Gold thin films produced from laser stimulated plasma reaction products


Material Information

Gold thin films produced from laser stimulated plasma reaction products
Physical Description:
ix, 235 leaves : ill. ; 28 cm.
Simon, Charles George, 1954-
Publication Date:


Subjects / Keywords:
Gold films   ( lcsh )
Gold coatings   ( lcsh )
Thin films   ( lcsh )
Metallic films   ( lcsh )
bibliography   ( marcgt )
theses   ( marcgt )
non-fiction   ( marcgt )


Thesis (Ph. D.)--University of Florida, 1988.
Includes bibliographical references.
Statement of Responsibility:
by Charles George Simon.
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 - 001129780
notis - AFM7016
oclc - 20177934
sobekcm - AA00004826_00001
System ID:

Full Text







To my wife, Diane,
to my parents, Sam and Bernice,
to Bobby Sammons.


The author wishes to express his gratitude to all who contributed

to this work and to his education at the University of Florida. In

particular, he wishes to express his appreciation for all of the effort

and guidance provided by his research director and committee chairman,

Samuel Colgate.

The author would like to thank his peers and the faculty and staff

members at the University of Florida (U.F.) and the University of South

Florida (U.S.F.) who contributed to this effort: at U.F., Mark Hail

and Roy King for quadrupole MS analyses, Mehdi Moini for FTICRMS

experiments, Dan Lesson for cluster beam TOFMS experiments, Paul

McCaslin and Jack Davis, Jr., for XPS analyses; at U.S.F., Alfred

D'Agostino and Carl Biver for XPS analyses, and Mike Ammons and Alicia

Slater for SEM/EDS analyses. Specials thanks are also extended to

Stephanie Boggess at U.F. for her invaluable assistance throughout the

course of this work, and to Bobby G. Sammons of Stone Container

Corporation, Port Wentworth, Georgia, for providing that special


Finally, the author wishes to thank his committee members for

their contributions to this work, especially Martin Vala and John Eyler

who provided laboratory equipment and time, and The Florida High

Technology and Industry Council for their financial support (FHTIC

Grant #85092726).



ACKNOWLEDGEMENTS........... .......................... iii

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

ABSTRACT............................................. viii


I INTRODUCTION........................................ 1

II EXPERIMENTAL........................................ 13

Production of Gas-Phase Gold Species: The Plasma
Reaction System ................................. 13
The Carbon Dioxide Laser, Optics, and
Operation .. ............................... 14
Gas Handling System and Procedures............. 18
The Plasma Reaction Chambers................... 28
The glass plasma reaction chamber......... 28
The aluminum plasma reaction chamber...... 34
The Product-Recovery/Substrate-Coating Systems. 40
Heated quartz tube........................ 40
Staged cold traps......................... 41
Substrate coating chambers................ 46
Heated substrate mount.................... 50
Production of Gas-Phase Gold Species: Experimental
Conditions and Materials......................... 51
Screening Study Experiments.................... 51
Substrate Coating and Annealing Experiments.... 64
Product Volatility Study............................ 75
Sublimation Experiments........................ 76
Mass Spectrometer Experiments.................. 77
Time of Flight MS Experiment .................. 80
Gold Thin Film Characterization..................... 83
Visible Absorption Spectroscopy............... 84
Optical Microreflectometry..................... 85
X-Ray Photoelectron Spectroscopy............... 86
Scanning Electron Microscopy with Energy
Dispersive Spectroscopy...................... 89
Electrical Resistance Measurements............. 90


III RESULTS AND DISCUSSION.............................. 92

Production of Gas-Phase Gold Species................ 92
Screening Study Results........................ 92
Substrate Coating and Annealing Results........ 111
Product Volatility Study Results.................... 115
Sublimation Experiments........................ 115
Mass Spectrometer Experiments.................. 116
Time of Flight MS Experiment................... 138
Gold Thin Film Characterization Results.,........... 151
Visible Absorption............................. 151
Optical Microreflectrometry.................... 153
X-Ray Photoelectron Spectroscopy.............. 158
Scanning Electron Microscopy with Energy
Dispersive Spectroscopy ..................... 164
Electrical Resistance Measurements............. 193

IV SUMMARY AND CONCLUSIONS............................. 210

Discussion of Plasma Reaction Product Formation..... 210
Conclusions................................... 219


A GLASS FIBER FILTER BLANKS .......................... 222



REFERENCES................................................. 230

BIOGRAPHICAL SKETCH........................................ 235


Table Page

1. Gases Used in the Plasma Reaction Study.............. 54

2. Screening Study Plasma Reaction Experiments.......... 55

3. Substrate Coating Experiments........................ 65

4. Substrate Annealing Conditions....................... 69

5. Volatility Study Mass Spectrometer Experiments....... 77

6. Filter Catch Plasma Experiment Results............... 107

7. GC/MS Results from an Acetone Extract of H2C2H2/Au
Plasma Reaction Products ............................ 118

8. XPS Analytical Conditions and Results................ 161

9. Glass Fiber Filter Blanks............................ 214

10. Repetitive Weighings on the Mettler Analytical
Balance................................... ..... 215

11. R(stnd) Values for SiC................................ 220

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


Charles George Simon

December 1988
Chairman: Samuel 0. Colgate
Major Department: Chemistry

Superfine particles of gold in the 100-400 nanometer size range

were produced in a plasma stimulated by 10.6 micron radiation from a

pulsed carbon dioxide laser. The production of the particles was shown

to be highly dependent on the types and pressures of the gases used in

the plasma, and the incident power density. The best conditions

determined for production of the particles were a partial pressure of

350 Pascals of methane gas mixed with hydrogen to a total pressure of

700-1400 Pascals, and an incident power density of 3 x 109 watts/cm2.

The particles were used to coat tungsten, glass, and polymeric

substrates. The coated tungsten and glass substrates were subjected to

air and hydrogen anneals convectively at temperatures up to 600 K, and

by 10.6 and 0.532 micron laser radiation. Selected sample films were

then subjected to analyses by optical microreflectometry, X-ray

photoelectron spectroscopy, scanning electron microscopy, energy

dispersive spectroscopy, and electrical resistance.

The results of these analyses showed that the gold films had

highly variable properties depending on the substrate and annealing


conditions. The average mass concentration of gold in the films was

95% (54 atomic %). High quality continuous gold lines could be written

on coated glass substrates using continuous wave 0.532 micron argon ion

laser radiation. Three distinct colors could be developed in the films

using thermal input. This property makes the product a viable

candidate for use as a multi-level optical encoding media.

Film collapse did not occur until 490 K and there were no other
contaminants in the films due to the containerless evaporation

technique. These properties indicate that the plasma reaction products

may be suitable high temperature replacements for gold blacks.


Methods for producing submicron gold film patterns are of current

interest to the microelectronic industry for use as interconnect

metallization as device sizes shrink in VLSI (very large scale

integration) circuitry. Interconnects provide communication routes

between single devices and each other, between single devices and the

outside world and between active regions within a single device. Since

propagation time delay along the interconnect is directly proportional

to film resistivity and inversely proportional to the film thickness,

as well as varying with the square of the length,1 it is important to

produce short and relatively thick (compared to the width) films with

good electrical properties in order to minimize operating times.

Current micron-size interconnects are patterned using aluminum, but

industry progression to submicron-size devices will eliminate aluminum

for this use because of electromigration problems. Gold is an ideal

substitute for aluminum because of its low resistivity, high

conductivity, and resistance to electromigration.

Chemical vapor deposition (CVD) is the only currently available

technique for producing gold thin films that met both the selective

deposition and high throughput requirements necessary for mass-

producing submicron patterns in VLSI circuits. Volatile gold compounds

are used in the CVD process to form the thin metal films by decomposing


on activated surfaces, leaving the metal atom on the surface and the

remaining molecular constituents in the gas phase. Thus, there is a

concurrent interest in high yield, cost effective methods for

synthesizing suitable compounds for use in gold CVD processes.

An alternative method to conventional solution based synthetic

chemistry that offers several advantages is the in situ gas phase

production of suitable gold compounds. This technique could be

integrated into a CVD system and produce volatile gold compounds on

demand, eliminating the need to store and handle such compounds, and

removing the processor's dependency on off-site precursor production.

The ideal system would use solid gold and non-toxic reactive gases to

form the volatile species upstream from the CVD chamber. After passing

through the CVD chamber, any unused gold compounds would be easily

recovered for recycle or gold reclamation. The following report

describes efforts to explore and understand novel chemical means which

might support the development of such a system based on production of

gas phase gold compounds in a pulsed laser stimulated plasma using a

solid gold target and non-toxic reactant gases.

A brief review of the physical properties and mechanisms of

currently available materials and methods utilized in interconnect

formation will illustrate the limitations that need to be overcome.

The advantages of using gold for interconnect metallization and the

effectiveness of CVD for submicron patterning will be described.

Evidence of the gas phase production of gold compounds using energetic

plasmas will also be presented. Finally, a brief description of the

proposed technique will be given, followed in Chapter II by a complete


description of all equipment and procedures and in Chapter III by a

discussion of the results.

Among the elements only gold, silver, copper, and aluminum have

high enough electrical conductivities at device operating temperatures

(> 300 K) to be considered for use in interconnect formation.2 Both

silver and copper are highly reactive and subject to catastrophic

oxidation. While aluminum is more stable to oxidation, it is far more

susceptible to electromigration than gold. Electromigration results

from mobility of the metal ions in the film driven by the friction

forces exerted by the flowing electrons. The ultimate result of

electromigration is the formation of a void large enough to cause a

discontinuity in the interconnect resulting in device failure.3 When

the factors responsible for electromigration are considered, the

advantages of gold over aluminum become clear.

In the polycrystalline thin metallic films which make up

interconnect lines, electromigration can be evaluated by considering

the total atomic flux, F, which can be expressed as the sum of the

atomic fluxes through the interior of the grain (equivalent to the flux

in the bulk metal) and the flux across the grain boundaries.

