19], because of their excellent

surface smoothness and hardness.

Optical. Though DLC films are less transparent than diamond in the visible

spectrum, they nonetheless have high transparency in the IR and near IR range.

Moreover, DLC films can be made at low temperatures and have an atomically smooth

surface, making them more suited as protective coatings for IR windows. On the other

hand, CVD diamond can only be deposited at relatively high substrate temperatures, thus

limiting the available optical windows for use as substrates. In addition, because CVD

diamond films often display growth facets with roughness up to several microns, extensive

post-deposition modification of the film surface is also required to minimize light

scattering. ZnS and ZnSe IR windows coated with DLC films were found to be highly

transparent in the 3 15 .tm range and exhibited large improvement in chemical and

abrasion resistance over uncoated ones [17]. DLC could suit well as protective layers for

similar soft IR windows such as MgF2, Ge, and CdTe, allowing them to be used in a

corrosive environment. By varying the refractive index of DLC films, they can also be

used as anti-reflection (AR) IR window coatings. Furthermore, due to their high damage

threshold, hard DLC films with high sp3 content could serve well as protective layers for

high-power laser windows [20].

Microelectronics. DLC films with high extinction coefficients have a dark

appearance and they have been used to provide an increased contrast enhancement ratio in

liquid-crystal displays (LCD) [21]. With a high electrical resistivity and breakdown

voltage up to 106 V/cm, DLC films are being used as insulating layers in some electronic

devices. These include high-energy-density, high-voltage capacitors [22], metal-insulator-

semiconductor (MIS) devices [23], as well as metal-insulator-metal (MIM) devices [24].

DLC films are also being considered as photoresists in lithographic applications. Through

selective graphitization of DLC film by laser/oxygen exposure, the patterned DLC film can

act as an in-situ mask for underlying silicon substrates [25].

Biological and medical. DLC is a biocompatible material [26]. It may find use as

biomedical coatings for heart implants, hip joints and other prostheses.

High internal stress and relatively low thermal stability are among the factors

limiting the usefulness of DLC films. Hydrogen-free DLC films deposited with ion beam

or lasers routinely have compressive stress up to several gigapascals. This high internal

stress prevents the deposition of DLC films beyond a few microns on rigid substrates. The

compressive stress depends strongly on preparation method. It increases with film

thickness and is relieved through peeling and buckling of the films [27,28]. Compressive

stress also contributes to poor adhesion between the DLC film and the substrate. This

problem, however, can be alleviated through the deposition of carbide-forming materials

such Si, Ti, W, or Ge as buffer layers. Alternatively, a better film adhesion can be

achieved by increasing the carbon ion energies to form an intermediate carbide layer at the


Most DLC films exhibit high hardness, low optical absorption and high electrical

resistivity at room temperature, but their properties rapidly deteriorate at elevated

temperatures. A graphitization process starts at a couple of hundred degrees Celsius.

This thermal annealing process breaks down the tetrahedral sp3 bonds and encourages the

carbon atom diffusion to form trigonal sp2 bonds, which are thermodynamically more

stable. Transformation from DLC to a fully graphitic structure is completed at about 500

to 600 C, at which point the films become soft, electrically conductive, and highly

absorbing. Hua et al. have deposited DLC films with a KrF excimer laser beam (248 nm,

12 ns FWHM), at a laser intensity of 8 x 10s W/cm2 [29]. They measured the film density

as a function of substrate temperature, using the grazing-angle X-ray reflectivity

technique. From 22 C to 300 C, the film density was found to have decreased from 3.1

g/cm3 to 2.4 g/cm3 (corresponding to sp3 fractions of 77% and 17%, respectively), along

with a decline of optical bandgap from 2.2 eV to 0 eV. This observation clearly indicated

a graphitization process in progress. In an earlier experiment conducted by Capano et al.

[30], a KrF excimer laser (248 nm, 25 ns FWHM) was used to deposit DLC films at

substrate temperatures ranging from room temperature to 600 C. Raman spectroscopy

and grazing-angle X-ray reflectivity technique, respectively, were used to study the DLC

microstructure and film density. A graphitization process was observed to have occurred

at temperatures higher than about 200 C, leading to a decreased film density [31].

Similar findings were reported by Ong et al. [32].

Though many applications are being discovered for diamond-like carbon films,

industrial scale deposition of DLC films has so far been very limited. This is largely due to

the fact that there still lacks a suitable technique, which can sustain a reasonable deposition

rate without compromising the film quality. Instead, the current state of research has

mainly been focused a somewhat different subject: what is the deposition mechanism that

leads to the best film quality, i.e., DLC films with the highest sp3 bonding contents?

Deposition Mechanisms

Maximization of the diamond-like sp3 bonding content has been a main objective in

making DLC films. Ever since the successful synthesis of CVD diamond, much attention

has been directed at finding a suitable low pressure, low temperature technique for the

deposition of DLC films. This is because both DLC and diamond contain the metastable

sp3 bonds. The deposition mechanism for the stabilization of sp3 versus sp2 bonding

structure in a DLC film is still under dispute, not unlike the situation on CVD diamond

film depositions several years ago. Various deposition techniques have been tested,

generating a vast amount of information on the DLC film properties as a function of

deposition parameters. From these experimental results, several models on the formation

and stabilization of sp3 bonds have been proposed. Some of the most notable include:

1. Preferential sputtering of sp2 versus sp3 sites, first proposed by Spencer et al. from an

earlier DLC work in the 70's [33];

2. Stabilization of the sp3 structure due to high intrinsic stress in DLC film, suggested by

McKenzie et al. [34,35];
3. Sp3 bonding formation resulting from energetic carbon ion quenching on substrates,

proposed by Robertson [36].

In the preferential sputtering model, Spencer [33] hypothesized that the tetrahedral

sp3 bonds will be more resistant to sputtering than the trigonal sp2 bonded carbon atoms.

At ion impact energy range of 50 500 eV, the incident ion flux can simultaneously

deposit new carbon materials and resputter non-sp3 structures during film growth. Lifshitz

et al. [37] offered a similar "sub-plantation" mechanism based on the estimation that the

displacement threshold for graphitic carbon atoms is less than half that of diamond (30 eV

vs. 80 eV). Thereby a preferential displacement of the sp2 sites by influx ions would

eventually lead to an accumulation of the sp3 sites. However, neither of these two models

could explain the initial nucleation of sp3 bonded structure. They also have ignored the

fact that resputtering of the sp2 bonds is an unlikely event at the energy range considered

because carbon has extremely low sputter yield when compared to other materials [38,39].

Moreover, the large difference between the suggested displacement threshold of graphite

and diamond sites was also put into question, by a more recent measurement from Koike

et al. [40], who found both of them to be approximately 40 eV.

Through their work on mass-filtered, plasma vacuum arc deposition of DLC films,

McKenzie et al. [34] presented a stress induced sp3 formation mechanism. They

suggested that the high compressive stress generated by energetic ion impact could serve

as the driving force for converting sp2 bonds to the metastable sp3 bonding structure, in

the interior of DLC films. This proposition was based on the pressure-temperature phase

diagram of diamond and graphite. They found in their experiment that by negatively

biasing the substrate at between 20 and 70 V, a maximum compressive stress of 7 GPa

was present inside the films. These DLC films had a maximum density of 3.1 g/cm3,

corresponding to a sp3 bonding percentage of 70%. If the substrate bias was lower than

20 V or higher than 70 V, the compressive stress was found to fall back to lower values,

accompanied by a decrease in film density. The exact ion charge state was not clearly

stated in their experiment, as the mass filtering mechanism did not provide a complete

selection of single species. However, if assuming singly charged C' ions are the

predominant species in the ion beam, one concludes that the highest sp3 bonding

percentage can be expected at an ion kinetic energy between 20 and 70 eV.

Kelires' [41] work appeared to support the above observation. He concluded that

the high intrinsic stress in ion deposited DLC film was a result of high free energy of the

sp3 sites. These sp3 sites in DLC films have a free energy approximately 0.3 eV higher

than that of the sp2 sites. The extra free energy can introduce a compressive stress of up

to 15 GPa in the film. This intrinsic stress can be interpreted as the work done by incident

ions to compress/convert the sp2 sites to sp3 sites in the McKenzie model.

More recently, Robertson [36] has suggested a coupled energetic ion condensation

/thermal quenching model. He reasoned that the bonding hybridization in a DLC film

would adjust itself according to the local density: more sp2 bonds at lower density and

more sp3 bonds at higher density. A high film density requires an incident carbon ion to

have at least enough energy to penetrate the first few atomic layers and enter an interstitial

site. The accumulation of these ions will lead to an increase in the local film density. As a

result, the local bonding will reform around these sites to become bulk bonding at the

appropriate bonding hybridization, i.e., DLC film with high sp3 contents. Conversely, if

the ions have excessively high energies, they will be able to penetrate deeper into the

layers. The potential film density increment induced by these ions, however, will not be

able to materialize due to the lack of a sufficient energy dissipation mechanism in the local

area. A thermal relaxation process will instead take over, resulting in a decreased film

density and hence an overall more graphitic bonding structure with mostly sp2 bonds.

Using a thermal spike model to describe the density increment/reduction process, he

obtained an estimated optimal ion kinetic energy of 60 eV.

Fallon et al. [42] have, in a related experiment, observed this energy to be around

140 eV using a mass-filtered, singly charged C' ion beam.

Rossi et al. [43] have also conducted similar experiments and they later determined

that the optimum carbon ion energy to be somewhere between 150 and 200 eV.

Laser Deposition of DLC

Ion beam deposition (IBD) of DLC films, using a mass-filtered carbon ion beam

has attracted much attention as of late. State of the art IBD techniques have been able to

produce hydrogen-free DLC films with up to 75% sp3 carbon bonds, as well as high

hardness and electrical resistivity, as demonstrated by the work of Fallon and Rossi

[42,43]. A relatively precise control over the carbon ion kinetic energy distribution and its

charge state is the main advantage of this technique. However, a low deposition rate (up

to several tens of nanometers per hour at best) continues to be its drawback. Over the

past few years, another versatile thin film deposition technique called pulsed laser

deposition, or PLD, has emerged as a worthy alternative in making DLC films with

comparable qualities.

In a typical pulsed laser deposition process, a solid target is irradiated by a high-

powered laser beam in a high vacuum environment. Target materials are removed by the

intense laser beam in the form of a highly excited plasma. Due to the large pressure

gradients present within the hot plasma, it expands rapidly in a beam-like fashion in front

of the target surface. This plasma is then allowed to condense onto a substrate placed at a

short distance away from the target. Background gas pressure up to several hundred
millitorr and/or elevated substrate temperatures are sometimes applied to achieve desired

deposition rate and film properties. The usefulness of the PLD technique resides primarily

in its simplicity in set up and the ability to generate a highly energized plasma, the latter

being the key to the formation of metastable materials such as DLC.

Early attempts to make DLC films with laser beams were initiated by Fujimori et

al. in the early 80's [44,45]. They used a continuous wave (CW) CO2 laser to vaporize a

powdered graphite target. Carbon films were deposited onto fused silica substrates at

room temperature, with a laser intensity of 5 x 103 W/cm2. Only conductive, graphitic

films were formed at first. They were then able to improve the film quality, indicated by

higher film electrical resistivity and optical transmittance, by introducing a simultaneous

Ar ion bombardment on the substrate. In light of these results, subsequent research efforts

have since shifted to further increase the laser-solid energy coupling, with laser sources

capable of generating much higher intensities. Pulsed excimer lasers, with their combined

features of moderately high energy and short pulse duration, became a logical choice. The

experiments carried out in the mid-80's by Marquart [46], Pompe [47], and Sato [48]

earmarked the maturing of PLD technique in making DLC films.

The research groups led by Marquart [46] and Pompe [47], respectively, used Q-

switched nanosecond Nd:YAG lasers in their experiments. Marquart et al. [46] were able

to produce DLC films at peak laser intensities ranging from 3 to 13 x 1010 W/cm2. The

extinction coefficient and optical bandgap of their films were measured to be 0.3 and 0.4

eV, respectively. They studied the plasma electron temperature as a function of laser

energy, by taking X-UV emission spectra during deposition. Scratch resistance tests were

also performed on the DLC films. They discovered that both the plasma electron

temperature and the film hardness increased as a function of increasing laser energy. The

electron temperature and film hardness appeared to have reached a maximum value at

laser intensities of 5 x 1010 W/cm2 and higher. This observation prompted them to

conclude that there existed a threshold laser intensity of 5 x 1010 W/cm2, beyond which

hard, diamond-like carbon films were formed, and below which only soft, graphitic ones

were made. They further suggested that a plasma electron temperature of 40 eV,

measured at the threshold laser intensity of 5 x 1010 W/cm2, coincided with the

resputtering energy for sp2 bonded carbons as proposed by Spencer [33] in the

preferential-sputtering mechanism. Pompe et al.[47] also asserted that carbon films with

some diamond-like characteristics could be deposited with Nd:YAG laser pulses. Raman

spectroscopy was used for the first time as an analytical tool in DLC film characterization,

by linking the Raman band position to the presence of sp3/sp2 carbon bonds. However,

Pompe's experiment did not report any direct measurement of the film properties such as

film transmittance or hardness. The laser intensities used in their experiments were also

lower than the critical value of 5 x 1010 W/cm2 observed by Marquardt.

Sato et al. [48] took a slightly different approach by employing a XeCI excimer

laser (308 nm, 10 ns FWHM) to deposit DLC films. Typical peak laser intensity used in

their experiment was estimated to be 3 x 108 W/cm2, well below the critical density of 5 x

1010 W/cm2 observed by Marquardt [46]. Quality DLC films with an electrical resistivity

up to 108 a-cm, an optical bandgap of 1.4, and a refractive index of 2.1 2.2 were

obtained at substrate temperature of 50 C. Higher substrate temperatures produced only

opaque and conductive carbon films. At even higher laser intensities an increase of the

deposition rate was compromised by the presence of higher particle densities on the film

surface. One notable feature emphasized by these researchers was the presence of small

percentage of carbon ions (estimated to be ~ 0.01%) in the laser induced plasma. The

authors went on to suggest that this ionized plasma was likely to be responsible for the

production of carbon films with diamond-like characteristics. This is believed to be the

first time a direct correlation between the DLC film quality and an ionized plasma was


The late 80's and early 90's saw an explosion of research activities in the laser

deposition and characterization of DLC thin films. Several research groups, notably those

led by Collins [49], Pappas [50], and Murray [51] have been among the most active in this


After realizing that a highly activated laser plasma is the key to produce carbon

films with substantial diamond-like properties, researchers began to direct their focus to

further enhance the laser induced plasma. By doing so, they hoped to increase the degree

of plasma excitation and to improve the uniformity of the strongly forward-directed laser

plume. Several different paths were taken, including the use of laser beams with different

wavelengths, deposition at higher laser intensities, and the addition of auxiliary plasma

enhancement devices. A combination of these methods led to the so-called hybrid pulsed

laser depositions.

Two of the most widely adopted hybrid techniques were introduced by Collins

[49] and Krishnaswamy [52], respectively. The basic idea behind their modifications was

the implementation of supplementary electrical discharges to the laser plasma.

Collins et al. [49] used a Nd:YAG laser (15 ns FWHM) operated at the

fundamental wavelength of 1064 nm in their experiment. Peak laser intensity up to 5 x

1011 W/cm2 was found to be necessary to produce quality DLC films. Furthermore, they

incorporated an auxiliary discharge to further increase the laser plasma temperature by

Joule heating in the volume of the ablation plume. A voltage ranging from 2000 to 2000

V was applied to the discharge electrode placed between the target and substrate. When

the space between the target and substrate was filled with laser induced plasma, a strong

discharge plume with current up to 10 A was measured. A Rogowski coil was used to

monitor the current change during the plume discharge. With this set up they were able to

produce optically smooth DLC films at growth rates of up to 0.5 Vm/hr over an area of 20

cm2. In several ensuing experiments using the same technique, Davanloo et al. [53,54]

claimed to have deposited a unique form of hydrogen-free DLC film, which appeared as a

conglomerate of fined grained particles. These particles, when studied under a scanning

tunneling electron microscope, were found to have grain sizes ranging from 100 to 200 A.

They were speculated to be constructed of diamond-like sp3 bonds while the intergrainular

material was made of defected graphite. The DLC film quality was found to depend

strongly on the nature of the laser induced plume. Close to the center of the forward-

directed plasma, where the carbon ions have the highest charge states/kinetic energies, one

obtained DLC films with the best optical transparency (with an extinction coefficient of

0.16 and an optical bandgap of 1.15 eV) and microhardness (with a film density of 2.9

g/cm3 and a hardness of 1500 kg/mm2). Higher laser intensities also resulted in lower

extinction coefficients and higher optical bandgaps. Growth rate of 0.5 gtm/hr over a 100

cm2 area was obtained on various substrates, including Si, Ge, ZnS, and glass. The

authors later termed these DLC films as amorphicc diamond' to distinguish them from

those deposited from ion beam deposited DLC films. In a more recent paper presented by

the same research group[55], the authors showed the TEM picture of a gold coated

replica of the amorphicc diamond" film. It was compared with that of a DLC film

condensed from a mass separated C' ion beam with a kinetic energy of 2 keV. The

topography of these amorphicc diamond" (alternatively termed as "nanophase diamond"

by these authors) films appeared to more closely resemble that of a polycrystalline

diamond structure: the nodules seemed to be disordered mixtures of the cubic and the rare

hexagonal polytype (Lonsdaleite) of diamond. The authors went on to theorize that at

laser intensities higher than about 1011 W/cm2, the predominant species generated near the

target surface would be multiple charged C3' and C4+ ions. These highly charged carbon

ions expanded in vacuum with their charge states frozen in, giving rise to a high energy ion

source at distances up to 10 cm away from the target. The energetic condensation of

these C"3 and C41 ions, which possess kinetic energies up to 2 keV, facilitated the

formation of the "nanophase diamond" structure. This energetic condensation process

was also believed to have resulted in a unique chemical bonding, such that little residual

compressive stress (0.7 0.8 GPa) was measured in the film. The film hardness was

estimated to have a minimum value of 8000 kg/mm2. This experiment is presumed to be

the biggest success in making hydrogen-free DLC films to date. However, by closely

examing some of the claims by these researchers, one finds areas of controversy that need

to be further investigated. These include the polycrystallinee diamond" microstructure of

the films, the presence of multiple-charged carbon ions as the predominant plasma species,

as well as the absence of neutral or clustered microparticles in the laser plasma. Though

high intensity laser plasma has been found to generate carbon species with moderately high

kinetic energies, the presence of multiple-charged ions such as C3' and C4+ at a distance up

to 10 cm away from the plasma source has yet to be experimentally confirmed. Moreover,

if the microstructure of this "nanophase diamond" is indeed that of polycrystalline

diamond, a film density measurement should have yielded a value substantially higher than

2.9 g/cm3 observed in their experiments. The absence of any microstructual techniques,

such as Raman spectroscopy or X-ray diffraction to verify the existence of diamond

structure has left several unanswered questions as well.

Krishnaswamy et al. [52] have deposited DLC films using a XeCl excimer laser

beam (308 nm, 40 ns FWHM). They employed a capacitor-discharge hybrid technique: a

0.5 iF capacitor was connected between a graphite ring electrode (2 cm in diameter,

positioned at 0.65 cm away from target) and the graphite target. The capacitor was

charged by a high DC voltage supply at up to 3000 V. This capacitively stored energy

was automatically released when the laser plume filled the space between the target and

substrate. A large discharge plume extending from the ablated spot onto the substrate was

observed, indicating a further activation of the laser plasma. The enlarged plume led to a

better film uniformity on various substrates. Insufficient data were available to compare

the film qualities deposited with and without the plasma enhancement modifications.

Leppavuori et al. [56] also deposited DLC films with a XeCI excimer laser (308

nm, 20 ns FWHM), without any additional laser enhancement modification. They

demonstrated that the best quality DLC films were made at room temperature with the

highest available laser intensity of 8 x 109 W/cm2.

Pappas et al. [50] have synthesized DLC films with a KrF excimer laser (248 nm,

30 ns FWHM) at a relatively low peak intensity of 1.4 x 10s W/cm2. Depositions at or

below room temperatures produced hydrogen-free, transparent DLC films with an

extinction coefficient of less than 0.4 at 300 nm wavelength. Electron energy loss

spectroscopy (EELS) revealed that the sp3 bonding fraction was on the order of 70% -

85%. The film electrical resistivity was measured at room temperature to increase from

102 to 108 92-cm, as a function of increasing laser intensity. These improved film

properties were attributed to the energetic nature of the laser plasma, because the carbon

ion kinetic energy was also observed to have increased as a function of increasing laser

intensity. The use of an auxiliary plasma enhancement modification, similar to that

employed by Krishnaswamy [52], was found to have yielded slightly inferior DLC films to

those made with laser plasma alone.

In the experiment carried out by Xiong et al. [57], an ArF excimer laser (193 nm,

21 ns FWHM) was used to deposit hydrogen-free DLC films at a peak laser intensity of 5

x 101 W/cm2. At room temperature, the DLC films were measured to have a mechanical

hardness up to 4000 kg/mm2 and a maximum optical bandgap of 2.6 eV. The highest sp3

bond fraction was estimated through EELS measurement to be 95%. This result is

believed to be the highest tetrahedral bonding percentage ever measured on a diamond-

like carbon thin film, hydro- or unhydrogenated. The films were also found to have good

thermal stability up to 800 C, where the optical bandgap showed only a slightly lower

value than that measured at room temperature. This also marked the first time a thermally

stable DLC film was claimed to have been made, by any type of deposition techniques.

Though no film electrical resistivity measurement was conducted in conjunction to further

verify this result, the authors nonetheless made two interesting observations through their


1. High intensity laser induced laser plasma alone was sufficient to produce high quality

DLC films, without any need for an auxiliary plasma enhancement system;

2. Laser beam with shorter wavelength in the UV range, in contrast to those in the IR

range, appeared to have produced DLC films with superior qualities. The reason was

attributed to a change in the optical penetration depth which decreases with shorter

wavelength. Higher photon energy associated with shorter laser wavelength leads to a

more efficient coupling of the laser energy to the target, resulting in the production of

atomic ion species and an increase of their kinetic energy. The authors speculated that

atomic C' ions with a mean kinetic energy of 50 eV and over were present in the laser


Film deposition by sub-nanosecond lasers. The phenomena associated with the use

of ultrashort, intense laser beams have been of interest in the high field physics community

for well over a decade. High-harmonic generation, high-brightness, short pulse X-rays

formation, wakefield acceleration, and possible thermonuclear fusion reactions are among

the research subjects [58]. The use of ultrashort pulsed lasers in the deposition of thin film

materials, however, remains an uncharted territory. Researchers are only beginning to

explore the potential applications this unique technique might bring to the materials

science related areas, such as medical-surgical procedure, micromachining, as well as thin

film material processing [59-61].

We are aware of only one other research group that has conducted some

preliminary experiments on the femtosecond laser deposition of thin film materials [62].

In their experiment a KrF excimer laser, with a 30 ns temporal duration (FWHM) at 248

nm wavelength, was compressed to deliver 500 femtosecond (FWHM) pulses.

Depositions of DLC, silicon and copper thin films were carried out using both the 30 ns

and 500 fs pulses, at laser intensities of 2.3 x 10' and 1.4 x 1013 W/cm2, respectively. For

the femtosecond laser pulses, the particle kinetic energy, mass distribution and optical

emission spectra were found to be significantly different from the respective nanosecond

results. While only atomic carbon ions were detected in the 500 femtosecond pulse

induced plasma, cluster formation was found in the time-of-flight (TOF) spectra during the

30 ns laser ablation. This observation suggested a higher degree of target species

atomization (existence of atoms or singly charged ions instead of molecular clusters in the

plasma) with the use of shorter laser pulses. Both DLC and Si films showed improved

qualities by using the femtosecond laser pulses, due to the fact that higher kinetic energies

(in the keV range) were obtained in the femtosecond laser induced plasma, without the

formation of large particulates. The authors also suggested that there may exist a

threshold laser intensity somewhere between 109 and 1013 W/cm2, below which molecular

carbon particles were favored and above which atomic carbon ions would be dominant.

Table 2 summarizes the properties of hydrogen-free DLC films deposited by

various laser deposition techniques.

Table 2. Properties of DLC films deposited by laser beam.

Laser X(nm) Intensity sp3 n** k** Eg* Resistivity Reference

(W/cm2) %* (eV) (O-cm)*
CO2 10600 5x103 0 0 0 44, 45

Nd:YAG 1064 3- 0.4 46

13xl010 _

Nd:YAG 1064 5x10" 75 2.35 0.21 1.0 >107 49

Ruby 694 6.7x108 2.55 0.07 1.25 107 14

XeCl 308 3x108 2.2 0.04 1.4 48

XeCl*** 308 1.25x108 2.4 0.14 1.27 52

XeCl 308 1.25x108 2.7 0.85 0.47 52

XeCl 308 8x109 2.5 0.12 2.2 >104 56

KrF*** 248 1.4xl0' 15-27 2.53 0.14 1.5 >106 50

KrF 248 1.4x108 68 2.55 0.04 1.7 >106 50

KrF 248 8xl08 77 2.2 29

ArF 193 5x108 95 2.56 0.02 2.6 57

* measured at room temperature

** measured at 632 nm wavelength

*** with plasma enhancement modification

There are tremendous amount of information available on the subject of DLC film

fabrication using various laser deposition techniques. Though there still lacks a universally

accepted description of the DLC formation mechanism, many research groups have

reached the conclusions that:

1. Pulsed laser deposition is likely an ion assisted energetic condensation process;

2. The resultant DLC film quality is largely a function of the carbon particle kinetic

energy and the subsequent energy dissipation mechanism;

3. Higher laser intensity usually gives rise to a more energized plasma, resulting in DLC

films being more diamond-like.

In comparison, knowledge of the laser induced plasma, particularly its dynamic

transport between the target surface and the substrate, which ultimately determines the

resultant film quality, appears to be relatively scarce. The inadequacy in understanding the

laser plume dynamics is partly due to the difficulties involved in analyzing the highly

transient laser-plasma interaction processes, and partly because of the phenomenological

approach often taken by the materials science community with respect to laser deposition

of thin film materials.