Huntington and Grone4 derived an expression for the flux in bulk

material, while Ho and d'Heurle5 did so for the added atomic flux

across grain boundaries. Both approaches are based on statistical

distribution of mobile atoms and adjacent vacancies in the crystal

lattice or grain boundaries. Combining their two expressions yields an

equation for F as a function of the current density J, the atomic

density N, the atomic diffusion coefficient D, the absolute temperature

T, the effective charge Z*q, the bulk conductivity cb, and the film

conductivity cf:

F = NDZ*qJ/cbkT + (NDZ*qJ/cfkT)(s/d)

where k is the Boltzman constant, s is the average separation distance

between grain boundaries, and d is the average grain size. This model

is only an approximation since it ignores the geometrical orientation

of the grains and impurities adsorbed on the grain boundaries, and it

assumes that transport across grain boundaries occurs via channels

between the grains.

It is immediately apparent that the lower atomic density of gold,

0.098 moles/cm3, compared to aluminum, 0.167 moles/cm3,6 leads to a

lower F value for gold. Also, the lower effective nuclear charge of

the gold atom and the gold (+1) and (+3) ions (Z* = 0.65, 1.0, and 3.0,

respectively, using Slater's Rules) compared to the aluminum atom and

(+3) ion (Z* = 3.15 and 4.2, respectively, using Slater's Rules)7

results in a further lowering of the F value for gold.

Since movement of atoms within a crystal is highly restricted, and

the grain boundary area in these polycrystalline metal thin film

interconnects is many times greater than the external surface area, the

atomic flux within the thin film interconnect is dominated by diffusion

across the grain boundaries.8 Thus, the heavier, more bulky gold

atoms, possessing a lower relative charge than aluminum, but having a

similar net bulk resistivity of ca. 2.5 microohms/cm, are far less

prone to electromigration at high current densities. Typically,


current densities > 105 amp/cm2 cause serious electromigration problems

in aluminum and its alloys, while gold does not experience serious

problems until current densities exceed approximately 109 amp/cm2.9

The next generation of electronic devices are expected to be of

submicron dimensions and operate at current densities up to 108

amp/cm2.10 Thus gold appears to be a suitable material for

interconnect metallization in future devices. There are several

problems associated with the use of gold in silicon based circuitry.

Most notably, the high diffusivity of gold in silicon and silicon

dioxide necessitates the use of a barrier metal such as tungsten

between the two materials.11 While industrial methods for mass-

producing submicron patterns of tungsten based on CVD techniques

currently exist,11-16 there are no methods currently available for

mass-producing submicron patterns of gold. If an inexpensive, readily

available source of volatile gold species with low toxicity and good

deposition characteristics could be developed, it would greatly

facilitate efforts in this area.

A review of the methods available for producing gold thin films

reveals significant problems associated with the application of each

method to the mass-production of submicron patterning. There are seven

general categories of techniques currently employed to form gold films:

1) electroplating

2) sputtering (including reactive sputtering)

3) vacuum evaporation

4) ionized cluster deposition

5) superfine particle deposition

6) chemical vapor deposition (CVD)

7) direct write (including laser, electron, and ion beam


Electroplating gold from solutions has been used extensively by

the electronic industry to form thin film devices and structures with

dimensions greater than several microns.17 However, the films produced

are typically porous and have significantly lower conductivities than

bulk gold. These plated films also contain alkali ion contaminants

that migrate in the electric fields and accumulate at junctions where

they change the device operating characteristics. There is also the

problem of dendritic growth experienced in electroplated gold films,

which can lead to formation of short circuits in adjacent conductors.18

A recent improvement in conventional electroplating was developed

by C. Patton using a laser-enhanced jet plating technique.19 This

method produced high quality gold spots and lines by directing a 25-

Watt continuous-wave argon ion laser beam through a focusing lens and

into a free-standing jet of electrolyte that acted as an optical wave

guide and a gold ion source. The laser energy and the fine jet stream

of electrolyte were directed onto a rastered cathode surface where

deposition rates as high as 30 microns/sec. were reported. The

inherently low throughput of this rastered system precludes its use as

a technique for mass-production of submicron gold patterns.

Sputtering is a physical vapor deposition (PVD) technique that

uses ionized gases (usually argon) to bombard a target composed of the

metal to be deposited. The energetic ion collisions dislodge surface

and near surface neutral metal atoms which then travel randomly to the


substrate where they are deposited. Masks and etching techniques are

used to form patterns. Several variations have been developed, such as

adding a reactive gas to the ambient, or heating and/or applying a

radio-frequency (rf) bias to the substrate. These modifications

improve step coverage on the substrate surface, and well-defined step

coverage is essential for small scale device and interconnect

formation.11,20-23 Non-uniform step coverage results in "thin" areas

that lead to early device failure. The improvements have not

completely solved the step coverage problems for small dimensions, and

the methodology is still considered non-conformal.22 This major

disadvantage along with the non-selectivity of the deposition process

precludes the use of sputtering techniques for mass-production of

submicron gold patterns.

Vacuum evaporation is also a physical vapor deposition process.

Thermal convection, resistance heating, laser vaporization, or electron

gun vaporization is used to evaporate the metal in a vacuum. The atoms

then travel by line of sight to the substrate where they condense and

form a thin film.21 Masking and etching techniques are used to form

patterns. This methodology is limited by non-conformal step coverage

and non-selectivity of the deposited films.22 These limitations

exclude evaporation techniques from use in the mass-production of

submicron gold patterns.

Ionized cluster beam deposition is a recently developed technique

that utilizes ionized clusters of up to several thousand atoms to form

thin films.21,24 The clusters are formed by homogeneous or

heterogeneous nucleation either in the vapor phase in the plume exhaust

from a nozzle fitted onto a hot crucible containing the molten source

material, or on the side walls of the crucible itself before exiting

through the nozzle. The clusters are subsequently ionized by electron

bombardment with a current of several milliamps and energies in the 80-

100 eV range. They are then accelerated through a 1-10 KV potential

and directed to the substrate surface where they deposit non-

selectively with enough energy per atom (1-10 eV) to facilitate surface

mobility without inducing defect damage associated with higher impact


A recent study by Knauer and Poeschel showed that under optimum

conditions only about 1% of the total metal flux from a crucible filled

with molten gold or silver was in the form of clusters in the size

range of several hundred to several thousand atoms.25 Since only a few

percent of these clusters were ionized in the electron bombardment

step, the net efficiency of metal use was very low. This limitation,

together with the need for line of sight deposition (which implies non-

conformal step coverage) and the non-selectivity of the process,

preclude its use for mass-production of submicron gold film patterns.

Another recently developed technique for producing gold thin

films, also based on homogeneous and heterogeneous nucleation of

particles in the gas phase, involves the production of "ultra-fine"

particles in the 5-65 nanometer size range (103-107 atoms/particle)

that are subsequently deposited non-selectively on a substrate.26-28

Metal vapors are produced using a laser beam, arc discharge, or

electron beam in the presence of inert or reactive gases. At least one

group applied the technique to gold.28 The same limitations of non-


selectivity and non-conformal step coverage eliminate this method from

consideration as a viable procedure for submicron patterning of gold

films. However, investigators claim that high yields of metal

compounds can be formed by the activated clusters in reactive gas

streams.27 This is an interesting point which will be discussed

further below.

Chemical vapor deposition techniques do not depend on a flux of

condensible species reaching the substrate by direct line of sight and

hence are not subject to non-conformal step coverage problems.22

Substrates are surrounded by metal-bearing volatile species that do not

decompose without some form of additional energy input.20 In

principle, this allows close packing of the substrates in the

deposition chamber and selective deposition via selective energization

of pre-patterned barrier metal surfaces. There are many variations of

the basic CVD methodology designed to optimize performance of a

particular metal deposition system, but they all utilize a volatile

species that is forced to decompose on the substrate areas of interest

and leave behind a thin film of relatively pure metal.16,20-23

Chemical vapor deposition is the only category of currently

available metallization technology that meets the requirements of

selective deposition and high throughput necessary for mass-production

of submicron gold thin film patterns. To date, no reports of

simultaneous patterning of large arrays of submicron features with gold

have been published. Non-selective CVD of gold from triarylphosphine-

goldchlorides using thermal decomposition was demonstrated by Mann,

Wells, and Purdie in 1937.29 House and Colgate recently reported


similar results using the same compounds and rf stimulated thermal

decomposition.30 There are other examples of non-selective CVD of gold

from organometallic compounds such as AuR(CNR'), and (R) PAuOSi(R),

where R and R' are methyl, ethyl, or phenyl groups.31,32 The most

recent compounds to be used for CVD of gold are dimethylacetylaceto-

natogold(III) and its tri- and hexafluoro derivatives. Larson and Baum

reported the formation of high quality gold films from these compounds

by thermal decomposition at 300'C.33 Films produced from the

fluorinated derivatives were relatively free from contamination and had

resistance values about twice that of bulk gold. This work represents

the current state of the art for gold CVD methods. While the compounds

used performed well, they are difficult to synthesize and have low

synthetic yields.34 These compounds are available commercially at a

cost of several hundred dollars per gram.35 Clearly a cheaper, more

readily available source of volatile gold species would be of benefit.

The only examples of submicron selective gold deposition reported

utilized various direct write methods, such as the laser-jet

electroplating technique described earlier,19 or ion beam pyrolysis of

solid (spun-on) organometallic films.36 Most groups have used focused

laser, ion, or electron beams to induce selective CVD of gold from

volatile compounds.37-44 Again, the fluorinated derivatives of

dimethylacetylacetonatogold(III) formed the highest quality submicron

sized gold lines and spots. Despite the high quality of the gold films

produced by these methods, their inherently slow rate of throughput

precludes their economic use in mass-production scenarios.