Pappas et al. [63] have studied the vapor plume generated by a KrF excimer laser

irradiation of a graphite target, at a peak laser intensity of I x 108 W/cm2. This intensity

resembled the conditions under which actual DLC films were produced. Laser induced

fluorescence (LIF) and a Langmuir probe were used, respectively, to characterize carbon

molecules and ions present in the plasma. They learned that both neutrals (in the form of a

variety of small clusters) and ions were abundant in the laser plume, and the ions had

significantly higher kinetic energies than the neutrals. The kinetic energy of both ions and

neutrals increased as a function of increasing laser fluence (in units of J/cm2), along with

that of the electron temperature. At a 700 laser incident angle, the most probable kinetic

energies for the C2 molecules were measured to be 12 eV, and up to 80 eV/atom for the

C, ions. The ions were not mass resolved in their experiment, the authors thereby

assumed that all ions were singly charged. Laser ionization mass analysis (LIMA) using

266 nm radiation at similar laser intensity level revealed that the mass spectrum consisted

almost entirely of C, ions, with n varies from 1 to 12. Most of the yield was weighted

toward the lower n values. At a distance of 10 cm from the target surface, an ion fraction

of 5 10% was estimated. The authors conclude that the films become more diamond-like

as the laser intensity is increased, indicating that higher kinetic energy of the depositing

particles is a critical factor in obtaining high quality DLC films. This observation seemed

to have confirmed the energy-barrier model proposed by the same group in an earlier

report [64]. In that study, the researchers suggested that conversion of the low density,

low energy sp2 graphite geometry to the higher density, metastable sp3 geometry can be

envisioned on a potential-average atomic distance diagram. On this diagram, the sp2 and

sp3 bonding arrangement can be viewed as two minima which are the product of multiple

atomic interactions that occur as the particles are incorporated onto the surface. Between

them is the so-called potential barrier, which represents the energy required for the film

structure to change from one configuration to the other. When deposition techniques such

as thermal or electron beam evaporation is employed, carbons with thermal energy mostly

less than eV (leV = 1.16 x 104 K) are generated. They will subsequently condense into

the thermodynamically stable sp2 structure. In contrast, carbon particles generated by IBD

or PLD techniques have much higher kinetic energies (generally between tens and

hundreds of electronvolts). The impact of these highly energetic particles will force the

initially sp2 bonded carbon atoms to overcome the energy barrier into the metastable sp3

configuration. Furthermore, the excessive energies carried by these particles, if not

promptly quenched by the substrate, may again drive a sp3 bonded atom back over the

barrier to the equilibrium sp2 structure. This qualitative study is in essence a prototype of

the energetic ion bombardment/thermal quench model later proposed by Robertson [36].

Germain et al. [65] studied the plasma created during the interactions between a

graphite target and a KrF laser beam (248 nm, 15 ns FWHM). Time and spatially-

resolved spectroscopy were collected at a distance of less than 2 mm away from the

target, to detect the carbon particle charge state as well as its expansion velocity. At a

vacuum pressure of 5 x 10-' Torr and a laser fluence of less than 7 J/cm2, the laser plume

was found to contain mostly C2 molecules (atomic C signal was lost in the background

noise). These carbon neutrals had an expansion velocity of 9.4 x 105 cm/s, corresponding

to a kinetic energy E of 11 eV. C+ and C2+ ions, along with larger carbon clusters (charge

state unresolved) became abundant at higher laser fluences. Their expansion velocities

were determined to be in the 2 9 x 106 cm/s range (E = 25 500 eV), increasing as a

function of laser fluence. The authors also reported the possible existence of carbon ions

with even higher charge states. Good DLC film/substrate adhesion was achieved at room

temperature with high laser fluences.

A more comprehensive investigation on the dynamics of laser-graphite interactions

was later conducted by Murray and Peeler [66]. Through time-of-flight (TOF)

experiments, they measured the kinetic energy of the carbon particles ablated from a

graphite target with different laser beams. They used a Nd:YAG laser (15 ns FWHM) at

the fundamental 1064 nm and frequency-doubled 532 nm wavelength, in addition to an

excimer laser system (15 ns FWHM) running at 248 nm and 193 nm wavelength. Their

results indicated that ablation of graphite with the 1064 nm radiation produced a series of

low energy (E < 5 eV) clustered carbon Ct ions (1 < n < 30). Irradiation of graphite with

532 nm and 248 nm laser beams resulted in the ejection of molecular carbon ions (C2',

C3+), which have kinetic energies between 20 and 60 eV. Also, ablation by the 193 nm
laser beam generated mostly atomic C+ ions with a mean kinetic energy of 600 eV. The

DLC films deposited at various wavelengths were analyzed by electron energy loss

spectroscopy (EELS). The authors found that large, low energy carbon clusters tended to

induce the sp2 graphitic film structure, while highly energetic, atomic carbon ions

encouraged the formation of sp3 bonded diamond-like films. The strong correlation

between the film quality and the particle kinetic energy/laser wavelength supports

Cuomo's model [64] and the work by Xiong [57]. Table 3 tabulates the results from

Murray's experiment [66].

Table 3. Kinetic energy of carbon particles as a function of laser wavelength [66].

Wavelength Laser Intensity Predominant Ions Mean Kinetic Energy

(nm) (W/cm2) (eV)

1064 1.8 x 10 Ci 3.3

C15+ 4.5
532 1.8 x l0 C2+ 38

C3+ 17
248 2x 10' C2+ 55

C3+ 18
193 2 x 10' C+ 600

In a related experiment, using a KrF excimer laser at a laser intensity of 2 x 10s

W/cm2, Peeler and Murray [67] compared DLC films made with and without charged

carbon particles in the plume. Filtration of the carbon ions was achieved through the use

of ion deflectors mounted in the path of the laser plume. DLC films made with neutral

particles only and with both ions and neutrals were studied by valence/core level EELS.

All the films were found to have similar sp3 fractions at around 85%. Though the authors

supported the notion that the formation of DLC film by PLD is likely an energetic
condensation process, they questioned whether it is also an ion assisted process. A study

by Diaz's group [68] seemed to endorse Peeler's argument. Using a frequency-doubled

Nd:YAG laser beam (532 nm) at laser intensities in the 1010 W/cm2 range, they produced

DLC films with and without ionized carbons in the plume. Ions were filtered from the

plume by imposing a strong magnetic field (up to 8 kGauss) in the plasma path. Raman

spectroscopy and thickness measurement were carried out on two films deposited under

otherwise identical conditions, except the inclusion of ions for one sample. From the

thickness difference between these two DLC films (800 A with ions, and 700 A without),

a plasma ion fraction of -12% was estimated. Obvious similarity between the two Raman

spectra prompted the authors to conclude that the carbon atom electronic hybridizations

are essentially the same for the two DLC films. The authors also conducted TOF

experiments with a Faraday ion collector. They discovered that around 50% of the ejected

ions were singly charged monomers (C+), with the rest being clustered ions up to

hexamers and hepamers. However, some of the authors' claims warrant a more careful

look. It has been noticed by various other researchers that the Raman spectra of DLC

films with vastly different sp3/sp2 ratios can exhibit rather similar features (explained in

Chapter 4). Therefore without any other measurement of the film properties, such as their

electrical resistivity or mechanical hardness, it would only be plausible at best to assert that

ions are not a factor in affecting the DLC film formation.

In summary, along with ion beam depositions (IBD), pulsed laser deposition

(PLD) has established itself as one of the premium techniques in producing high quality,

hydrogen-free DLC films. The formation of diamond-like, tetrahedral sp3 bonds is

believed to be enhanced by the energetic plasma generated with high intensity laser beams.

With accommodating substrate conditions such as low temperature, sufficient thermal

quenching ability, DLC films with high hardness, high electrical resistivity, and a relatively

large optical bandgap can be condensed from small carbon particles with high kinetic

energies. The optimum kinetic energy, depending on the predominant species present in

the laser induced plume, is projected to be between several tens to several hundreds of

electronvolts. Higher laser intensity usually leads to the creation of carbon particles with

higher kinetic energies, and in most cases, better film qualities. But the advantage gained

from a more energized plasma is sometimes compromised by the addition of larger,

clustered carbon particles found at high laser intensities. Also, due to the nature of the

laser beam profile and the beam-solid interactions, a wide spread of the particle energy

distribution is often inevitable. This may also contribute to particulate formation on the

film surface from those particles that do not have sufficient energy to break up upon

impact. Though pulsed laser deposition of DLC film may be adequately described as an

energetic condensation process, whether it is also an ion assisted one remains somewhat

controversial. Moreover, there seems to exist a strong correlation between the film

quality and laser wavelength, at least for laser beams with nanosecond pulsewidth. Lasers

with shorter wavelengths in the UV range appeared to have produced more atomized

carbon species with higher energies, leading to films with better qualities. This is in

contrast to those lasers with wavelengths in the IR or near IR range, which seemed to

generate larger, less energetic carbon clusters. It thus requires higher intensities for long

wavelength lasers than short UV lasers to achieve similar plasma characteristics, otherwise

the resultant films tend to be more graphitic. This can be contributed to the fact that

absorption coefficients would increase as one moves to shorter wavelength, and the

penetration depth into the target material is correspondingly reduced. It is a favorable

situation as the laser energy is now consumed by thinner layers of the target, thereby the

thermal coupling between laser and solid will be more efficient. The stronger absorption

at shorter wavelengths is also believed to reduce the ablation density threshold. In

consideration of this intriguing but not yet well resolved phenomenon, a high "effective"

laser density, or volume laser intensity expressed in W/cm3, seems to be more appropriate

than the planar laser power density (W/cm2) in describing the deposition parameter that

favors the sp3 bond formation.

It should also be noted that little information is available at the present time, on the

subject of sub-nanosecond, high intensity laser deposition of DLC films. Nonetheless,

early evidence suggested that at very high intensities, the laser induced plasmas are likely

to have a higher degree of ionization and atomization, accompanied by high ion kinetic

energies up to the keV range. These are essentially the same ingredients that have proven

to produce high quality DLC films for nanosecond laser pulses.

Laser-Solid-Plasma Interactions

The success of PLD technique in making DLC films is determined by the physical

processes associated with the laser-target-plasma interactions.

Common features in the PLD technique include a laser generated plasma near the

target surface and the subsequent hydrodynamic expansion of the plasma in vacuum. This

high temperature, high pressure plasma contains different target carbon species, which

move toward the substrate with high speed in a strong forward-directed fashion. The

quality of the resultant DLC films is therefore a function of the energetic condensation of

these particles upon reaching the substrates.

When considered in the time domain, the PLD process can generally be divided

into three phases:

1. Interactions of the laser beam with the target material, resulting in heating/melting

vaporization of its surface layers;

2. Interactions of the evaporated target species with the incident laser beam, giving rise

to plasma heating and absorption;

3. Hydrodynamic plasma expansion in vacuum, leading to the characteristic nature of the

PLD process.

For simplicity, the first two phases are often considered to start with the onset of

the laser pulse and continue through the laser pulse duration, while the third phase

commences after the pulse terminates. In actuality, however, these phases are often not

separable due to substantial overlap amongst them. Since our investigation involves the

use of a Nd:YAG and a Ti:Sapphire laser system, the following discussion of the laser-

solid-plasma interactions will be limited to those generated by a single pulse of these nano-

and femtosecond lasers.

Many thermo-physical phenomena associated with the beam-solid-plasma

interactions are shared by these two types of laser pulses. However, due to each one's

unique pulse duration and thus different attainable laser intensity levels, certain aspects of

these three regimes could have distinctively different attributes.

The ensuing sections first address certain aspects of the laser-solid-plasma

interactions, including a brief discussion of the plasma hydrodynamic model used to

describe the evolution of plasma expansion. This will be followed by several examples of

studies having dealt with the plasma generation and expansion induced by these two

classes of lasers, in the intensity regimes that are relevant to DLC film depositions.

Emphasis is given to those instances where measurements (or simulations) of carbon ion

charge/kinetic energy distribution have been carried out as a function of laser intensity.

With a typical single pulse output, the laser-target-plasma interactions can be

summarized as follows:

At the onset of laser pulse incident on the surface, the target electrons immediately

absorb incident laser radiation and a thermal equilibrium state is reached amongst

themselves within several femtoseconds (-10"14 s). These electrons then start to couple

their energy to the lattice atoms, through electron-atom collisions. The lattice temperature

is expected to rise and become the same as that of the electrons within about one

picosecond (~10-12 s) [69]. As the surface layer is further heated, it will eventually melt

and vaporize, creating a collections of electrons, ions, neutrals, as well as Rydberg state

atoms in front of the target surface. In another word, a plasma is generated. There is no

absolute definition of a plasma, but it can best be interpreted as a physical state of matter

with high electrical conductivity and largely gaseous mechanical properties.

The initial plasma temperature and charge density, represented by those of the

electrons, are relatively low. The plasma is thereby transparent to the incident laser beam.

Electron-neutral inverse bremsstrahlung dictates the absorption of laser energy and the

plasma temperature will rise rapidly as a function of time. As the plasma becomes more

ionized the electron-ion inverse bremsstrahlung takes over as the main absorption

mechanism. Saha equations can then be used to predict the plasma ionization ratio. At

sufficiently high temperatures (10000 C or higher), a fully ionized plasma will develop.

Here "fully ionized" refers to ions in the plasma with different charge states and not

necessarily fully stripped ions.

With successively higher electron and ion densities, this hot plasma progresses to

become more absorbing to the laser light, thus preventing the trailing laser pulse from

reaching the solid target. In fact, depending on the laser wavelength and its intensity, the

incident laser beam could be totally reflected at the so-called "critical surface", before the

peak of the laser pulse arrives. This critical surface is defined as the plasma layer where

the electron density reaches a threshold. The critical electron density n. can be calculated,

in cgs units, from the equation that describes a uniform, collision-less plasma with an

unperturbed electron density:

co = cop = (47te2nC/m)0.5 (2-1)

where me is the electron mass, e the electron charge, co the laser frequency, and op the

plasma frequency. For a frequency-doubled Nd:YAG laser of 532 nm wavelength, co =

3.54 x 1015 s-1, and n< = 3.9 x 1021/cm3. For a Ti:Sapphire pulse of 780 nm wavelength, co

= 2.41 x 1015 s-1 and n, = 1.8 x 1021/cm3. At the critical surface, the magnitude of the

complex plasma index of refraction decreases from nearly one to zero, yielding a

reflectivity of one.