As previously mentioned, one possible alternate source of

organogold compounds for use in CVD systems that offers the advantages

of low cost and availability on demand is the in situ production of

such compounds from solid gold and inocuous gases. Techniques exist

for producing gold oxides and halides from solid targets using sputter

gun evaporation in the presence of the reactive gases.45 Araya et al.

claim that goldcarbide superfine particles can be generated in the gas

phase from Nd:YAG pulsed laser stimulated plasmas using solid gold

targets in the presence of methane or Freon gases.27 There are also

examples of organogold ions formed in ion cyclotron resonance mass

spectrometer experiments using high-voltage spark and carbon dioxide

pulsed laser stimulated vaporization/ionization of solid targets in the

presence of aliphatic or aromatic hydrocarbons, alkylhalides, and


In another related field of research, investigators have been

producing gold bearing plasma polymerized films using PVD sputtering or

evaporation techniques.50-58 These studies have shown that the metal

is incorporated in the films in the form of small particles that

drastically alter the optical and electrical properties of the


Gold carbonyls and organogold compounds have been formed in matrix

isolation studies by cocondensation of gold atoms with carbon monoxide,

acetylene, ethylene, and propylene.59-62 These studies illustrate the

high reactivity of gold atoms with carbonaceous molecules, even at very

low temperatures.


Taking this evidence into consideration, it would seem possible to

develop a continuous process method for generating gaseous gold species

from solid gold and gaseous compounds in an energetic plasma. Such a

process could be incorporated in a CVD production line to provide

volatile gold compounds on demand. This technique would also eliminate

the need to store and handle synthetic compounds, a major economic and

safety factor advantage in itself.

Materials produced by the method would be evaluated by considering

the following criteria:

1) production rate

2) volatility

3) thermal stability

4) temperature of decomposition

5) film properties (resistance, purity, morphology, stability)

6) properties of by-products (should be non-contaminating to the

substrate and non-toxic)

7) ability to recycle or reclaim gold.

In addition to these considerations, the toxicity of the starting

materials would be a primary concern.

A plasma reaction system was designed and constructed to

investigate the feasibility of producing gas phase gold compounds in

large enough quantities to be used in mass-producing CVD submicron gold

patterns. The procedures and results of that investigation are

presented in the following chapters.


The scope of this study was limited to the investigation of

reactions occurring between non-toxic or low-toxic gaseous species and

pulsed C02 laser stimulated gold vapors and plasmas. Particular

emphasis was placed on utilization of permanent carbonaceous gases.

The study consisted of three major parts: 1) production of gas phase

gold-bearing species in high yields in the plasma, 2) determination of

volatility of high-yield species, and 3) characterization of gold thin

films produced from the high-yield species. Detailed descriptions of

experimental equipment and procedures are presented together in this

chapter. Each major part of the study is treated autonomously.

Conditions leading to equipment or procedure modifications are

mentioned but are not discussed until the following chapter.

Production of Gas-Phase Gold Species:
The Plasma Reaction System

The plasma reaction system was composed of four sub-systems:

1) the pulsed carbon dioxide laser and associated optics

2) the gas handling system

3) the plasma reaction chambers

4) the product-recovery/substrate-coating systems.

Each sub-system is described in detail together with pertinent

operational procedures followed during the course of the study.

Carbon Dioxide Laser, Optics, and Operation

The foremost criterion for development of a successful method for

producing gaseous gold compounds for the intended use was a high

production rate. An energy source was needed that couTd provide both a

high flux of gold vapor and the additional energy needed to form a

reactive plasma from the volatilized gold and reactive gases. A

Lumonics pulsed TEA (transversely excited atmosphere) carbon dioxide

laser, model 101-1, that had previously demonstrated high

volatilization rates for refractory metals63 was available for use in

this study. This laser was capable of producing an optical pulse

energy of 2.5 joules with a pulse halfwidth of 250 nanoseconds and a

power density of approximately 3 x 109 watts/cm2 when optimally focused

through a 200 mm focal length lens. Maximum pulse repetition rate was

1 hz, giving the laser a duty time of approximately one millisecond per

hour. The emitted wavelength of 10.6 microns corresponds to the P(20)

line in the 001-100 band.

The laser's pulse energy could be varied by adjusting the nitrogen

gas pressure and flow rate to the spark gap trigger, by adjusting the

composition and flow rate of the helium/nitrogen/carbon-dioxide lasing

gas mixture, or by adjusting the operating voltage. During the course

of this study these parameters were maintained in a state that resulted

in maximum energy output.

The laser energy output was measured after major maintenance

events, such as disassembly and cleaning of the optics or electrodes,

using a Scientech 365 laser power/energy meter. During subsequent

operation, the laser output was monitored by exposing a piece of

thermal paper (Hewlett Packard #820-51A) to single pulses of the

unfocused, 28 mm diameter beam and comparing the resultant discoloration

patterns to standard patterns that were produced under measured

conditions (see Figure 1).

The laser beam was directed into the plasma reaction chamber

through the optic path using a 5 cm square face coated silver mirror

during initial screening experiments. This mirror was subsequently

removed from the system and the laser was positioned in a direct line

of fire with the optics and the chamber at a distance of ca. 0.5 meters

in order to transmit as much energy as possible to the target.

An externally mounted 5 cm diameter germanium meniscus lens with a

200 mm focal length (Oriel model 4365) was used to focus the laser beam

during the first half of the study. A similar lens with a focal length

of 250 mm (II-VI, Inc., model 5-5741-1) replaced the initial lens for

the final half of the study. (This was necessary as a result of

physical damage sustained by the 200 mm lens during a cleaning

procedure accident.) Both lenses had an anti-reflection coating for

maximum transmission at 10.6 microns.

The lenses were held in a clamp mounted to an optical stand with

three degrees of freedom. This arrangement allowed the lens to be

positioned closer to the target than the optimum focal distance during

the course of an experimental run, thus distributing the laser energy

Figure 1. Thermal Paper Exposed to Single Pulses of the Carbon
Dioxide Laser
(a) Laser performing well (nominal)
(b) Laser performing poorly (gas mixture not
optimized for power level and pulse repetition


(b) 5 ..

^' -y%"' 1 ^


LL- .

over a larger surface area. This procedure was used in all experiments

where the effects of lower than maximum power densities were


The optimum focal length for each lens was determined

experimentally by exposing thermal paper (in air) to the partially

focused beam at distances equivalent to 10, 60, 85, 135, and 195

percent of the expected focal lengths. The average spot diameters from

each point were plotted against the distance from the back edge of the

lens. The intersection of the two lines drawn through the points gave

the optimum focal length (Figure 2). The value determined for the two

lenses were 199 mm and 254 mm. These values were confirmed by

microscopic examination of target impact craters after exposure to the

highly focused beam.

After passing through the focusing lens, the beam entered the

reaction chamber through a 3.8 cm diameter zinc selenide window (II-VI,

Inc., #-746-7) which also had an anti-reflection coating for maximum

transmission at 10.6 microns. The distance between the lens and the

window was kept to a minimum in order to prevent damage of the latter

by interaction of the partially focused laser beam with surface

contaminants. This distance was ca. 0.5 cm for the 200 mm focal length

lens and ca. 2 cm for the 250 mm focal length lens.

Gas Handling System and Procedures

All of the gases used in this study, except one, were gaseous at

STP and were supplies from pressurized cylinders. Standard one and two

stage gas regulators and associated fittings were used to deliver these

























(un) Ja4aDWPa 4ods


gases to the mixing system through 0.635 cm o.d. polypropylene


The one compound that was liquid at room temperature, dicyclo-

pentadiene, was placed in a 50 ml glass impinger and helium gas from

the gas handling system was bubbled through 30 ml of the liquid at 120

ml/min before entering the plasma chamber. This effectively saturated

the gas with the volatile liquid. Teflon tubing and fittings were used

to connect the impinger to the plasma chamber during this experiment

(Figure 3).

The gas mixing system used in all experiments, except the one

noted above, consisted of the following components:

1) three variable area flowmeters with built-in needle valves

(Cole-Parmer models J-3217-12 and J-3217-18), mounted together on an

aluminum plate,

2) a 120 ml acrylic gas drying column with 0-ring sealed aluminum

end caps (Altech Associates 100 psi model, sold by Cole-Parmer as model

J-1418-100) half filled with indicating drierite and half filled with

molecular sieves,

3) a 100 ml glass bubble-tube flowmeter,

4) a brass on/off plug valve with elastomeric 0-ring seals (Nupro

model B-4P4T) and two brass fine metering valves (Nupro model B-SS4),

5) an oil-filled mechanical vacuum pump (Precision Scientific

model 25), and

6) a capacitance manometer (Baratron Type 220BA-01000A2B) with a

digital readout (MKS type (PDR-D-1).










7 c

*r- a> Q CT
t > Mr- (0
0-) 0 +
aE >

- Lf

A diagram of the system components and the gas flow pattern is

presented in Figure 4.

Gases were supplied to the flowmeters at 69,000 Pascals (10 psi)

pressure and the flow rates adjusted using the variable area flowmeter

needle valves. Copper tubes (1/4 in o.d.) from the flowmeter exit

ports were teed together a few centimeters downstream using brass

compression fittings. The gases then passed through the drying column,

which also acted as a mixing chamber. Nominal total flow rates were

200-400 ml/min. After exiting the drying column the gases flowed

through a teflon tee. One arm of the tee went to the bubble-tube

flowmeter (which was vented to the atmosphere through the building hood

exhaust), and the other arm went to the plasma chamber on/off valve.

Thus, when the on/off valve was closed, the total gas flow passed

through the bubble-tube flowmeter. When the on/off valve was open,

part of the gas stream (about half) flowed through the plasma chamber

while the remainder was vented through the bubble-tube flowmeter. This

arrangement assured that no atmospheric gases were introduced into the

plasma reaction chamber through the gas handling system.

Flow rates were set using the flowmeter needle valves and measured

using the bubble-tube flowmeter and a stopwatch while the chamber

on/off valve was closed. When more than one gas was used for an

experiment, the flow rates were determined by addition, starting with

the lowest flow and ending with the highest flow. The bubble-tube

flowmeter was also used to monitor the excess gas flow to the

atmosphere during plasma reaction experiments. The variable area











0 uV E

Q. U

*r- 0W

a0 -




cn C


-0 U

L 1

w -' s~r >

*- E

n3 C
c cr-





flowmeters were used to monitor the individual gas flows (comprising

the reaction mixtures) during experiments.

The gas flow to the plasma chamber was controlled by one or both

of the metering valves in most experiments. These valves were located

just downstream from the on/off valve and were used to separately

control the gas flow through two inlets into the plasma chamber. (One

inlet was located at the bottom of the chamber, beneath the target, and

the other inlet was located on the optic side arm adjacent to the zinc

selenide window. (See the plasma chamber description for further

details of the gas inlets.) In experiments where the chamber pressure

was 530 Pascals (4 torr) or greater, the gas flow was split equally

between the two inlets using the respective needle valves. In

experiments where the pressure was 133-533 Pascals (1-4 torr), one

needle valve and inlet were plugged and the entire gas flow was

directed through the optic side arm inlet. In experiments where the

chamber pressure was < 133 Pascals (< 1 torr), the Nupro metering

valves were replaced by an in-house fabricated, low-flow, 1000:1 taper


All experiments used the mechanical vacuum pump to provide the

driving force for gas flow through the system. The plasma chamber

pressure was monitored with the capacitance manometer which had a range

of 0-1000 torr in 0.1 torr increments (0-133,323 Pascals in 13 Pascal

increments). The manometer was calibrated using a standard mercury-

filled McLeod gauge.

The Plasma Reaction Chambers

Two plasma reaction chambers were used during the course of this

study. Their function was to provide a controlled environment for the

interaction of solid gold and gaseous compounds and elements. An

initial screening study was conducted using an all glass reaction

chamber. The remainder of the study was conducted using a more

versatile aluminum reaction chamber. A description of the all glass

chamber is presented first, followed by a description of the aluminum


General features common to both chambers included a target

holder/manipulator for positioning the solid gold target, a zinc

selenide window for introducing the laser light, gas inlets for

introducing the reactant and inert gases, and a gas outlet for

delivering the reaction products and unreacted gases to the product

collection system and the vacuum pump. The major difference between

the two chambers was the design pressure rating. The glass chamber

used in the screening study was designed to operate in the pressure

range of 667-186,650 Pascals (5-1,400 torr), while the aluminum chamber

used throughout the rest of the study was designed to operate in the

pressure range of 0.133 to 93,000 Pascals (0.001 torr to 0.9


The glass, plasma reaction chamber. This chamber was designed to

be utilized for initial screening of prospective gases and gas mixtures

for their effectiveness in producing and transporting gas phase gold-

bearing species. This involved the determination of optimum pressure

and laser energy density conditions for gases showing positive results.

The protocol also included the study of the interaction of the focused,

pulsed CO2 laser beam with several inert and reactive gases and gas

mixtures at pressures approaching 200,000 Pascals (ca. 2 atm.). The

transparent chamber facilitated the observation of gas breakdown. This

phenomenon resulted in absorption of most of the laser pulse energy by

the gases, leaving very little energy to vaporize the solid gold. (A

further discussion of this topic is presented in the following


A diagram of this plasma reaction chamber is presented in Figure

5. The two-piece borosilicate glass body was fabricated in the

Scientific Glass-Blowing Shop at the University of Florida and had a

total volume (including side arms) of 588 ml. Two 50 mm i.d. glass 0-

ring seal joints (Kimax No. 50) were used as the base structures. One

of the joints was modified by tapering the blank end and butt-sealing

it to a 12 mm o.d. by 9 mm i.d. by 4 cm long piece of glass tubing.

The overall length of the resultant bell-shaped structure was 20 cm. A

9 mm o.d. by 6 mm i.d. by 14 cm long piece of perforated teflon tubing

was press-fitted into the narrow end of the bell structure from the

inside. This tube extended about 1 cm into the 12 mm o.d. glass

extension tube. The rest of the teflon perf-tube extended through the

center of the bell structure and ended about 2 cm from the wide end. A

50 mm dia. by 2 mm thick teflon perforated plate was then attached to

the teflon tube by threading a stainless steel screw through the center

hold of the perf-plate and into the end of the teflon tube. This unit

comprised the main gas inlet to the plasma chamber. The perforated










> 1
4- U
s- c
LC a

0 E E-


'n a

M \n


cn rc :
r-'- a)- '
O a rI
' I .4-)


c % m F=

Ln 0 0 (A

>E ( r- ,

CP 4- E (A

tube and plate distributed the gas flow evenly to the plasma reaction

zone. A length of copper tubing coming from one of the gas handling

system metering valves was attached to the narrow end of the glass,

gas-inlet structure with a brass, 0-ring sealed, quick-connect fitting.

The other 50 mm i.d. glass 0-ring joint was also tapered and butt-

sealed to a 12 mm o.d. by 4 cm long piece of glass tubing to form a 20

cm long bell-shaped structure. The narrow end was the plasma chamber

outlet and was attached to the product recovery system and a mercury

manometer using a brass quick-connect fitting. Another 12 mm o.d. by 4

cm long piece of glass tubing was sealed to the large diameter sidewall

of the bell structure 5 cm (on center) from the mouth and served as a

guide shaft for the sample holder. A 20 mm i.d. glass 0-ring joint

(Kimax No. 20) was sealed to the opposite sidewall 6 cm (on center)

from the mouth and served as the optic side arm attach point. The

offset center lines allowed the use of a large diameter target that was

rotated past the beam impact point.

The sample holder shaft was an aluminum rod 15 cm long by 9 mm

o.d. that was slip-fitted through the chamber side arm, as mentioned

above, and sealed by sliding a length of tightly fitted, 3 mm thick

silicon rubber tubing over the joint. The part of the aluminum shaft

that contacted the silicon rubber tube was coated with vacuum grease

(Dow Corning silicon high vacuum grease) to facilitate sealing during

shaft rotation. A 7 cm dia. aluminum cogwheel was attached to the

exterior end of the aluminum shaft and used to rotate the sample

manually. The other end of the shaft extended into the chamber and had

a threaded center hole that was used as a target attach point. The


shaft, and target, could be moved toward the beam input opening a

maximum of 3 cm in order to shorten the optic path.

The two piece optic side arm was constructed from glass and

aluminum. The piece closest to the plasma chamber was made from a 20

mm i.d. glass 0-ring joint butt-sealed to a 30 mm i.d. glass 0-ring

joint (Kimax No. 30). There were actually two such pieces fabricated.

One was 5 cm long and the other was 8 cm long. This provided different

path lengths between the zinc selenide window and the target surface.

By exchanging these pieces, and moving the target back and forth, the

optical path length between the backside of the window and the target

surface could be varied by 6 cm. This allowed adjustment of the

incident energy density of the laser beam while maintaining a minimum

distance between the external focusing lens and the window. (This

arrangement minimized the probability of damage to the window surface,

as previously mentioned.)

The aluminum window holder was machined to match the 30 mm i.d.

glass 0-ring joint on one end. The other end was machined to hold the

zinc selenide window described earlier. The window was sealed using a

rotary 0-ring gland that incorporated the window edge and the inside

circumference of the aluminum holder as sealing surfaces. Two teflon

rings with inside diameters 2 mm smaller than the window o.d. were used

as spacers on either side of the window. The window and spacers were

held in place using an aluminum ring with the same i.d. as the teflon

rings and an o.d. equal to that of the aluminum holder. This ring was

attached to the end of the holder with four screws. A brass swage

fitting (1/4 in. pipe thread to 1/4 in. tube) was located 1 cm (on


center) from the edge of the window seat on the side of the holder. A

1/4 in. copper tube coming from the gas handling system's other

metering valve was connected to this fitting and provided a sweep of

incoming gases across the window surface to help minimize contamination

of the surface.

Butadiene elastomeric 0-rings were used where required throughout

the system. Pinch clamps held the glass 0-ring joints together when

the system was not under low pressure. The rotary seal on the window

and the seal on the sample holder shaft had moderate leak rates and the

chamber could be pumped down only to 13 Pascals (0.1 torr). While this

operational parameter met the rough-vacuum design criterion, the affect

of the residual pressure of air on the proposed reactions was unknown.

Therefore, a study of the affect of increased levels of air in

successful gas mixtures was included in the screening study protocol.

The affect of very low concentrations of residual air in successful gas

mixtures was investigated using the aluminum reaction chamber.

The aluminum plasma reaction chamber. This chamber was fabricated

by the author from aluminum (Alcoa alloy 2024) using available

facilities in the Department of Chemistry at the University of Florida.

A diagram of the chamber, including the sample holder, optic side arm,

and observation ports is presented in Figure 6. The screening study

was completed, and all other experiments were conducted using this











0 0

>S -

3 V)
10 r_

0E r

0 >,
+ u)


(U C
.- (

ru E

Li) t
D- -




S- c0
W >


4-) E
0 O(
S- (V)


SL m E
4- C a

10 *- >
m E Ln

The aluminum plasma reaction chamber consisted of five parts:

1) the main body, including two gas inlet inserts, a gas outlet

flange, and a capacitance manometer insert,

2) two glass observation ports and their flanges,

3) the baseplate,

4) the optic side arm, and

5) the sample holder.

All of the components were fabricated from aluminum except the glass

ports and the main body inserts and outlet flange (which were machined

from brass). Viton or butadiene elastomeric 0-rings were used for all

seals. All tubing, fittings, and inserts leading from the gas handling

system and capacitance manometer that were subject to vacuum were

either brazed or connected with 0-ring quick-couples.