Absorption of the laser energy in the plasma is a nonlinear process. It increases as

square of the electron density and cube of the ion charge state. The high electron

temperature generated by this inverse bremsstrahlung induces large thermal pressure

gradients within the plasma. These steep thermal gradients will lead to a rapid expansion,

or "blow-off' of the plasma in the direction largely perpendicular to the target surface,

thus converting plasma thermal energy into expansion energy of the electrons and ions.

Since an electron has negligible mass compared to that of an ion, all the expansion energy

can be considered to have been converted into ion kinetic energies. As a result of this

expansion, the plasma density will quickly drop below the critical density. If this happens

before the laser pulse terminates, any remaining laser pulse will again be able to reach the

target surface. Further heating of the target is also provided through radiation loss from

the plasma and via heat conduction in the solid. The ablated target species in the

expanding plasma will recombine and cool, through radiative and three-body

recombination processes. The degree of recombination and cooling depends on initial

plasma conditions and background pressure. At pressures lower than about 104 Torr, any

recombination between the plasma particles and ambient molecules can be considered


The phenomena discussed above are a series of dynamic processes. A successful

modeling of these processes is difficult. This is because a plasma usually contains a large

number of charged particles and each one of them generates its own electromagnetic field.

The motion of each particle is in turn modified by this field. To fully describe the state of

a plasma, each particle would require six equations to calculate its motion

(position/velocity) and associated electromagnetic field. This makes a complete dynamic

description of the above sequence of events practically impossible.

Historically, studies of laser induced plasma have instead treated it as a fluid: the

individual motion equation of charged particles is averaged to give a distribution function

of the plasma. This phase space distribution function f (x, u, t), which gives the plasma

density at spatial position x and time t with velocities between u and u + 5u, can therefore

characterize the plasma as a fluid and its collective behavior.

When dealing with a carbon plasma, a two-fluid model, one for the electrons and

one for the carbon ions, is usually considered adequate. By coupling the two-fluid model

equations with the Maxwell's equations, a complete description of the carbon plasma

evolution can be obtained, for given initial conditions.

The Maxwell's equations relate the electric and magnetic fields to the charge and

current density of the plasma, in cgs units:

V E = 47rp (2-2)

V B=0 (2-3)

1 8B
VXE= (2-4)
c Ot

VXB= -4J+ OE(2-5)
c c 8t

where E and B are the electric and magnetic fields, respectively, p = nq (n is the plasma

density, q the particle charge) is the plasma charge density, J = Z nqu (u is the mean fluid

velocity) is the plasma current density, and c is the light velocity.

The logarithm defining the plasma as a fluid includes three basic equations. They

are derived by taking moments with respect to velocity of the Vlasov equation. These

include the Euler's equation of motion (conservation of momentum), the equation of

continuity (conservation of mass), and the equation of state isothermall or adiabatic

compressibility) [70]:
,u eu nq ux B 1 Op
n( + u ) (E + ) (2-6)
Ot 9x m c m 1x

S+ -0 (nu) = 0 (2-7)

p = nO isothermall] (2-8)

r constant [adiabatic] (2-9)

here m is the particle mass, p is the plasma thermal pressure and u is the fluid velocity.

The plasma pressure p is related to its density by equation of state (2-8) or (2-9),

depending on the condition of the charged fluid under consideration. The isothermal

equation (2-8) applies when 0)/k << v, where co and k are the characteristic frequency and

wavenumber (cm1) of the process being considered, and v is the particle thermal velocity.

In this case, p = nO, where temperature 0 is a constant. This condition is generally

assumed during the laser pulse span, when heat flow is so fast that the temperature of the

charged fluid remains constant. On the other hand, when o/k >> v, the adiabatic equation

(2-9) applies. An adiabatic process occurs when the fluid has no exchange of heat or

energy with other medium outside. The plasma expansion after the laser pulse termination

is usually considered to satisfy the adiabatic condition. The plasma pressure is then related

to its density through equation (2-9), where ui = 1/T is the effective electron-ion collision

frequency (r is the mean collision time), y = c,/cy = (F + 2)/F, with cp/cy being the ratio of

specific heat, and F being the number of degrees of freedom. For a fully ionized plasma F

= 3, which corresponds to a y vaule of 5/3.

The simulation of a spherical laser plasma expanding into a vacuum typically relies

on self-similarity solutions to the hydrodynamic equations considered above. Self-

similarity flow is defined as one for which the expansion scale-lengths, therefore the

density profiles, are considered to be independent of time. The expansion velocity profile

is regarded as linear, increasing with distance from laser focus to the plasma front edge.

Also, the plasma is treated in both the spatial and time domains as if it were originated

from a point source. Fader [71] has carried out a complete hydrodynamic calculation of a

spherical plasma expanding into vacuum. He demonstrated that some time after the

plasma ignition, its radial velocity profile v (r, t > 0) always becomes linear, the ion density

n (r, t > 0) invariably assumes a Gaussian profile, and its temperature drops adiabatically.

This was found to be true regardless of their respective initial conditions. The fact that the

plasma density and velocity remain similar in the time limit t << t < oo (t being the laser

pulse duration) is the reason for using self-similar solutions to describe the plasma


When solving the hydrodynamic equations of the proceeding sections, the laser

energy transfer to the plasma is usually assumed to be fully symmetric. The energy

transfer and equilibration time is considered to be instantaneous when compared to the

laser pulse duration. These assumptions are justified when the laser intensity is not

exceedingly high or the pulse extremely short. Potential modeling inadequacies associated

with the use of ultrashort femtosecond laser beams will be addressed later in this chapter.

Nanosecond Laser Plasma

For nanosecond laser-solid interactions, Singh [72] has developed a model to

describe the thermo-physical phenomena associated with the plasma expansion. This

model treated the laser instigated plasma as a high-temperature, high-pressure gaseous

phase that is initially confined to small dimensions and then allowed to expand freely in

vacuum. Gas dynamic equations were set up to simulate expansion velocity of the plasma.

With intensity level in the 109 1011 W/cm2 range considered for a nanosecond

Nd:YAG laser pulse, initiation of the beam-solid interactions is believed to be largely a

thermionic emission process. Removal of the target material by laser irradiation depends

on beam-solid coupling, which leads to melting and/or vaporization of the surface layers.

Energy dissipation is carried out exclusively through conduction in the solid. Radial

transport can be neglected as long as the laser beam is relatively well focused. Properties

of the target material, such as its optical reflectivity, absorption coefficient, thermal

conductivity and specific heat are the principal parameters defining this process. The

evaporation attributes as a function of laser and material parameters can be modeled with

the solution of a one-dimensional heat flow equation, constrained with appropriate

boundary conditions. The energy deposited onto the target, by simple heat balance

consideration, should be equal to the energy needed to vaporize the surface layers plus the

conduction loss by the substrate and the absorption by the plasma. This yields:

Ad = (1 R)(E Eth)/(AH CvAT) (2-10)

where Ad, R, AH, Cv, AT, and E are the evaporated surface thickness, reflectivity, latent

heat, volume heat capacity, temperature rise, and laser energy density (J/cm2),

respectively. Eth represents the energy density threshold above which evaporation is

observed. Nonlinear changes in Ad may be observed due to self-induced variations of the

coupling parameters used in the above equation.

After the surface layers reach the melting point or sublimation temperature, a

transient plasma is generated above the target, while the liquid-solid interface penetrates

into the bulk. The plasma absorption coefficient a (cm'1) can be expressed as [73]:

a = (3.69 x 10')(ZWN2/T05v3)[1 exp(-hv/kT)] (2-11)

where Z, N, T, v, h, and k are the average ion charge, plasma density, plasma temperature,

laser frequency, Planck constant and Boltzmann constant, respectively. This equation

predicts stronger plasma absorption at longer laser wavelength. Numerically, it would

require a relatively high plasma density on the order of 1019/cm3 or higher for a significant

absorption to occur. At laser intensities considered in our experiment, this density level

should be easily achieved. The 1 exp(-hv/kT) term represents the loss due to stimulated


Based on the above discussion, a schematic of the laser-solid-plasma interactions

can be shown in Figure 2-3 (dimensions not drawn to scale). When considered in the

spatial domain, three separate regimes can be distinguished:

I. Unaffected bulk material;

II. Melting/evaporating target surface;

III. Laser induced plasma, which is highly absorbing at near the target surface and

becomes transparent at the fast expanding edge.

1-D Kundsen Coronal | Electrostatic
flow layer Y region Y acceleration

Laser Beam

Solid Liquid Plasma

Fig. 2-3 Schematic of laser-solid-plasma interactions.

The laser plasma region III itself consists of several overlapping sub-regions
marked as A, B, C, and D in Figure 2-3. Region A, whose dimension is much smaller than
the laser focal spot, is generally treated as a highly absorbing, one-dimensional steady state
fluid. Depending on laser intensity and wavelength, this layer could become reflective to
the incident beam if its electron density exceeds that of the critical density. Region B is
the so-called Knudsen layer resulting from the three-dimensional expansion of layer A. It
is a highly collisional regime with an explicit center-of-mass velocity, leading to a beam-
like plasma expansion [74]. This regime also sees a reduced plasma temperature and
density. Further plasma expansion into Region C yields the coronal region where
collisions among electrons and ions become less frequent. The hot electrons created in the
previous regions could then decouple from the Boltzman-Maxwellian distribution and
deviate from local thermal equilibrium (LTE). A small number of these electrons may
actually escape from the main plasma, due to the lack of energy-exchanging collisions. In
Region D, these escaped electrons can create a high electrical field which is much higher

than that represented by the plasma temperature kTe. This electrostatic field in turn

accelerates ions in the plume to high kinetic energies. In fact, due to even lower electron-

ion collision frequencies in this region, the electrons and ions may form an oscillating

cloud while they continue to expand in vacuum, during which process the ions can acquire

successively higher energies. These ions will reach an asymptotic energy distribution

when all the electron energies have been captured. Keep in mind that when a nanosecond

laser pulse is considered, only Regions A and B may have sufficient time to form during

the span of the laser pulse, while Regions C and D are not likely to develop well after the

pulse has terminated.

During the laser pulse span, it is often assumed that the energy absorbed by the

plasma is distributed uniformly over its entire mass. This assumption is validated by the

fact that the plasma thermalization time (on the order of one picosecond) is much less than

the time it takes the plasma to expand to any significant dimensions. Consequently, a

uniform temperature in the plasma can be established. The plasma in this regime is

therefore at an isothermal state. It will continuously absorb laser radiation and expand


Rapid expansion of the plasma in vacuum results from large thermal pressure

gradients. This isothermally expanding plasma can be treated as a high-temperature, high-

pressure ideal gas. By applying gas dynamics, the expansion can be simulated by its

density, pressure, and velocity profiles as shown in Figure 2-4 [72].

The plasma density can be expressed as a Gaussian function for t < T, T being the

laser pulse duration:
Nt x2 y2 z2
n(xyzt) r (t)Y(t)Z(t) 2X(t)2 2Y(t)2 2Z(t) [t <']


here n(x,y,z,t) is the plasma density at any point and time, Nt is the total number of plasma

particles at the end of the laser pulse (t = t), and X(t), Y(t), Z(t) are dimensions of the


plasma in three orthogonal directions and correspond to distance at which point the
plasma density is 60 65% of its maximum value. Since the plasma is being treated as an
ideal gas, the plasma temperature at any point is then related to its density by:

P = n(x,y,z,t)kTo

[t < t]]


where To is the isothermal temperature of the plasma.
Dawson [75] has argued that in order to maintain a Gaussian density profile, the
plasma expanding velocity needs to be proportional to the distance from the target
surface. A self-similarity expression for the spatial and temporal dependence of the
velocity can be expressed as:

x dX y dY. z dZ
X() dt Y(t) dt Z(t) di




Velocity, V (x, y, z, t)

.Density n (x, y, z, t)
Pressure P (x, y, z, t)

Distance in x direction

(X, Y, Z)
Plasma edge

Fig. 2-4 Schematic profile of plasma density, pressure, and velocity gradients
perpendicular to the target surface [72].






where dX/dt, dY/dt, and dZ/dt are plasma expansion velocities at positions X, Y and Z,

respectively. This equation manifests that the inner edge of the plasma contains a

maximum of density and a minimum of velocity.

Since the equation of gas dynamics governing the expansion of plasma is

essentially the equation of continuity and motion, a solution for the plasma expansion can

be computed by substituting the expression for density, velocity and pressure into the

continuity and momentum equations discussed in the earlier sections. This yields:

I dX d2X I dY d2Y I dZ d2Z kT
X(t)[ + -] = Y(t)[- + Z(t)[-- + [t ]
t dt dt tdt dt2 dt dt2 M

The above equation determines the expansion of the three orthogonal plasma

edges, with the initial plasma dimensions on the order of tens to hundreds of microns in

the transverse direction and less than one micron in the direction normal to the target

surface. It also suggests that the plasma has a low velocity and high acceleration at the

initial expansion stages. But as velocity increases, acceleration will start to diminish and

eventually drop to zero, giving rise to an elongated plasma shape.

Since no laser-plasma interaction exists after the laser pulse terminates, an

adiabatic plasma expansion will take place, where the plasma temperature is related to its

dimension by the adiabatic thermodynamic equation:

T[X(t), Y(t), Z(t)] -1 = Constant [t > T] (2-16)

where y = cp/c, is the ratio of specific heat capacities at constant pressure and volume.