The completed system was leak checked using a mass spectrometer

helium leak detector (Veeco Leak Indicator MS17) previously calibrated

with a standard leak (Veeco Sensitivity Calibrator Type SC-4). The

detector response indicated that the leak rate was < 2 x 10-9 cc-

atm/sec (air equivalent) at each union and overall under He flood


The main body was machined from a single block of aluminum 9.5 cm

wide by 7.6 deep by 15.2 cm high. The reaction zone was formed by

boring a 5.1 cm dia. by 9.5 cm deep hole through the center of the

block, starting on the bottom face. The top of this cavity was tapered

300 and narrowed to 1.0 cm in two steps before exiting the center of

the top face of the block. Static face seal 0-ring glands and four-


hole bolt circles were machined around both holes for attachment of the

gas outlet flange and the chamber baseplate.

Several different outlet flanges were used to connect the plasma

chamber to the various product recovery systems. All of the flanges

were sealed to the chamber using the 0-ring face seal and connected to

the product recovery system by an 0-ring quick-connect type coupler. A

1.2 cm dia. hole was bored on one of the 9.5 cm wide sides 2 cm (on

center) from the bottom of the chamber. A 6 cm long insert with a

rotary 0-ring gland was inserted through this opening. The insert had

a 6 mm hole bored from the incoming end and turned 90' at the outlet

end. This design allowed the incoming gas to be directed toward or

away from the plasma reaction zone by rotating the insert. Another 1.2

cm dia. hole was bored in the same face 9 cm (on center) from the

bottom of the chamber. A 6 mm i.d. plug insert with a rotary 0-ring

gland was used to connect the plasma reaction zone with the capacitance

manometer via a 30 cm length of 1/4 in. o.d. tubing.

A 2.0 cm diameter hole was bored into one of the 7.6 cm wide faces

5.5 cm (on center) from the bottom of the chamber. The 0-ring gland

and four-hole bolt circle were machined around this opening. This was

the optic side arm attach point.

A 3.3 cm dia. hole was bored into the other 7.6 cm wide face 4.7

cm (on center) from the bottom of the chamber. The sample holder was

inserted through this opening. A rotary 0-ring gland was machined into

the circumference of the opening 3.0 mm from the outer edge. This

0-ring sealed the sample holder even when it was being rotated. A


four-hole bolt circle was machined around the port to secure the sample

holder with its 0-ring face seal.

Two 5.0 cm dia. ports were cut into the 9.5 cm wide sides directly

opposite from each other and 5.0 cm (on center) from the bottom of the

chamber. Static face seal 0-ring glands and four-hole bolt circles

were machined around each port. Two 1.0 cm thick port flanges with

matching 0-ring glands and bolt holes held and sealed 6.0 cm dia. by

6.0 mm thick lead glass windows over each port.

The baseplate was 12.7 cm square by 2.5 cm thick and had four bolt

holes that matched the bolt circle on the bottom face of the chamber.

In addition, the baseplate had four holes at the corners that were used

to attach the chamber assembly to a table. Block spacers were placed

under the baseplate to raise the chamber optic arm to the same height

as the laser beam.

The single piece, 3.0 cm i.d., optic side arm used the same type

of rotary 0-ring gland as described above to center the zinc selenide

window. An 0-ring face seal gland was cut into the aluminum holder and

used to form the primary vacuum seal against the face of the window. A

precisely machined teflon spacer, and an aluminum ring flange similar

to the one described above, provided the necessary squeeze to form the

seal without unduly stressing the window. A 6.0 mm dia. hole was bored

into the side arm 5.0 mm (on center) from the window face seal edge and

used as the sweep gas inlet. A brass insert with a rotary 0-ring gland

connected the inlet to one of the gas handling system metering valves.

The optic side arm was attached to the plasma chamber using the bolt

circle and face seal described earlier.

The sample stand was also made from a single piece of aluminum.

The sample attach point was a threaded hole in the end of a 1.4 cm dia.

by 2.3 long section of the stand. This section ended in a 3.3 cm dia.

by 1.5 cm long section that acted as the sealing surface for the rotary

seal in the chamber throat. The outer section of the one-piece stand

was 7.5 cm in diameter and 1.3 cm thick and acted as the flange. This

section contained a face seal 0-ring gland that formed the primary

vacuum seal, and a series of arced bolt slots that matched the bolt

circle on the chamber. The rotary seal allowed the sample to be

rotated manually while under vacuum by loosening the face seal bolts.

(Depending on target thickness, this was done once or twice an hour

during most experiments.) The bolts were retightened after the sample

rotation was completed. This was necessary because particles and

frictional wear frequently caused minor leaks in the rotary seal.

The Product-Recovery/Substrate-Coatina Systems

There were several product-recovery/substrate-coating systems used

throughout the study. Many of the systems used common components

assembled in different sequences and maintained at different

temperatures. Many of these arrangements were designed to investigate

the volatility, or other physical properties, of the plasma reaction

products. The system descriptions are presented below along with their

respective operational parameters. Results are discussed in Chapter


Heated auartz tube. A 2 mm i.d. by 4 mm o.d. by 50 cm long quartz

tube was attached directly to the glass plasma chamber outlet using a


silicon rubber sleeve seal. The tube was placed in a Linberg tube

furnace with a 30 cm long high temperature zone. The furnace was

heated to 600-890 K during experiments in order to decompose any gold

bearing species that passed through. The gas flow rates were adjusted

to allow a residence time of 3-194 milliseconds in the hot zone. In

practice this resulted in visible deposits varying in color (from black

to purple to red to gold) and position along the tube. In some experi-

ments the quartz tube was followed by a glass wool plug and/or a cold

finger (size 14/22 or 24/40) in a dry-ice/acetone bath (see Figure 7).

Staged cold traps. Several experiments used multiple cold traps

at progressively colder temperatures to observe product collection

characteristics. The various schemes used are pictured in Figure 8.

The first such scheme used a cold finger (CF) in water-ice (273 K)

followed by a second CF in dry-ice/acetone (195 K), and a third CF in

liquid nitrogen (77 K). Clogging of the CF in liquid nitrogen resulted

in its replacement with a second CF in dry-ice/acetone. The next

modification incorporated a glass wool plug before the cold traps.

This scheme was then modified by placing the glass wool between the

first CF, which was kept at room temperature, and a second cold trap

made from a 1/4 in. (6.35 mm) o.d. by 25 cm long glass U-tube kept in

crushed dry ice. Another scheme used a small test-tube inserted in the

gas stream at the plasma chamber outlet ahead of the U-tube in dry ice.

The test-tube was either maintained at ambient temperature or was

filled with dry ice. The flange which held this tube could also

accommodate a quartz window in place of the test-tube. Finally,












I (fl-&-

1 Cfl-


I 0













to .-

Staged Cold Traps Used for Product Recovery
(a) Three cold fingers
(b) Cold finger with U-tube
(c) Test-tube with U-tube

Figure 8.

silicon rubber



glass test-tube

glass wool plug

gas flow from
plasma reaction zone


- b to


the system was limited to the U-tube in crushed dry-ice (or dry-

ice/acetone) since this proved sufficient for trapping the plasma

reaction products for later use. A fresh trap was used for each

experiment. The 1/4 in. o.d. glass U-tubes were eventually replaced

with 3/8 in. (9.525 mm) o.d. glass U-tubes. One of these U-tube cold

traps was used in all subsequent experiments either alone or after a

substrate coating chamber.

Substrate coating chambers. Substrates were coated with the

plasma reaction products during experiments by sticking them to the

inside of the chamber windows (using Scotch clear adhesive tape), by

placing them in the connecting tubing leading to the U-tube cold trap,

or by placing them in one of the three coating chambers pictured in

Figure 9. The first two chambers were fabricated from glass in the

Scientific Glass Blowing Shop at the University of Florida. The third

chamber was constructed from aluminum by the author using available

facilities in the Department of Chemistry at the University of Florida.

All of the chambers were placed in line after the plasma chamber and

before the U-tube cold trap. The two glass coating chambers were

mounted perpendicular to the plasma chamber via an elbow quick-couple.

The aluminum coating chamber was connected directly to the top of the

plasma chamber using a straight quick-couple.

The first chamber was made by truncating a 20 mm i.d. glass 0-ring

seal joint (Kimax No. 20) 3 cm from the sealing end and attaching 1/4

in. o.d. by 5 cm long glass tubes on opposite sides, perpendicular to

the seal face. These tubes acted as the inlet and outlet. A 25 mm

dia. by 1 mm thick quartz window was placed on the 0-ring and held in

Substrate Coating Chambers
(a) Small glass chamber showing a ceramic feed-
through in place
(b) Large glass chamber showing the 3-tier
sample rack

Figure 9.

,5--quartz window

in -

(a) ceramic -

40 mm 0-rinq joint -


3-tier substrate
(b) holder

to thermocouple


to heater controller

FI > gas out

perforated teflon
- plate (for filter
catch experiments)

heated substrate

gas in

place by the system vacuum. Total volume of the chamber was 20 ml.

Substrates could be placed in this chamber and exposed to a 5 watt

Argon ion laser beam (Spectra Physics 165) while plasma reaction

products flowed through.

A second, larger coating chamber was constructed from two 40 mm

i.d. glass 0-ring seal joints (Kimax No. 40). The joints were tapered

and sealed to 1/4 in. tubing on the ends. The two halves were held

together with a pinch clamp when the system was not under vacuum. The

assembled unit was 20 cm long and had a total volume of 145 ml. A

three-tiered substrate holder rack was made by the author using three

38 mm dia. by 2 mm thick perforated aluminum plates connected by a

central stainless steel shaft. Substrates were placed on the different

rack plates in either equatorial or axial positions in order to observe

shadowing effects during coating.