The thermal energy is then rapidly converted into ion kinetic energy with the fast

expanding plasma. Depending on the value of y, the maximum attainable expansion

velocity is usually several times that of sound. In the adiabatic regime, the plasma density,

pressure, and velocity can be expressed similarly as those in the previous isothermal

regime by neglecting the t/T term which takes into account the injection of particles into

the plasma. In addition, a solution that controls the plasma expansion can be obtained by

substitution of density, pressure, and velocity into the differential equations of energy and

temperature in the adiabatic regime. This yields:

d2X d2Y d2Z kT XoYoZ
X( dt- = (t) = Z(t)[- ] = -M X(t)Y(t)Z(t) [t >]


where Xo, Yo, and Zo denote the initial orthogonal dimensions of the plasma after pulse

termination. This equation indicates that acceleration of the plasma depends on its

temperature and dimension, as well as the mass of the ablated species.

The initial transverse Y and Z dimensions are much larger than the perpendicular X

dimension, which represents the expansion length in the isothermal regime. Therefore the

highest velocity is expected in the X direction as the velocity is dictated by these lengths

shown in the above equation. This leads to the characteristic plasma shape elongated

outward from the target surface. The above equation also implies that if the plasma is

initially longer in Y than in Z direction, the plasma will be accelerated more in the latter

direction. As the plasma expands, most of the thermal energy is converted into the kinetic

energy, thereby no more energy is left for further expansion. The plasma becomes

elongated in the shorter X direction and retains its profile until it is intercepted by the


The same approach can be applied to all species, including ions, atoms, and small

molecular clusters. Since the above model is based on the equations of fluid flow, and the

expansion velocity is controlled by pressure gradients in the plasma.

In addition to a theoretical model of the hydrodynamic plasma expansion as

discussed above, the following sections explain by example some of the numerical

descriptions of this process. The parameters used in these examples are more closely

matched with those used in our investigation.

Puell [76] has developed a hydrodynamic model for the plasma generated by an

intense laser beam focused on a plane, solid target. The laser light used in his modeling

was a nanosecond (7 ns FWHM) ruby pulse at 694 nm wavelength. The plasma

production was treated as a steady state process. Taking into account the finite focal spot

size, he also divided the plasma into three separate regions:

1. Undisturbed solid target;

2. A zone extending from the target surface out to a distance of the focal spot radius R.

In this so-called heating zone, a one-dimensional plasma flow was assumed;

3. A zone beyond R where a three-dimensional plasma was allowed to show a lateral


By considering an energy balance between the incident laser beam and the thermal

and kinetic energy transported by the plasma flow, Puell was able to quantitatively

establish a relationship between the highest plasma (electron) temperature kT, the total

number of particles N generated during the laser-solid interactions, and the ion kinetic

energy E as a function of the incident laser intensity I. They are summarized as follows (in

cgs units):
kT = a2(3MCR/50)2'9I4/9 (2-18)

N = xrR2(a-3(2MCR/3)29 f (I / 5)5/9dt (2-19)

E = 5ZakT = 5Za'3(3MCR/50)27914/9 (2-20)

where M is the ion mass, k is the Boltzmann constant, R is the focal spot radius, Z is the

average ion charge, a = (Z + 1)/Z, and C = 2.55 x 10-55 is a constant associated with the

absorption coefficient for inverse bremsstrahlung. When laser pulse other than that of
Ruby is considered, C = 2.55 x 10"55 (COR/C0)2, where co is the frequency of light under

consideration, and C0R = 2.7 x 1015 s-1 is the frequency of the Ruby laser. These equations

apply to the layer within a distance of R from the target surface.

Heat conduction and radiation losses in the plasma are neglected in this model.

Also, there exist two pre-conditions under which the above equations are derived from.

The first one being the plasma stays largely transparent to the laser beam throughout its

pulse duration, i.e., the validity of the above equations holds only if the electron density

stays below the critical value of 2.3 x 1021/cm3 for Ruby laser. The critical laser intensity

associated with this electron density is calculated to be 8 x 1013 W/cm2. The second

condition is that the plasma can be treated as a stationary flow. This assumption is

satisfied by the fact that the time an electron-ion pair spends within the heating zone is

much smaller than the nanosecond laser pulse duration. Under this steady state model, the

ion kinetic energy is postulated to be equal to the laser intensity divided by the ion flux

leaving the target surface. This assumption is based on the argument that during the

heating of the plasma, the absorbed energy is partitioned between thermal and kinetic

energy of the particles. At distances far away from the heating zone, however, all of the

energy is transformed into ion kinetic energy due to the plasma expansion.

In a subsequent experiment, Puell et al. [77] used the same ruby laser to ablate

lithium deuteride (LiD) and carbon (C) targets in vacuum. The laser intensities used were

between 2 x 1011 and 5 x 1012 W/cm2. Both the electron temperature and the ion kinetic

energies of these two target materials were found to increase almost linearly as a function

of increasing laser intensity. The electron temperature To of the carbon plasma was

measured to be in the 80 330 eV range at an angle of 70 off the target normal, and the

average ion kinetic energy E varied from 2 to 13 keV.

Both Te and E in Puell's experiment are somewhat higher than those observed by

some other researchers working in the same area. These include the results from

Ambartsumyan [78], Haught [79], and Demtroder [80], where the plasma temperatures

were found to be not higher than a few tens of electronvolts, and ions attained asymptotic

(t -> oo) velocities on the order of 106 107 cm/s. These velocities correspond to kinetic

energies ranging from several hundred to a few thousand electronvolts.

One possible explanation to the discrepancies among these results is that in Puell's

case, a higher maximum laser intensity level was used on target materials with lower

atomic weights. This could lead to higher initial electron temperatures and ion kinetic

energies. Another possibility lies in the adoption of a one-dimensional heating zone

geometry in Puell's model, where calculations are based on the assumption that the plasma

in this heating zone absorbs all the laser energy and then adiabatically expands. In reality,

this plasma layer would have expanded in a three-dimensional fashion before the laser

pulse terminates. For example, with an ion velocity of 1.5 x 107 cm/s, it travels in 7 ns

through a distance of mm, much larger than R 50 jpm used in Puell's model. Thereby

only a fraction of the total energy carried by one laser pulse is actually consumed in the

heating of this plasma zone, and a lower plasma temperature should be expected. In

addition, since the inverse bremsstrahlung plasma heating is proportional to the square of

the electron density, a rapid plasma expansion would effectively lower the plasma density

and hence its temperature. Finally, recombination and cooling effects may be responsible

for the lower electron temperatures observed in these other experiments [78-80], since

their data were taken at distances further away from the ablation source.

In the experiments carried out by Demtroder and Jantz [80], a ruby giant laser

pulse (30 ns FWHM), similar to the one used by Puell [76] was focused onto aluminum

and copper targets in vacuum to generate plasmas. By combining the time-of-flight (TOF)

and retarding potential measurements, the authors were able to detect the Cu and Al ion

charge state as well as their kinetic energies, as a function of laser intensities in the 1010 -

1011 W/cm2 range. They found that the kinetic energy of the ions was proportional to

their charge and about 100 times larger than their thermal energy. At a laser intensity of

3.3 x 10" W/cm2 and 1.7 meters away from the target, Cu2+ ions were determined be the

predominant species in the laser induced plume. Copper ions with up to 6-fold charge

state were detected in the plasma. The median kinetic energy corresponding to the peak

of the TOF spectrum was calculated for Cu' ions to be 200 eV, and close to 2000 eV

for Cu" ions. An upper limit of 11 eV was detected for the electron temperature at this

intensity level. The maximum Al ion kinetic energy, defined by the onset of the TOF

signal, was found to increase from 500 to 2000 eV for laser intensities ranging from 2 x

1010 to 2.5 x 101 W/cm2. The disparity between the large ion kinetic energy and the low

thermal energy was explained with a simple model proposed by these researchers: an

oscillating electric space charge field transferred electron thermal energy to radial kinetic

ion energy. In their hypothesis, a fast electron leaving the plasma cloud is electrostatically

attracted back and forth, each oscillation sees a fraction of its energy being transferred to

the ions. At a laser intensity of 3.3 x 10" W/cm2, up to keV ion kinetic energy is

achievable if the electrons oscillate at a frequency of 2 x 1014 s"'. The final ion kinetic

energy E can be expressed as:

E=(nZ + 1)kT (2-21)

where Z is the average ion charge, T is the plasma temperature (same as ion or electron

temperature), and n is a constant that describes the average number of times an electron

loses and regains its thermal energy.

Based on Puell's theory [76], Stevefelt and Collins [81] carried out a computer

simulation of the expansion of a carbon plasma generated by a focused Nd:YAG laser

beam. Their model considered the hydrodynamic plasma expansion in vacuum after the

initial ignition by the intense laser beam. Ion charge states and corresponding kinetic

energies at a distance up to several centimeters away from the target surface were

computed. These parameters are of more interest than the initial conditions predicted by

Puell's model, since they more closely resemble the conditions under which laser

depositions of DLC films are actually carried out.

The processes included in their modeling were adiabatic cooling of the plasma,

electron-ion three-body and radiative recombinations. Three hydrodynamic equations

were used to describe the plasma flow:

-d(Nvr2) = 0 [mass density conservation] (2-22)

d 5 Mv2
[- (1 + Z)kT + ] = [laser energy conservation] (2-23)
dr 2 2 v

dT 2T dN 2Q
dT 2T =2Q [first law of thermodynamics] (2-24)
dr 3N dr 3v

where r is the distance from the target surface, N the ion density, v the ion velocity, T the

plasma temperature, M the ion mass, and Q the laser energy delivered to the electrons per

ion per unit time. The above equations were then coupled to the recombination kinetic

equations to obtain the evolution of the carbon ions with various charge states, as a

function of distance from the target and laser intensity:

Cz+ + e7 + e o> C(z-"1 + e- [three-body recombination] (2-25)

Cz+ + e" => C(z+ + hv radiativee recombination] (2-26)

Shown in Figure 2-5 and Figure 2-6, respectively, are plots adopted from the

authors' paper [81]. These two figures illustrate the simulated ion concentration as a
function of distance from the surface at a laser intensity of 5 x 1010 W/cm2, and the ion

fractional concentration as a function of laser intensity.

It is demonstrated in Figure 2-5 that, for a distance r larger than about 0.1 cm from

the target, the relative carbon ion concentration largely remains the same. This so-called

"ion freezing" phenomenon was also observed earlier by several other research groups

[82,83]. In this region, three-body recombination is the dominant process, and the

recombination rate 0 approximately scales as P oc r-"1. The electron density decreases as a
function of rf2, while the ion velocity and charge assume asymptotic values. The authors

also argued that because of the recombination processes, the plasma cooling should be
substantially slower than a simple adiabatic expansion. Figure 2-6 shows that even at a

distance of a few centimeters away from the target, multiple charged C4' and C3" are still
the most predominant species in the plume. Using the kinetic energy equation developed



o IE+16


0- IE+14

IE+ 10 --- I F I I it+EIu IE+11
0.001 0.01 0.1 1 10
Distance from target (cm) Laser Intensity (W/cm2)

Fig. 2-5 Ion concentration as a function
of distance from the graphite target, at 5
x 01o0 W/cm2 [81].

Fig. 2-6 Fraction of various carbon ions
characterized by 'frozen ionization' as a
function of laser intensity [81].

by Puell [76], with Z = 4, R = 15 pm, and a laser intensity of 1.2 x 10" W/cm2, an ion

kinetic energy of 1000 eV can be estimated. Even higher ion energies are expected at

greater laser intensities. The authors used these results to suggest that diamond-like

carbon films are the product of multiple charged plasma ions with high kinetic energies.

Femtosecond Laser Plasma

One apparent difference between the laser-solid interactions induced with

nanosecond and femtosecond lasers is that much higher intensity levels can be obtained

with the short femtosecond pulses. With near-UV excimer and Q-switched Nd:YAG

lasers most commonly used in PLD of thin film materials, the highest attainable laser

intensity is usually in the 10" W/cm2 range. On the other hand, a laser pulse with a 50 fs

pulsewidth and 100 mJ energy can produce a laser intensity up to 101 W/cm2 when

focused. Laser intensities in the 1014 1016 W/cm2 range can be easily achieved with a

CPA Ti:Sapphire system, such as the one used in our study.

At laser intensities greater than about 1011 W/cm2, atoms and molecules become

unstable. The oscillating electric field E (V/m) is inherently related to the laser intensity I

(W/m2) by [84]:

I = 2n(6-)O5IEl2= o EI2 (2-27)

where n is the refractive index, co = 8.85 x 10-12 F/m, 4n = 4t x 10-7 H/m, and Zo = 377

Q. With a laser intensity of 1014 1016 W/cm2, an electric field on the order of 108 109

V/cm can be induced on the target surface. This value is comparable to the Coulomb field
binding the carbon electrons to their atomic cores. As a result, an intense femtosecond

laser pulse can ionize a carbon atom through non-thermal processes, even when the

photon energy is much less than the atom's ionization potential. Multiphoton ionization

(MPI) and tunneling ionization (TI) are two of the possible mechanisms. These two

ionization mechanisms are closely related to each other.

MPI is a nonlinear material response to a strong laser field, where an atom with an

ionization potential 9 ((p > hv, hv is the laser energy) simultaneously absorbs n photons

and releases an electron. In this process, n equals the integer part of (p/(hv + 1). When n

>> 1, the TI mechanism better describes the nonlinear ionization process: the ac laser field

greatly distorts the atomic Coulomb field so that a bound electron has a high possibility of

escaping by quantum tunneling through the Coulomb potential well during half an optical

cycle, near the maximum of the optical field. The transition from MPI to TI can be

characterized by the so-called Keldysh or adiabatic parameter u:

87r2mv2q 05,
u= ( e2E2 ) (2-28)

When u > 1, multiphoton ionization process applies, otherwise tunneling ionization

is more appropriate in describing the ionization process.

Further ignition of the plasma by an intense laser field is believed to be an electron

avalanche process. It is initiated by a small number of free electrons floating in space

close to the target surface. These electrons, which may originate from laser induced

multiphoton ionization, can gain enough energy from the laser light to impact ionize an

atom in collision. Repetition of this process leads to a rapid multiplication of electron, i.e.,

an electron avalanche process is said to have started. These electron-ions (atom)

collisions, generally termed as inverse bremsstrahlung, may be accompanied by cascade

ionization to further increase the plasma ionization ratio. Free-free transitions between the

plasma electron and incident light cause intense heating of the plasma, followed by a rapid

hydrodynamic expansion of the plasma in the form of a shock wave.

The above sequence of events play out differently for a short Ti:Sapphire laser

pulse (- 10'13 s), when compared with a relatively long Nd:YAG pulse (-10-9 s).

For a nanosecond laser pulse, when the peak of the pulse reaches the plasma, its

density gradient at the solid-vacuum interface would have already relaxed to a density

scale-length defined as:

d = n.( )- = v,(/2) (2-29)

where n. is the electron density, is the laser pulse duration, and v. = (ZkTe/M)05 is the

plasma acoustic velocity. For a 7 ns Nd:YAG pulse with a laser intensity of 1010 10"

W/cm2, assuming Z = 4 and kTe = 10 eV, v, is calculated to be on the order of 106 cm/s.

Its density scale-length d thereby is at least 10000 nm, which is much longer than the laser

wavelength of 532 nm. Under this situation, the laser light will propagate in a long

underdense plasma up to the critical density surface, where electron density equals 3.9 x

10 21/cm3, and be reflected. Beyond this critical density layer the laser field can only

tunnel (decay exponentially) through a distance up to the skin depth defined as:

5 = c/Op (2-30)

where c is light speed, cop is the plasma frequency. For a Nd:YAG laser with a cop of 3.54 x

101 s"', 5 is estimated to be -135 A. This skin depth is much less than the laser

wavelength of 532 nm and the density scale-length, therefore no energy carried by the

trailing laser pulse is directly deposited into the solid surface. Rather, this energy is

absorbed by the plasma to further elevate its temperature.

On the other hand, when an ultrashort Ti:Sapphire laser pulse hits the target, a

distinctively different plasma will be created. This plasma would have an electron

temperature up to at least a few hundred electronvolts and a plasma density on the order

of several 1023/cm3, both of which are much higher than those induced with longer

nanosecond pulses at lower intensities. This plasma density is now comparable to the

solid density of graphite (ng = 1.13 x 1023/cm3), i.e., a solid density plasma is created. The

target lattice, or more precisely the target ions, however, will remain relatively cold during

the span of the femtosecond pulse. This is because the electron-ion equilibrium time is on

the order of about one picosecond. As a result, a two-temperature plasma with "hot"

electrons and "cold" ions is formed. Thermal equilibrium among themselves is only

achieved after the pulse has terminated. In the meantime, this solid density plasma

undergoes little expansion because of the extremely short pulse duration. The plasma-

vacuum interface remains relatively sharp when the pulse peak arrives. For instance, when

a 100 fs pulse strikes a carbon target at laser intensities of 1014 1016 W/cm2, assuming Z

= 6 and Te = 200 eV, v. is calculated to be on the order of 107 cm/s. The density scale-

length d is thus on the order of several tens of angstroms, similar to the laser skin depth

and much less than the laser wavelength of 780 nm. This would allow a significant

amount of laser energy to be deposited at densities much higher than the critical density,

which is calculated to be 1.8 x 10 21/cm3 for the 780 nm Ti:Sapphire laser light.

Figure 2-7 illustrates the differences between the plasmas induced with a 100 fs

and a 7 ns laser pulse.

It would be interesting to compare the plasma ion kinetic energy and charge state

induced by these two types of laser pulses, on a time scale relevant to laser deposition of

thin film materials. However, little information is available at the present time on

femtosecond laser induced plasma expansion. Most research efforts have so far been

concentrated on the high field physics associated with this short scale-length, high-density

plasma. Time scales of interest for these topics are usually within a few picoseconds after

the laser onset, and ion kinetics seems to be of less importance.

As mentioned earlier, within the first picosecond or so after the laser incident, the

femtosecond laser generated plasma has two distinctive temperatures: one for the "hot"

electrons and one for the "cold" ions. Temperature, meanwhile, is a representation of the

average kinetic energy of a plasma constituent in thermal equilibrium. A two temperature

plasma is hence nonthermal, and thermal equilibrium is only achieved after the pulse

termination. During this non-thermal span, no existing model can sufficiently describe the

100 Femtosecond Laser Pulse 7 Nanosecond Laser Pulse

Ne >1023/cm3 Ne ~1020-1021/cm3

Laser Laser

-100A 1 mm

Solid Density Plasma Long Density Scale-length Plasma
High Electron Temperature Strong Plasma Absorption
High Ion Charge Medium Electron Temperature and Ion Charge

Fig. 2-7 Differences between femtosecond and nanosecond laser induced plasmas.

plasma kinetics, such as the electron temperature or ion state evolution inside the plasma.
On a longer time scale, however, both the nanosecond and femtosecond plasmas can be

treated similarly as in local thermodynamic equilibrium, or LTE. LTE refers to the

situation where the average plasma thermal energy is a function of space, and yet the ions

kinetic energy has a Boltzmann-Maxwellian like distribution. LTE is a valid assumption
only if the collisional process dominates the radiative process, and if the

electrontemperature does not change rapidly on time scale of the inverse ionization or

recombination rates. It has been demonstrated that regardless of the laser pulsewidth,
LTE can adequately estimate the plasma ion charge states as long as the electron densities
are higher than 1017 1018 /cm3, and electron temperature does not exceed a few hundred

electronvolts [85]. Saha equation can then be equally applied to both nanosecond and
femtosecond plasmas considered in our study, to predict the ion density with different

charge states:

N 2Zj (T) 23 mekT exp( ) (2-31)
Nj Zj (T)Neh kT

here Zj(T), Zj+i(T) are the partition functions of ion state j and j + 1, respectively. Uj is the
ionization potential of ion state i, Ne is the electron density, and Te is the electron

temperature. The partition function Zj(T) is defined as:

Z,(T) = g,, exp(-U, / kT) (2-32)

where gij is the statistical weight of quantum number i in the ion state j, and Uiy is the
excitation energy of that state.
Liu [86] has studied the laser-plasma interactions induced by a 400 fs (FWHM)

pulse on an aluminum target, at laser intensities in the 1015 1017 W/cm2 range. A
Ti:Sapphire laser at 527 nm wavelength was used in his investigation.
For these femtosecond laser pulses, where the gradients in the transverse directions

(determined by the laser focal spot) are much less than that in the direction normal to the
target surface (defined by the density scale-length), one-dimensional hydrodynamic

equations were developed to describe the evolution of the plasma:

Dp du
Dt + p u = 0 [mass conservation] (2-33)

Du 1 0 p, f
S+ [momentum conservation] (2-34)
Dt p Ox p

DE + [energy conservation] (2-35)
Dt p Ox p

The above equations are written in the Lagrangean coordinates. The notations are
defined as: p is the mass density, u is the mean fluid velocity, ph is the thermal pressure, f
is the pressure force density, E is the internal energy per unit mass, and Q is the total heat
absorbed by the plasma.

These equations were then coupled to a LTE model to estimate the initial

ionization degree and electron temperature. For the aluminum target, an initial electron

temperature of 100 200 eV with an average ion charge Z = 11 (Al"h) was estimated at a

laser intensity of 1 x 1016 W/cm2. The ionization potential for the last L shell electron in

Al is 442 eV, similar to that of the last K shell electron of carbon (490 eV for C6+). We

thereby have adopted kTe = 200 eV and Z = 6 for the carbon plasma in our study,

considering C has a lower atomic weight than Al yet the laser intensities used in our study

are also somewhat lower.

Often treated as an ideal gas, the thermal pressure of a laser plasmais related to the

plasma temperature by ph, = nekTo + nkTi = (1 + Z)nikTe, assuming Te = Ti. As elucidated

earlier, a femtosecond laser plasma would have much higher density than that of a

nanosecond plasma, in addition to the expected higher kTe and Z. This should lead to a

much higher initial thermal pressure and correspondingly higher pressure gradients. With

laser intensities up to 1016 W/cm2, a thermal pressure up to several megabars can be

expected. It then seems logical to conclude that a femtosecond plasma will have a higher

expansion energy, i.e., the ions would have higher kinetic energies. This seemingly

plausible assumption, however, does not necessarily have a bearing upon closer

examinations. The reason is that at lower laser intensities such as those induced by the

nanosecond Nd:YAG pulses, the plasma expansion is determined solely by the thermal

pressure force. In this free expansion situation the plasma velocity is related to the sound

speed of the plasma, which is in turn given by the electron temperature. However, at the

intensity levels associated with ultrashort femtosecond pulses, the pressure induced by the

laser field itself can no longer be neglected. This laser force acting upon the plasma,

which is commonly termed as the ponderomotive force Fp, is related to the gradients of the

laser electric field E by [70]:

F 2 VE2 (2-36)
4 4m

where co is the laser frequency.

The ponderomotive force Fp is a nonlinear force induced by the gradients in the

electric field, resulting from plasma wavebeating with the laser light wave. Since the laser

field has a positive slope at the critical surface, the net effect of this ponderomotive force

is thereby to counteract the thermal pressure force. Fp acts on the plasma and forces it to

pick up an inward momentum opposing the outward momentum induced by the thermal

pressure gradients. Though the laser ponderomotive force is a very transient phenomenon

and it no longer exists after the laser pulse terminates, it will nonetheless alter the initial

plasma kinetics. As a result, the plasma expansion may be temporarily impeded, and the

local electron density profile further steepened. Equation (2-36) also implies that

independent of its wavelength, a high intensity laser light will penetrate to a specific

density in the plasma until its ponderomotive pressure is balanced by the thermal pressure.

It was shown that the ponderomotive force became important at laser intensities

higher than about 1014 W/cm2 [87]. Liu's work [86] reaffirmed this observation. He

measured the expansion velocity of the plasma critical surface through Doppler shift

experiments. By increasing the laser intensity from 1 x 1014 W/cm2 to 4 x 10o W/cm2 on

an Al target, he noticed that instead of a large increase of the critical layer velocity, the

Doppler shift remained largely unchanged within the experimental uncertainties. This

observation clearly demonstrated the repulsive effects of the ponderomotive force to the

plasma kinetics. It is expected that a linear increase in laser intensity would have resulted

in a roughly linear increase in the plasma temperature. Thus if the ponderomotive

pressure played no role, the resulting linear increase in thermal pressure should have led to

an increase in the Doppler shift by at least the square root of this change. An even greater

increase of the Doppler shift should be expected because of the increased Z at higher

plasma temperatures. In Liu's work, if no ponderomotive force was considered, a

simulated Doppler shift of the critical layer showed an overestimation of the expansion

velocity by a factor of three.


The experiments in our study can be divided into two parts. The first part includes

the synthesis as well as characterization of the diamond-like carbon thin films, while the

second part deals with the understanding of the laser plasma dynamics.

DLC films were made with 100 femtosecond (FWHM) Ti:Sapphire pulses at laser

intensities in the 3 x 1014 6 x 1015 W/cm2 range. The film properties were analyzed with

various techniques and correlated with the laser intensity. Carbon ion kinetic energy as a

function of laser intensity was measured through time-of-flight (TOF) experiments.

Using a Q-switched 7 nanosecond (FWHM) Nd:YAG laser beam, we also

deposited DLC films at laser intensities in the 3 x 1010 6 x 10" W/cm2 range. The

deposition parameters used in these experiments are more representative of the

conventional PLD techniques. The DLC films made with these nanosecond pulses, along

with the carbon ion kinetic energy measured as a function of laser intensity, were studied

and served as a reference to the femtosecond experiments.

Chapter 3 summarizes the experimental set-ups in our study. These include a

discussion of the laser systems employed, an outline of the vacuum system configuration

and DLC film deposition conditions, followed by a description of the TOF experiments.

Laser System

Femtosecond laser deposition of DLC films and plasma analyses were carried out

with a solid state Ti: Sapphire laser system, which has been developed as a high field

research laser at the Center for Ultrafast Optical Science, University of Michigan. It is

based on the principle of chirped pulse amplification (CPA) technique, which enables the

generation of high peak intensity laser pulses with moderate beam energy and extremely

short pulsewidth [88].

Figure 3-1 is a schematic of the CPA process. The pulses generated by a

broadband laser oscillator are coupled to a passive medium, producing a train of low

energy (~10-12 10-9 J) laser pulses with short pulse duration (< 10-12 s ). One of these

pulses is selected to pass through a pulse stretcher, after which the pulse duration is

increased by a factor of 103 to 104 while maintaining its beam coherence. This laser pulse

then undergoes a chain of amplifiers that have a combined gain on the order of 1010 1012,

thus increasing the laser energy to several hundred millijoules. This high energy pulse is

eventually sent through a compressor which reverses the function of the stretcher.

Ideally, the final compressed laser pulse can be as short as the original pulse with a

Fourier-transform limited bandwidth. The CPA process thus enables the generation of

laser pulses with peak power up to the terawatt (1012W) range.