A third, more versatile coating system was machined from aluminum

stock (Alcoa alloy 2024). This chamber was also a two-piece design and

had an 0-ring face seal between the halves. The bottom half started

with a 1/2 in. (12.7 mm) o.d. by 1.9 cm long inlet segment that sealed

directly to the top of the plasma chamber via a 1/2 in. brass quick-

couple. This section had an inside diameter of 1.0 cm. The next

segment of the bottom piece had an o.d. of 5.0 cm and an i.d. of 4.0 cm

and was 4.4 cm long. The final segment was 7.6 cm wide by 1.2 cm thick

and contained the 0-ring gland, a four-hole bolt circle, and a seat on

the inside diameter for a 4.0 mm wide by 3.0 mm thick teflon ring

spacer. This segment acted as a flange to connect the two pieces of

the chamber.


The top half started with a matching flange segment with a 3.0 mm

thick by 48 mm dia. teflon perforated plate press-fitted into the 4.0

mm wide seat on the inside diameter. This was followed by a 5.0 cm

o.d. by 4.0 cm i.d. by 1.9 cm long segment and a 1.0 cm i.d. by 1/2 in.

(12.7 mm) o.d. by 2.3 cm long outlet segment. The assembled chamber

was 13 cm long and had an internal volume of 87 ml.

The teflon perforated plate and spacer were used to hold a 47 mm

diameter borosilicate glass fiber filter that had a retention rating of

0.98 at 0.2 microns (Micro Filtration Systems GB 100 R47 mm). The

filter faced the gas flow and was backed by the perforated plate. The

teflon ring held the filter in place without damaging it. This

arrangement was used to trap particles that were produced in the plasma

chamber. Particles that were transported 20 cm up from the plasma

reaction zone, through the narrow plasma chamber outlet and up to the

filter, were very efficiently trapped. Other particles that impinged

on the plasma chamber sidewalls or fell to the bottom of the chamber

were not collected by the filter. Most large particles ("chunks" and

"flakes" tens of microns in size) that made it to the filter rebounded

on impact and were found in the bottom of the collection chamber.

Large particles that adhered to the filter were easily distinguished

using visual microscopy. Thus, the arrangement provided a semi-

quantitative method for comparing relative abundances and gold content

of particles entrained in the gas stream using gravimetric and atomic

absorption analyses, respectively.

Heated substrate mount. Several experiments were performed with

substrates attached to a heated mount shown in Figure 9. The 3.5 cm


dia. by 3.0 cm high mount consisted of a potted nichrone heating

element covered with a 2 mm thick aluminum plate that had a series of

threaded holes. Flat head screws were used to attach substrates to the

plate using the threaded holes. The mount was positioned in the top of

the aluminum collection chamber with the heated substrates facing the

gas flow. Three small ceramic posts minimized the area of contact

between the mount and the top of the collection chamber.

The heating element wires extended through the top of the chamber

through two glass inserts that were epoxy sealed (J. B. Weld Co.,

Sulphur Springs, TX) to the aluminum. Rubber septa and sealing putty

(Apiezon Q) were used to seal the wires at the exit of the glass

inserts. There was a small leak at this point, ca. 0.01 ml air/min

determined by differential pressure measurements. However, its

location downstream from the plasma reaction zone and the substrate

coating zone, and the results of earlier experiments that showed the

relative insensitivity of selective plasma reactions to air, minimized

concern over the leak.

Temperature control for the heated mount was provided by a 12 V dc

power convertor driven by a variable voltage auto-transformer.

Temperatures were controlled to + 5% of the set point. A thermocouple

junction (type K) was implanted in the aluminum hot plate and connected

to a digital multi-meter (Kiethly 177 Microvolt DMM). The thermocouple

was calibrated using water ice and boiling deionized water.


Production of Gas-Phase Gold Species: Experimental
Conditions and Materials

Screening Study Experiments

The purpose of the screening study was to identify gases and/or

gas mixtures, operating pressures, and laser power density parameters

that resulted in the production of relatively large amounts of gas-

phase gold species. The screening study was continued throughout the

duration of the overall study. The product recovery systems described

above were used to separate the gold species from the gas stream.

Visual observation was used as an initial method of evaluation since

the relative abundances of gold species collected in individual

experiments varied drastically. Cold traps containing the recovered

products were sealed and heated in a bunsen burner and then opened to

the atmosphere and reheated. Observations of the relative quantities

of gold films formed on the glass surfaces were utilized to identify

successful operating conditions.

Glass fiber filter catches were determined gravimetrically using a

Mettler analytical balance. (Filter blank measurement methods and data

are presented in Appendix A). Quantitative measurements of the amounts

of gold collected in some of the cold traps, and on most of the glass

fiber filters, were performed using Atomic Absorption (AA) analyses

(Perkin Elmer Model 303 AA) and standard analytical techniques.64

Samples were either extracted/ suspended in acetone, or dissolved in

aqua regia and diluted. Calibration data and response curves are

presented in Appendix C.


The initial gold target used in the screening study was formed by

melting and casting ca. 10 grams of Au wire (0.999 purity) into a 2-3

mm thick crescent shape. A machined graphite block was used as a mold

for the casting. A Canadian Maple Leaf 1 oz gold coin (0.999 purity)

was also used as a target during the screening study and in all

substrate coating experiments. This target required recasting after

extended use, but its total mass was never less than 23 grams, and its

thickness was never less than 3 mm. A 1.0 cm thick by 2.5 cm diameter

piece of graphite (mass = 23.7 grams) was used as a target in two

experiments and a 6 mm thick by 18 mm wide by 28 mm long OFHC copper

target (mass = 28.0 grams) was used in two other experiments.

Gases used in the study were obtained from several different

sources. The gases are listed in Table 1 along with their respective

vendor sources and purity ratings. No attempts were made to further

purify any of the gases beyond the removal of excess moisture in the

gas drying column already described, with one exception. The helium

gas used in test No. 861126B was passed through two liquid nitrogen

cold traps in order to remove traces of water which otherwise would

have reacted with the liquid dicyclopentadiene, causing it to


Table 2 lists all of the experiments contained in the screening

study along with their respective experimental conditions. Each

experiment was given a six numeral identification number based on the

year, month, and day that it was performed. A letter was added to the

numerical code when more than one experiment was performed in the same

day. For example, "860602A" refers to the first experiment performed

Table 1. Gases Used in the Plasma Reaction Study

Gas Vendor Purity Rating











carbon monoxide


natural gas



(laboratory ambient)




Aldrich Chemical

Allied Chemical Corp.

Allied Chemical Corp.

PCR, Inc.


Matheson Specialty Gases

Matheson Specialty Gases

Gainesville Regional
Gas Company (through
laboratory lines)

standard grade

standard grade


breathing quality

standard grade

standard grade

reagent grade

refrigerant grade

refrigerant grade


welder's grade


99.9% and 99%

commercial grade

I- u
0" 0)

~ E


-) cuE

0 U 0
I- a) >

GO 0
0 3


0 W*-'
C u


S *r- *
0 EC



9- r E0

s. a.

000000000 -4 C)0 C') 00 to 0 0

tio n mc 4000i c b02r -.(I-iC% iVo cJ-e
r-4-4ro-4 -4-?" -4 r4-4 r-4


0 c

Cr U

)N in
rh c^

00 O
C*-l C*


E EE 5 E
4J +- 4-) 4-1
. Q- *0. *- 0.
0 00 0 0

S- 0


n Lm
C l








Qj 0




CL) r-

0 c 0 C C

Ic I k- CCa to0c
1 wl



* r- *. r *. *. .r- *- *.- *. *r r- *r *-
+J 4-4+4-)+4- 4J+-J4 4J 4 4-l- 4-)4 4-J
0 0 00CL0.00QQ.0.0- L .MCl

4-; S-
C. 0 tv
* 4-)
LU _



- 0
=> r

>3 (1)
S0 +-j

S- 4) >>


) 0




0) 0 4)




CX 4-
S- 0-E


Q; Ur C

c U
0 f:'^*-

in oDo Cn
CN ON -- n
c'j 09 v-I


4) *-4


co10 r-


in In
S.. 9.I
3: 4
'*^in i

s-~ r~ i


rO C0 10 -.

o o o0 C<

E- E E E-



C ,

cv) z




- cM. cMj CM CM
10 "-~ 0 *-" .4' CM \J f ~
03 3 3 3 3
0c 0'" 02 02 Co 02
CO CO 0 00' 00

D C 0 >0 D 00

--c'.CM r. 02 C'.J -.
0 *.** *0 NM

c 0m 0w30 LOn o 0
W)0'-i 0'.O' 02

C0 <
00 -




-C34J4 4-4J4J+-+J 4-

3 3

0 0-
0 0

0D to

co) cv)


00 0
1 10 <*>

s E


LO 4 02
ro o c




) ,-4


j- S- I-

a-> I- I-- I- 0' aocy ao
U, E f C (A
o a n E in e S L
oS- 0 4 m a $m (A m n

O 0 N 0 O N
ae o c OO O a QI o c

a03 m O O 0 in 0 -I
*i- X -

C --
0o0 E E E E E E E E

02Jr 3 3 4 )
3 c E E E E E E E E

"- L JUE a6 C 0m CL .
-O 00 0 0 0 0 0 0 0 0
0 U-

0D ca
C 4-J

w Ur- C -

) -->. *
0 O? 0 0 0 0 0 0
3( C C CD 0 0 10 0 C) 0
Cl In c"+. m C" to C> I-n 10
w) a) LO m .4 1- .c m

Q- CL 0.