The use of CPA technique is necessitated by dielectric breakdown of the gain

media and optical components induced by the high intensity laser beam. By stretching an

intense, ultrashort laser pulse, more energy can be extracted without imposing permanent

damage to these elements. For a laser pulse traveling in a normally dispersive medium, its

pulse duration will become longer as a result of the group velocity dispersion (GVD)

effects. That is, the light group velocity decreases with increasing frequency. As a result,

in the original short pulse envelope, the "red" component (light at longer wavelength)

travels ahead of the "blue" component (light at shorter wavelength). One will first

observe the "red" light and then the "blue" light if he stands facing the direction of pulse

propagation. The pulse is in this case dispersed, or stretched. This carrier wavelength

variation with time is called a chirp, and thence the origin of chirped-pulse amplification.

Because a grating is a highly dispersive optical element, it is often used in pairs as the

stretcher in a CPA laser system.




o .
H Cu

0 0

*E w_

0 &



Amplification of the laser pulse is often carried out by several amplifiers working in

a stepwise fashion. A Q-switched, cavity-dumped regenerative amplifier, injection-seeded

with the stretched pulse, is often used when the laser energy is low. By convention, this

amplification stage is termed as the "regen". After the "regen" are a series of single or

multiple-pass amplifiers operating in the saturated gain mode, further boosting the laser

pulse energy to higher levels.

A second pair of gratings, arranged in a configuration to generate opposite group

velocity dispersion to the stretcher grating pair, serve as the pulse compressor. The red

component of the laser light, which travels faster than the blue component in the stretcher,

is made to travel a longer optical path in the compressor so that the blue component can

catch up and eventually overlap with the red component. By doing so, a Fourier-

transform limited pulse can be conceived. A Fourier-transform limited pulse, however, is

in practice unattainable. As this requires that all the higher order dispersions experienced

by the laser pulse, while traveling through various stretching and amplifying stages, are

completely compensated by the compressor.

An oscillator material with a broad gain bandwidth is essential in the CPA process.

This is because the Fourier transform relation between the laser pulsewidth At and its

spectral width Av has the form A'rAv = 0.5. The gain medium thereby must have broad

bandwidth to support short pulses. Titanium-doped sapphire crystal (Ti:Sapphire) has a

bandwidth over 400 nm, which can in theory generate laser pulses as short as a few


Figure 3-2 is a schematic of the Ti: Sapphire laser system used in our study.

A self-mode-locked Ti: Sapphire oscillator generates -~1 nJ of output energy, when

pumped by an argon ion laser at ~ 4 W of power. These 50 fs, 780 nm pulses are then

stretched temporallyy chirped) to -300 ps by double passing them through a stretcher,

which consists of two spherical mirrors and two 1200 line/mm, gold-plated holographic

gratings. Next, they are injected into a regenerative amplifier pumped by a Q-switched,




N 8


I -


S ^ _____________ ^N0
< .~ T^ g

internally frequency doubled Nd:YAG laser operated at 10 Hz repetition rate. Typical

pumping energy is approximately 40 mJ. This amplifier is made up of an optical resonator

that contains the gain media Ti:Sapphire crystal, a Pockels cell, and a thin film polarizer

(TFP). The Pockels cell is oriented to give a static quarter-wave birefringence that allows

the pulse to enter the cavity. Once the pulse enters the cavity, a quarter-wave voltage is

applied to the Pockels cell. The net birefringence is now half-wave, trapping the pulse

inside. The pulse is allowed to make approximately 20 round trips in the cavity, allowing

most of the energy to be extracted from the gain media and then forced to exit by a half-

wave voltage applied to the Pockels cell [89]. Before entering the final compressing

stage, these 300 ps pulses are sent through two more stages of amplification, after which

they obtain energy up to about 100 mJ. The first amplification stage is made of a two-

pass amplifier pumped by the same 40 mJ Nd:YAG laser used in the "regen", and the

second one consists of a four-pass amplifier pumped by a second Nd:YAG laser with a

maximum output energy of 600 mJ. The final compressor grating pair is comprised of two

2400 line/mm, gold-plated holographic gratings. The compressed pulse is now near

Gaussian-shaped, centered at around 780 nm wavelength, with a full-width-half-maximum

(FWHM) of 100 femtoseconds. Over half of the laser energy is lost due to compressor

gratings absorption, leaving the maximum laser energy out of the compressor at about 40


We used a Quanta-Ray GCR-4 pulsed Nd:YAG system, which also serves as the

second amplification pump laser for the Ti:Sapphire system as our nanosecond beam

source. A Q-switched solid state laser, this Nd:YAG laser is operated at frequency-

doubled 532 nm wavelength and 10 Hz repetition rate. The output beam has a pulsewidth

of 7 ns (FWHM) and delivers a maximum energy of 600 mJ. The spatial beam profile is

quasi-Gaussian with a near diffraction limited nominal beam diameter of ~ 8 mm. Beam

divergence was determined to be ~ 0.5 mrd.

Turn box

MgF2 Window

Laser beam line




TC gauge

Fig. 3-3 Schematic of the high vacuum thin film deposition system.


Thin Film Deposition System

The deposition system consists of a high vacuum deposition chamber coupled with

vacuum laser beam delivery tubes. Vacuum pumping was provided by an oil diffusion

pump backed by a single stage rotary mechanical pump. A freon cold trap was mounted

on top of the diffusion pump to prevent backstreaming of pump oil into the vacuum

system. A standard Bayard-Alpert ion gauge and several thermocouple gauges were

affixed to the system for pressure measurement. The base pressure of this system was

maintained at 2 x 106 Torr. Figure 3-3 shows the schematic of the deposition system.

A high purity (> 99.999%) graphite disk, with dimensions of 0.25" x 1.00" from

K.J. Lesker Company, was used as the ablation target. It was attached to a target holder

rotating at ~ 5 rpm to expose fresh surface for each laser pulse, thereby attenuating the

cratering effects caused by the intense laser irradiation. Separation between the target and

the substrate was kept at 4 cm during each deposition. Substrates were mechanically

secured onto a rotating substrate holder. The design of this triangular-shaped substrate

holder allows three samples to be deposited in one operation, yet preventing the laser

plume from reaching the substrate not directly facing the target. Alternatively, when high

temperature is desired, a self-heating ceramic substrate holder can be used. Inside this

hollow block, a high temperature (up to 2100 K) quartz lamp serves as the substrate

heating source. Substrate temperature during deposition was monitored by a

Chromel/Alumel thermocouple attached to the substrate. It has a measurement range of 0

- 1200 OF.

A plano-convex silica lens with a 12.5 cm focal length was used to direct the

nanosecond Nd:YAG laser beam onto the graphite target. The target was placed at a 300

angle with respect to the incident laser beam. Figure 3-4 demonstrates this experimental


Rotatable substrate

Laser beam
erl Convex
I I lens

Rotating target holder

Fig. 3-4 Deposition arrangement for the Nd:YAG laser beam.

The diffraction limited focal spot Ddiff (at 1/e2 intensity), assuming a Gaussian

spatial distribution of the laser beam, can be calculated as:

Ddiff = (fVhtr) = 5 4am (3-1)

where f is the focal length of the piano-convex lens, X is the laser wavelength and r is the

nominal radius of the beam.

The diffraction limited depth of focus Ldiff, defined by the points along the beam
axis on each side of the focal plane where the beam area doubles, is estimated as:

Ldiff = (f2 tr2) w 165 apm (3-2)

Both Ddiff and Ldiff are believed to be much smaller than the actual focal spot and
depth of focus on the target. A more realistic estimate of the actual focal beam diameter

D and depth of focus L can be obtained as [90]:

D = fW = 62.5 upm (3-3)

L = Df/r t 2 mm (3-4)

where y is the Nd:YAG laser beam divergence. Considering that the target is placed at an

angle of 30 to the incident beam, a minimum beam diameter of 73 p.m should be expected

on the target. This compares with a 110 pm damage spot measured on the target surface

under an optical microscope. Damage induced by the laser beam is a nonlinear thermo-

optical effect influenced by various parameters. These include laser intensity, beam spatial

profile, target material thermal conductivity and surface topography, etc. The damage

spot therefore is not a true representation of the focal beam size. It, however, has been

found to serve well as an upper limit for the beam spot estimation. The Nd:YAG laser

intensity cited in this paper was calculated with a focal beam diameter of 73 pm.

The delivery of the femtosecond Ti:Sapphire pulses from the compressor gratings

onto the graphite target, on the other hand, is very different from the conventional laser

systems. This is due to some of the nonlinear optical phenomena induced by these short,

high intensity laser pulses: when the laser induced electric field is higher than 104 V/cm,

nonlinear response in a medium can no longer be neglected.

As high intensity laser pulses pass through a medium, the dielectric properties of

the medium will be altered and they become a function of the local laser field strength.

The propagation of the laser beam will in turn depend on the newly modified dielectric

properties of the medium. The light-medium interactions are now coupled and nonlinear.

Two of the nonlinear effects, self-focusing in the spatial domain and spectral broadening in

the temporal domain, are the primary concerns in our experiment. These two effects

result from an increase in the refractive index with laser intensity expressed as An = n2E|I2,

where n2 is the nonlinear refractive index coefficient, and E is the magnitude of the electric

field induced by the laser pulse. For a near Gaussian-shaped Ti: Sapphire pulse, the highest

laser intensity and thereby the largest increase of refractive index is at the center of the

beam. Both the laser intensity and refractive index become less toward the edge of the

beam profile. In air or other nonlinear media such as a glass vacuum window, this self-

induced effect will lead to the curvature of a flat laser wavefront, much like the effect of a

positive focusing lens. This self-focusing effect can cause catastrophic damage to optical

components in the beam path and totally distort the beam profile. Further more, a

broadening or stretching of the short femtosecond pulse in time will lead to miscalculated

laser intensity on the target. Distortion and self-focusing of the laser wave-front were in

fact experimentally observed on this laser system [91]. Inversely, the plasma generated by

an intense laser pulse will generally act as a negative lens, defocusing the light beam. The

intensity dependent ionization of the plasma is also non-uniform in the transverse plane of

the laser beam, with higher plasma density near the center of the beam than toward the

edge. However, the refractive index of the plasma generally decreases with increasing

laser intensity, thereby effectively "diluting" the laser beam. The laser light in this scenario

is said to have self-defocused.

The combined effects of these nonlinear phenomena, including self-focusing, self-

defocusing, beam bandwidth broadening, as well as beam diffraction on the final beam

profile constitute a complicated dynamic process and thus beyond the scope of this paper.

Our main objective was to minimize some of the potential problems associated with these

nonlinear effects.

The nonlinear effects on the laser beam can be quantitatively described by the so-

called B-integral, defined as [92]:

B = n2I(z)dz (3-5)

where L is the propagating distance of the laser beam in a nonlinear medium, n2 is the

nonlinear index coefficient, and I is the laser intensity. When B exceeds a value of 4 5,

the beam will break up and can not be well focused. It is therefore imperative to keep the

B-integral to a minimum in our study.

In consideration of the brief discussion above, we implemented vacuum beam lines

and all reflective optical components in our femtosecond experiments. Strictly speaking,

even in the case of a vacuum, laser light can still interact through vacuum polarization.

However, at the intensity level employed in our investigation, nonlinearity caused by

photon-photon scattering and other nonlinear effects is too small to warrant any real


In our experiment, the Ti:Sapphire laser pulses out of the compressor gratings

were aligned and immediately transferred into the vacuum beam line. A thin 3 mm MgF2

laser window serves as the front vacuum window. MgF2 was chosen due to its lower

nonlinear index of refraction when compared to a conventional fused silica vacuum

window. A p-polarized, high damage threshold reflective mirror positioned inside the

turning box redirects the beam into the deposition chamber.

A gold plated aluminum off-axis parabolic mirror was used to focus the incoming

Ti:Sapphire laser beam onto the graphite target. The parabolic mirror has a 10 cm

reflected effective focal length and a 600 off-axis angle. The parabolic mirror was chosen

over conventional glass focusing lenses to eliminate the same nonlinear effects induced by

these lenses, which could lead to spreading and distortion of the ultrashort laser pulses. A

micron-thick plastic wrap was placed between the substrate holder and the parabolic

mirror to protect the mirror surface from laser plume during deposition. Figure 3-5

illustrates this experimental configuration. The final laser energy measured on the target

surface was about 60% of its original value out of the compressor.

With the parabolic mirror, a diffraction limited focal spot was calculated to be

roughly 1 j.m. The actual spot size, however, was measured to be 11 pm. This is due to

the combined effects of mirror surface roughness of 80 100 A, as a result of diamond

milling of the Al surface, and a slightly divergent beam profile. The beam focal size of 11

upm was measured using an image relay technique: at low laser intensity, a microscope lens

was positioned near the mirror focal and it projected the image of the focal spot onto a

charge-coupled device (CCD) camera.

The femtosecond laser beam was capable of breaking down air even at relatively

low energy levels. The target was visually placed at the air breakdown spot. Positioning

the target precisely at the focal spot, however, proved to be difficult due to lack of a

precision motion controller available for the experiments. During actual film deposition,

the target was placed at an off-focal position to maintain a larger spot size. This was

found to be necessary to keep the otherwise tightly focused beam from cutting a deep

trench on the target surface. A minimum beam diameter of 50 p4m was used in calculating

the Ti:Sapphire laser intensity in our study. This compares with a damage spot of- 80

p.m found on the target surface under an optical microscope.

Fig. 3-5 Deposition arrangement for the Ti: Sapphire laser beam.

Rotating target holder Laser beam

Parabolic mirror

Rotatable s bstrate holder

DLC Film Deposition Conditions

The diamond-like carbon films were deposited onto fused silica and silicon

substrates at base vacuum pressure. The SiO2 (> 99.99% purity) substrates from Kamis

Inc. have dimensions of 1 cm x 1 cm x 3 mm. They were optically polished on both sides

with roughness < 20 A. Silicon substrates are n-type, phosphorus doped (100) single

crystal wafers cut into 1 cm x 1 cm squares. These substrates were ultrasonically cleaned

in acetone and methanol before been loaded into the deposition chamber. Since DLC

films are amorphous at room temperature, no special treatment was given to remove the

native oxide layer from the silicon wafer.

During the deposition processes, different intensity levels were achieved by varying

the incident laser energy and the beam spot size, while keeping the laser pulsewidth


The deposition conditions for DLC films with both the femtosecond and

nanosecond laser pulses are summarized in Table 4 and Table 5, respectively.

Table 4. Experimental conditions for Ti: Sapphire laser deposition of DLC films.

Laser Source
Repetition Rate


Target-Substrate Distance

Minimum Spot Size

Peak Intensity


Substrate Temperature

Ti: Sapphire (780 nm)
10 Hz

100 fs (FWHM)

4 cm

50 p.m

3 x 1014 6 x 101' W/cm2

Si, Si02

Room Temperature

Table 5. Experimental conditions for Nd:YAG laser deposition of DLC films.

Laser Source Nd:YAG (532nm)
Repetition Rate 10 Hz

Pulsewidth 7 ns (FWHM)

Target-Substrate Distance 4 cm

Minimum Spot Size 73 pm

Peak Intensity 2 x 1010 6 x 1011 W/cm2

Substrate Si, SiO2

Substrate Temperature Room Temperature

Time-of-Flight (TOF) Measurement Set-up

Time-of-flight (TOF) experiments were conducted to measure the kinetic energy

of carbon ions generated by the laser beam. A field-free drift tube coupled with a Faraday

ion collector was used for this purpose. The experimental arrangement was similar to that

of Demtroder and Jantz [80].

Ion transit time between the target surface and the Faraday cup was measured as a

function of laser intensity. The carbon ion velocities were determined with the time lapse

between the laser onset and the peak of the TOF distribution curve. This peak

corresponds to the most probable carbon ion velocity Vm with a Boltzmann-Maxwellian

like distribution. Their kinetic energies can then readily be calculated from KE =

0.5MVm2, where M is the carbon ion mass. Figure 3-6 illustrates the experimental lay out.

Shown in Figure 3-7 is the configuration of the ion drift tube. An outer tube with

a diameter of 10 cm is constructed of aluminum. At one end of the tube is an entrance

aperture with 2 cm in diameter. At the other end inside this tube is a Faraday ion

collector, which is electrically isolated from the tube wall and linked through a BNC

Fig. 3-6 Schematic of the TOF experiment.

Field free drift tube

SI to oscilloscope


Electron retarding
potential E = -15V

Fig. 3-7 Layout of the field-free TOF drift tube.

connector to a feedthrough flange on the vacuum chamber. A cylindrical inner tube with a
3 cm diameter and 12 cm in length is positioned coaxially between the aperture and the
Faraday cup. It is mounted by four retaining screws to the outer tube. Isolation between
the inner and outer tube is achieved by Teflon spacers. On both ends of the tube are finely
meshed copper grids, a voltage can be supplied to these grids. This enclosed tube design
creates a field-free drift region for the laser plasma and shields the Faraday cup from
electromagnetic noises and strayed charged particles. Small holes were drilled on the
outer tube wall to allow sufficient evacuation. The outer tube was grounded through the
chamber wall and the inner tube was kept floating during the TOF experiments. Between
the inner drift tube and the Faraday cup is a fine meshed copper screen. A negative
voltage of -15 V was applied to this grid during the plasma ion collection process. This
voltage served two purposes. First, it was intended to retard the primary electrons in the

plasma entering the drift tube. The electrons have the same expansion velocity as the ions

in order to keep the plasma in quasi-neutrality. However, these electrons have much

lower kinetic energies due to their smaller masses, therefore a relatively low negative bias

should be sufficient to separate them from the ions. The second function of this negative

voltage was to repel the secondary electrons from the Faraday cup generated by the high

energy plasma ions.

Ion signals collected by the Faraday cup were monitored by a Tektronix 2440

digital oscilloscope (500 MHz) and relayed through a GPIB card to a computer. The

oscilloscope was operated in the external triggering mode with a 50 ohm impedance

match. When studying the Ti:Sapphire femtosecond laser plasma, the triggering signal

was provided by the Pockels cell wave voltage, which coincides with the onset of each

laser pulse. Conversely, the Pockels cell voltage in the resonant cavity generates the

triggering signal when working with the nanosecond Nd:YAG laser.

The distance L from the target surface to the entrance aperture of the drift tube

was kept at 15 cm for each TOF data acquisition. An additional 16 cm from the entrance

to the surface of Faraday cup leads to a total ion travel distance of 31 cm. With an

entrance aperture d of 2 cm, the Faraday cup has a collection solid angle of 0 =

0.257d2/L2 = 1.4 x 10-2 sr. The Faraday cup surface was kept perpendicular to the target

normal for all the TOF measurements.


This chapter details the experimental results of our investigation. First, results

from the time-of-flight (TOF) experiments, which measured the carbon ion kinetic energy

as a function of laser intensity, are presented. This is followed by a description of the

DLC film properties analyzed by various techniques. The DLC films made by the

femtosecond Ti:Sapphire pulses are compared with those deposited with the conventional

nanosecond Nd:YAG pulses. A correlation between the carbon ion kinetic energy and the

DLC film characteristics is established.

Time-of-Flight (TOF) Measurements

Laser plasma analysis was carried out by the time-of-flight (TOF) experiments

described in Chapter 3. Data were taken at a background pressure of 2 x 10-5 Torr with a

Tektronix 2440 digital oscilloscope, which has a rising time of 10 ns. The Faraday ion

collector surface was kept perpendicular to the target normal during the measurements.

Due to rather severe pulse to pulse energy fluctuation (up to 15%) observed for the

Ti:Sapphire laser beam, all TOF data were averaged over eight pulses on the oscilloscope.

For the nanosecond Nd:YAG and the femtosecond Ti:Sapphire pulse induced plasmas,

their TOF spectra were taken at laser intensities in the 2 x 109 6 x 10"1 W/cm2 and 1 x
1014 6 x 1015 W/cm2 range, respectively. For each laser system, the upper intensity limit

is typical of the maximum laser intensity used in making DLC films, while the lowest

intensity value is determined by the oscilloscope signal/noise ratio.

Figure 4-la shows a typical TOF spectrum collected on the Faraday cup, from the

Nd:YAG beam induced plasma at a laser intensity of 2 x 109 W/cm2. The "zero time"

in time zero


0 5 10 15 20 25 30 35
Flight time(106 s)

Fig. 4-la TOF spectrum plotted on the time scale at laser intensity of 2 x 109 W/cm2



c 0.6

o 0.4


1 106 1.5 106 2 106 2.5 106
Velocity (cm/s)

3 106 3.5 106 4 106

Fig. 4-lb TOF spectrum plotted on the velocity scale at laser intensity of 2 x 109 W/cm2.

0 Lo
5 105



0 4 8 12 16
Flight time(10-6 s)

Fig. 4-2a TOF spectrum plotted on the time scale at laser intensity of 2 x 1010 W/cm2.

1 .2 .. |, ,.. ,

S .8 / -..

~S 0.6

c 0.4 -
0.2 -

0 *
0 5 106 1 107 1.5 107 2 107
Velocity (cm/s)
Fig. 4-2b TOF spectrum plotted on the velocity scale at laser intensity of 2 x 1010 W/cm2.

0 2 4 6 8
Flight time(10-6 s)

Fig. 4-3a TOF spectrum plotted on the time scale at laser intensity of 7 x 10" W/cm2.

CU 0.8
c 0.6
c 0.4



0 5 106 1 107 1.5 107 2 107 2.5 107 3 107
Velocity (cm/s)

Fig. 4-3b TOF spectrum plotted on the velocity scale at laser intensity of 7 x 10" W/cm2.

2.5 10'

2 107

1.5 107

1 107

5 106

*0 O

- ci 0

0 Vmax
- Vm





100 200 300
Laser intensity



600 700

Fig. 4-4 Most

probable and maximum carbon ion velocity as a function of laser

100 200 300 400 500
Laser intensity (109 w/cm2)

600 700

Fig. 4-5 Most
laser intensity.

probable and maximum carbon ion kinetic energy as a function of

4 103

3.5 103

3 103

2.5 103

2 103

1.5 103

1 103



-, I I o Em a
U E m

o ,
,.a a
- LI
*. D a Em I
:a Emax

[ ., ,I I i ,i i .-


................... Illllllllllllll


corresponding to the moment the laser pulse strikes the target surface, is referenced by the

early spike on the spectrum preceding the main plasma peak. It is generated by the soft X-

rays from the plasma. The collected TOF spectrum, with a fast rising front edge and a

trailing tail, is characteristic of those produced by a nanosecond laser pulse [94-96]. The

main spectral peak yields the most probable carbon ion velocity Vm, and the maximum ion

velocity Vmax is represented by the rising edge of the spectrum. Figure 4-lb is the same

TOF spectrum plotted on the velocity scale. It was fitted with a shifted center-of-mass,

Boltzmann-Maxwellian like distribution of the form f(v) ~ vnexp[-m(v-Vcm)2/2kT]. The

best fit, shown as the solid curve in Figure 4-1b, was obtained with the power law factor n

= 2 and a center-of-mass velocity vem = 2.1 x 106 cm/s. The fact that their TOF spectrum

can be fit with a Boltzmann-Maxwellian function implies that the plasma ions, upon

reaching the Faraday cup, are in local thermal equilibrium. At higher laser intensities, the

TOF spectra exhibited similar features, except that the main ion peak as well as its rising

edge have shifted toward shorter time scales, indicating an increase in both the median and

maximum carbon ion velocities. Shown in Figure 4-2a and Figure 4-3a are the TOF

spectra taken at laser intensities of 2 x 1010 W/cm2 and 6 x 1011 W/cm2, respectively.

Their velocity distributions, however, appear to have deviated from Boltzmann-

Maxwellian function. This is because at higher intensity levels, the shape of the TOF curve

is determined by ions which have experienced acceleration based effects and therefore can

no longer be described by a thermal equilibrium state. The fits to data curves are also

plotted as solid lines in Figures 4-2b and 4-3b.

Figure 4-4 illustrates the most probable (Vm) and maximum (Vmax) carbon ion
velocity calculated as a function of laser intensity. Their values increase rapidly at lower

intensities and then start to level off in the 1011 W/cm2 range. Because the Faraday ion

collection design does not yield the carbon ion mass/charge information, its kinetic energy

was calculated based on the assumption that all collected ions are in the atomic form.

Figure 4-5 shows the corresponding most probable (Em) and maximum (Ema) carbon ion

kinetic energy as a function of laser intensity. At the highest laser intensity of 6 x 1011

W/cm2, up to 1.9 keV and 3.5 keV were estimated for Em and Emax, respectively. Data

fitting yielded Em cc 10.40 and Emax oc I-3, with I being the laser intensity. These measured

kinetic energy values are in agreement with the simulations presented by Stevefelt and

Collins [81], they also compare well with the experimental results of Demtroder [80] and

Gregg [93].

It should be noted that even at the lowest laser intensity of 2 x 1010 W/cm2 used to

deposit DLC films, the plasma has a most probable ion kinetic energy Em on the order of

300 eV. This is higher than those observed with UV excimer lasers running in the 108

W/cm2 range, where Em was found to be several tens of electronvolts [63,66]. It would be

interesting to compare the quality of DLC films deposited by Nd:YAG laser at similar

kinetic energies. However, this relatively low kinetic energy level could only be achieved

at just above the Faraday cup detection limit of 2 x 109 W/cm2 (Em ; 25 eV), where the

film deposition rate was extremely low.

As expected, the area under each TOF spectrum, which represents the total ion

charge collected by the Faraday cup, increases as a function of increasing laser intensity.

This indicates a larger number of generated carbon ions and/or an increase in average ion

charge at higher intensities. By measuring the total ion charge, in combination with the

knowledge of the film thickness and the ion collector solid angle, we were able to obtain

an estimated plasma ion fraction of 0.1 1%. This number was calculated based on the

assumption that the DLC films deposited at 4 cm away from the target have a density of 3

g/cm3, and the average ion charge is 3 [29, 53, 54, 81].

In comparison, the TOF spectra taken from femtosecond Ti:Sapphire laser induced
plasmas have some strikingly different features. Shown in Figure 4-6a is a typical TOF

spectrum taken at a laser intensity of 3 x 1014 W/cm2. Notice in addition to the main ion

peak marked as "Vm", there is a second peak marked as "V," in the spectrum preceding

this main peak. When converted into ion velocity in Figure 4-6b, this early peak



0 5 10 15 20 25
Flight time (10-6 s)

Fig. 4-6a TOF spectrum plotted on the time scale at laser intensity of 3 x 1014 W/cm2.




0 1 107 2 107 3 107 4 107 5 107 6 107
Velocity (cm/s)

Fig. 4-6b TOF spectrum plotted on the velocity scale at laser intensity of 3 x 1014 W/cm2.


0 5 10 15 20
Flight time (10-6 s)

Fig. 4-7a TOF spectrum plotted on the time scale at laser intensity of 1 x 10"5 W/cm2.

0 2107 4 107 6107 8107
Velocity (cm/s)

Fig. 4-7b TOF spectrum plotted on the velocity scale at laser intensity of I x 10"1 W/cm2.

0 5 10 15 20
Flight time (106 s)

Fig. 4-8a TOF spectrum plotted on the time scale at laser intensity of 6 x 10" W/cm2.

0 2 107 4 107 6 10' 8 10'
Velocity (cm/s)

1 108

Fig. 4-8b TOF spectrum plotted on the velocity scale at laser intensity of 6 x 10"5 W/cm2.