0 CM U C'cri- CCM Cm3

=- 32 3- ) =- U 3 ) = 3 -'=-u

0) O 1 In OD CM CM
CL 0 tv -0 =I CD 0 0
0.003 -3 J- -l 033 -3 03 03
XZ4-) 1.0 to <.cC IDCC <' Lal -t 0~ -~ O~ 2


u U U -u
mm Ln CL m- c-
m "- 3m 3 0= 3 o -
c-4j CD-4 CD PL c =D3 CD

= =3

L I-
0 e

I- C) 0O
MV r -

CJ C> C=) C)
C'0 0) 0
to K C7 ON
10 4 CT U

0) (V)

o) S

X0 04>)


C C)
Q) C)

0 O



C) 0

0o c0

S- 0 E
= 1-4 U
) 0

30) E E

0) -

U) U)
S- Q

0- 4

+* 0)

CL 0 cc
X 4.-)

0 0 0

n O f- 0 3o

cl N CJ NC

o 0 C
CV C0 0C

t -4
toj co

C0 C) 0

0 O
o 0n CM
5- 5--

0 0


s- S- S-

I--M -M 1 31-1 LM L.) I--
o a0n 0 CO LO La bto <- < < 0

j c m to \ w ON

0l =3 0 -4 -1- -q C -4 =5 5-" '" -50
kD 0 5- 0- -- w w w- M -

t0 C%

- -4


0n *u- 4- 4-- *4- *-- 6 -
* cO L. Q. 0L Q. 0. L 0- .
00 0 0 0 0 0 0 0 0

C- I-

0 0
CI- 0C-






0 0 0 0

cN N (-

E E 5 E E E E E
= = =I = 3 I 3
*+2* 44- 4-3 4-) -1 4-3 4-)
S 0. 0. 0. Q0 0. 0
C(%-J 0 0 0 0 0 0 0 0


0 =
-- l- X

C u
0* 0 E

I- *0-

"a E

o -

0 0


0 )




x -
LjJ -

to 0 0 0
C) irn un ce
" in I- l

) to l 0 Co O) -
01 1-4

0 0 0 0
u, en nX ID
Cv) Cv Cv) tO
-4 -4 -

0--O O 0 0I\ \ 0CD\ ?\
-, Ln Co O \1to C-, -t \CJ" lC
CIOO ('. r. i C%. cM cC.-. C\Z.s C~xl=

co cM m cv) "It
C i CJ CJ C0 0 C
Cm s CO .. CMi CMJ C% -% C m CQ -% CV) CM -.

c ~ c col Coo< cos co tL co co
CO>* C O ^ C O -' C '' C O*- O -





U 4) E

S- 0) >,
9. 9' C6

0 O
C )

0 0

m M-

F- 1- -
=3 3 =?

0 ( CD C.3

-0 0 4-)

S- 4) >

C 0

0) 0
r- -




Uw 0
10 o
Cj E


o r_

- *r-


0) -(

0) 0
Q- *-

I- h- -

Ll Ll Ll L:

OLD L D t-

LO r a

43 43 3 4- =4 4 3

C. C. 0-. 0. 0- mr- 0.
0 0 0 0 O O 0 0 0

o o Ln L n
w O to to
U-3 "0

o (C in L n

0D 4 C) 0 0
0C 4 to -
0- 10

LO 0 0 LO 0
0J I --

0 0 0 0 0 0 0 0 0
P r cn aw cn C> o r%. ^c
-t o c1. m cM
-4 -4 r" --I4 -4 -


-.CNI% c~- kD-
N-CJ ) 10t 1.. W- 0
LC) 0C 0 C Co oiw C') w- CJCJ-

o co '-, LC n q~r to '*^ -ri csxl MT C CM. a)r^ or Q) r> a) co a) P~.

C 4 C)1 s4- r- 0)m-i Wn 0)N
en _e M =n m -At -c lc

C) :3 CC) CCD C C40 CDD

0o 00 00C 00 0 CO CO 1 00 c 0 0 00 '"0= 0
-O o\ \ \ -\ o\ -\ w U \






0 0 CO 0
10 0 0C 10t
4 *- 0c'
cM C C4



> a)

O 0
s- w) >,

0 0

c -- *
4j) r


a )0


CD 0

0 u


e *)u
c U

V) 0A

S- O-

I s- 5-


t- s"
O o S O S
C .0 CM 0L C 0.
Nr- Ns .. i-.a

00 00 00


4-) +4-l
0. 0
0 0




-- to t .- *- *-


0 0CDCoCo 0 0 0
to co 00 in fl CA
r- P- LO r_


N0 00 0 0 00

N N A-l-l i-CO


*) -L

XZ c
0J 0 -t








11 L







= m

Un I4>

cl u
01 <


o+ w
i- a)


II l







3: '


on June 2, 1986. The gas mixtures, total pressures, and flow rates

through the plasma reaction system are also listed in this table. The

pulsed laser focal point diameter at target impact and the total number

of pulses for each experiment are included as well.

Volumetric flow rates were measured in ml/min at atmospheric

pressure using the bubble-tube flowmeter. Linear flow rates (L)

in cm/sec through the plasma reaction zone were calculated using the

following equation:

L = (F)(101,325/P)/irr2

where F is the volumetric flow rate in ml/sec measured at atmospheric

pressure, P is the plasma reaction chamber pressure in Pascals, and r

is the radius of the reaction zone in cm. Some values (listed in

parentheses) were estimated from measurements made during similar

experiments utilizing the same nominal gas mixture and pressure and the

same settings on the variable-area flowmeters.

The diameter (d) for the laser beam at the target impact point was

calculated using the absolute value of the difference (x) between the

focal length of the lens (f) and the measured separation of the lens

and target, and the measured diameter (D) of the unfocused laser beam:

(d/2)/x = (D/2)/f

The impact point diameter was varied by moving the lens closer to or

farther away from the target.

The first experiment listed in Table 2 investigated the onset of

gas breakdown in helium under pressures ranging from 18,800 to 161,500

(Pascals (141-1211 torr). The second test listed demonstrated the same

effect in air at 101,500 Pascals (761.5 torr). In both of these

experiments the laser beam was focused into the center of the plasma

chamber. The target and target stand were removed from the system.

The third experiment listed demonstrated gas breakdown in hydrogen at

106,390 and 110,900 Pascals (798 and 832 torr) pressure when focusing

the laser on a gold target. The focusing lens was moved back 3 cm from

the optimum distance in order to observe the effect on the gas

breakdown. This same phenomenon was also observed in several other gas

mixtures at lower pressures during the course of the study.

Observations and results are discussed in the next chapter.

Gases were supplied to the chamber under pressure during segments

of the first and third experiments when reactor chamber pressures were

greater than the ambient pressure. Flow rates were controlled by a

metering valve at the chamber exit, upstream from the mechanical pump.

At lower pressures (< 101,500 Pascals) in these and all remaining

experiments, the pressures and gas flow rates were controlled using the

gas handling system described earlier and the metering valve before the


The last four experiments listed in Table 2 were attempts to

generate pure carbonaceous and copper-bearing species in the plasma

chamber using the carbon and copper targets, respectively, and a 3:1

H2CH4 gas mix. The remaining experiments listed in Table 2 were all

attempts to generate gold species that would remain entrained in the

gas stream and subsequently be trapped in the product recovery system.

The reader is directed to the table for a detailed presentation of the

experimental conditions. The results of each experiment are discussed

in Chapter III.

Substrate Coating and Annealing Experiments

Several different types of substrates were coated with the plasma

reaction products during the course of the study. Experiments that

produced the coated substrates are listed in Table 3 along with the

respective substrates and their physical positions in the plasma-

reaction or product-recovery systems during the coating process. In

all of the coating experiments the pulsed CO2 laser was operated at its

maximum voltage level (40 KV) and the focusing lens was adjusted to

provide an optimum focus at the target impact point. All coating

experiments were carried out at a nominal pressure of 1,330 Pascals (10

torr). A nominal gas mixture of 75% hydrogen and 25% methane was used

in all of these experiments with four exceptions. Two experiments were

performed using 100% helium, one test used 78% helium and 22% carbon

monoxide mixture, and one test used a mixture of 75% hydrogen and 25%

natural gas.

Eleven different types of substrates were coated with the plasma

reaction products:

1) glass plates cut from standard 1 mm thick microscope slides
(Corning No. 2947), varying in size from 5 mm by 10 mm to 25 mm by 40


2) quartz optical windows, 1 mm thick by 27 mm diameter,

Table 3. *Substrate Coating Experiments

Expt. Total # Substrates Position in
No. of Laser Coated System
(x 1000)

861117 15.00
(78% He/22% CO)

(100% He)

(100% He)



870313 10.20
(75% H2/25% natural

870326 12.60

W coated polycrystal-
line (pc) Si chip

glass plates

W coated pc Si chip
with an etched area;
patterned W on single
crystal (sc) SI; glass

W ribbon coil

W ribbon

placed in U-tube inlet

taped to plasma chamber
windows; placed in bottom
of plasma chamber

W/Si chips taped to glass
plates; glass plates taped
to windows and placed in
bottom of plasma chamber

placed in U-tube inlet, Ar
ion laser beam directed
through glass tube onto
leeward area

placed in plasma chamber
outlet, Ar ion laser beam
directed through quartz
window onto W

870416 13.92

870427 40.74

871106 27.66

870721 63.12

870801 11.76

870819 23.10

870901 55.92

ceramic feedthrough

9 ceramic feedthroughs

glass plate

W on pc and sc SI
chips; patterned Al
on sc Si chip

glass plate

glass plate

3 patterned W, 2 pat-
terned Al on sc SI

placed in small


placed in glass coating
chamber (GCC) on three-
tiered rack

taped into GCC inlet

taped to first tier of
rack in GCC

taped into GCC inlet

taped into GCC inlet

taped onto glass plate
that was taped into GCC

Table 3 (continued)

Expt. Total # Substrates Position in
No. of Laser Coated System
(x 1000)

880527 33.55

880602 31.32

880617 27.56

880623 23.46

880629 20.58





3 W on pc Si and
3 W on sc Si chips

7 W on pc Si and
5 W on sc Si chips

3 W on pc Si and
3 W on sc Si chips

2 W on sc Si chips;
2 glass plates

3 W on sc Si chips;
5 glass plates; strip
of Scotch tape, ad-
hesive side up, with
an area of W thin film
stuck to the adhesive

2 charged aluminum
plates (+275 and -125)
volts dc potential)

Si chip with a central
strip of W on one side
charged with -125 volt
dc potential; two 2.5
cm dia. quartz optical
flats; 4 glass chips

taped to perf-plate in the
aluminum coating chamber

taped to perf-plate in ACC

clipped to 373 K hot plate
in ACC

clipped to 373 K hot plate
in ACC

taped to perf-plate in ACC

pressed into center of

placed near sidewall in

taped to plasma chamber

*All experiments
using a nominal

were conducted at a nominal pressure of 1,330 Pascals
gas mixture of 75% H2 and 25% CH4, except where noted.

3) tungsten ribbon, 1.5 mm wide by 0.2 mm thick (H. Cross


4) ceramic electrical feedthroughs, 1 cm diameter surface,

5) blanket coated CVD tungsten (500-600 nm) on a silicon wafer

with one polycrystalline (pc) side and one single crystal (sc) side

(Harris Corporation) cut into 5 mm by 15 mm chips,

6) same as (5), with areas of tungsten etched away (1 M

K3Fe(CN)6:0.1 M ethylene diamene) to expose the silicon surface,

7) VLSI circuit patterned CVD tungsten on sc silicon with silicon

dioxide surfaces surrounding the 5-50 micron size tungsten features

(Harris Corporation), cut into 5 mm by 15 mm chips,

8) VLSI circuit patterned CVD aluminum on sc silicon with silicon

dioxide surfaces surrounding the 5-50 micron size features (Harris

Corporation), cut into 5 mm by 15 mm chips,

9) aluminum plates, 10 mm by 20 mm by 1 mm thick,

10) Scotch brand adhesive tape, and

11) Scotch brand black electrical tape.

All of the substrates, except the two polymeric tapes, were cleaned in

spectral grade acetone, rinsed in spectral grade methanol, placed in

clean glass petri dishes, and dried in a 383 K oven for at least one

hour. After carefully mounting the cooled substrates onto the plasma

chamber windows or in the substrate coating chamber using rolled Scotch

tape, they were rinsed off with dry nitrogen and placed in line.

Coated substrates were stored in air in glass petri dishes or plastic-

stoppered glass vials.


In most of the experiments the substrates were positioned in one

of the product-recovery/substrate-coating systems described earlier and

coated at room temperature (ca. 300 K). In one experiment a 5 cm long

piece of tungsten ribbon was coiled and placed in the inlet arm of the

glass U-tube cold trap and irradiated with an unfocused cw Argon ion

laser beam (4.5 watts, 514.5 nm). In another experiment a 1.5 cm long

piece of the tungsten ribbon was placed in the top of the plasma

reaction chamber at the exit union with the product recovery system

flange (see Figure 9) and exposed to the unfocused Ar ion laser beam

through a quartz window. This caused the ribbon to glow a dull red

when the chamber was under vacuum. When the system gas flow was turned

on the glow disappeared. In both experiments the laser beam was

impacting the leeward face of the ribbon. These were attempts to

observe thermal decomposition of gold species entrained in the gas

stream and subsequent deposition of gold along the thermal gradient

induced by the laser beam.

Two experiments utilized the hot plate described earlier to heat

glass plates and CVD tungsten coated pc and sc silicon chips to 373 + 5

K during the coating process. Problems associated with channeling of

the gas stream around the hot plate limited the efficiency of this

process, but several substrates were effectively coated in each

experiment and then subjected to spectral and microscopic analyses.

A list of the coated substrates that were annealed and the

respective annealing conditions is presented in Table 4. Some of the

tungsten coated and patterned chips, aluminum patterned chips, glass

plates, and quartz windows that had been coated with plasma reaction

Table 4. *Substrate Annealing Conditions

Expt. Substrate Annealing Time Temp.
No. Method (K)

(78% He/
22% CO)


plasma chamber
glass window

W on pc SI chip;
patterned W on
sc Si chip

870326 W ribbon

870416 ceramic feed-

870427 ceramic feed-

870721 W on pc and sc Si;
patterned Al on sc
Si chips

870901 W and Al on sc Si

880224 glass fiber
filter (GFF)

880229 GFF

880304 GFF

(100% CH4)

(48% H2/
52% CH4)



pulsed CO2 laser,
40 KV, unfocused

--radio frequency
(rf) coil (in
flowing air);
--rf coil in still

Ar ion laser, 4.5
watts, unfocused

60 pulses

10 sec
60 sec

70 sec

211 min
(in situ)

slow convective 141 mi
heating (cv.h.)
in air while moni-
toring electrical

same as above, ex- 57 mi
cept under H2 flow

cv.h. under H2 flow 18 hr
+ 1 hr
+ 2 hr

cv.h. under H2 flow 18 hr

cv.h. in air 1 hr
+ 3 hr

cv.h. in air 24 hr
+ 1 hr

cv.h. in air

cv.h. in air

cv.h. in air

8 hr

8 hr

8 hr

dull red





@ 563
@ 723


ca. 400


Table 4 (continued)

Expt. Substrate Annealing Time Temp.
No. Method (K)



880527 W on pc and sc
Si chips
W on pc and sc
Si chips

880602 W on pc and sc
Si chips

cv.h. in air

cv.h. in air

--cv.h. under H2 flow
--cv.h. in air
--cv.h. under H2 flow
followed by cv.h.
in air

--cv.h. in air
--cv.h. in air
--cv.h. in air
--cv.h. in air
--cv.h. under H2 flow
--cv.h. in air, fol-
lowed by cv.h.
under H2
--cv.h. under H2,
followed by cv.h.
in air

880623 W on pc Si chip
glass plate

880629 W on pc Si chip
and glass plate

--rf coil in air
--drew lines with
focused Ar ion ,
laser beam (3.5

--rf coil, 0.013
--cv.h. in air
--drew lines with
focused Ar ion laser
beam (3.5 watts)
(2.0 watts)
(1.0 watts)

3 sec red hot
0.3 cm/sec

1 min

85 min


0.5 cm/sec
0.5 cm/sec
0.5 cm/sec

880708 (-) dc potential
aluminum plate

cv.h. in air

3 hr

24 hr

18 hr
18 hr
18 hr
+19 hr

18 hr
18 hr
18 hr
18 hr
1 hr
18 hr
+ 1 hr

1 hr
+19 hr



90 sec



Table 4 (continued)

Expt. Substrate Annealing Time Temp.
No. Method (K)

880711 pieces of GFF cv.h. in air 25 hr 383
23 hr (473)
1 hr (676)
90 sec (875)

*All experiments were conducted at a nominal pressure of 1,330 Pascals
using a nominal gas mixture of 75% H2 and 25% CH4, except where noted.

products were annealed in air or hydrogen at temperatures ranging from

373-670 K for time intervals ranging from several seconds to several

hours. These anneals were performed in the tube furnace described

earlier. An annealing chamber was constructed from a 20 mm i.d. by 50

cm long borosilicate glass tube with 0-ring seal joints on both ends

(see Figure 10). The ends of this tube were connected to two matching

0-ring seal joints that had ends tapered to 1/4 in. o.d. tubing. Quick

couples were used to connect the chamber with the gas handling system

for delivery of hydrogen at ca. 400 ml/min. Anneals in air were

performed by disconnecting the chamber's gas inlet line and allowing

ambient air to be drawn through the system by convective flow. A brass

sample holder with a stainless steel insertion handle was used to place

and hold the substrates in the center of the tube furnace, adjacent to

the thermocouple.

Some glass fiber filters were also annealed in air at temperatures

ranging from 390-873 K by placing them in borosilicate glass petri

dishes and then into a laboratory drying oven or a muffle furnace.

This was done to observe the onset of a development process that

changed the color of the plasma reaction product coating from black to

purple to pink/red.

Several patterned and unpatterned CVD tungsten on silicon chips,

and one glass plate, were inserted into a radio frequency coil and

heated inductively. Either a 30-year-old ham radio with 100 watts of

power at 13.56 Mhz, or an ICOM All Solid State HF SSB tranceiver

feeding a Heath Kit tube-type linear amplifier at 14.000 Mhz and

delivering 250 watts of power, were used to power the 3 cm diameter 3







V) 4-)

c 0

f0 C
O *r-




turn coil. No temperature measurements were made in these tests, but

several samples were heated to a red glow. Exposure time to the rf

field was varied from 3-70 seconds. Some tests were performed in

ambient air with or without a fan blowing air past the sample, and

other samples were heated while under a pressure of 0.01 Pascals.

The unfocused pulsed CO2 laser beam was used to anneal a coated

chamber window and a coated quartz window. Exposures were varied from

1-100 pulses. The laser was operated at its maximum power level of 40

KV during all of these tests.

The Argon ion laser was used to write gold lines on coated

substrates by focusing the beam through a 75 mm focal length quartz

lens (Rolyn #11.0140) onto the coated surface. The laser was operated

at power levels of 3.5, 2.0, and 1.0 watts. Glass plates and CVD

tungsten on Si chips were taped to a carbon block and manually rastered

in front of the laser beam at the focal point. Pull rates were varied

from 0.3 cm/sec to 0.5 cm/sec. The focused beam was also fixed on

one spot on a CVD tungsten on sc SI chip for 10 seconds.

Product Volatility Study
The search for volatile gold compounds in the plasma reaction

products was divided into three categories, based on technique. The

first category of tests involved attempts to sublime gold compounds at

low pressure from the reaction products thermally and collect them in a

cold trap. The second category of tests utilized quadrupole and ion

cyclotron (ICR) mass spectrometers and several different methods of

sample delivery in an attempt to isolate and identify gold compounds.

compounds. The third category consisted of one series of experiments

using a time of flight mass spectrometer (TOFMS) to detect gold cluster

compounds formed in a Nd/YAG pulsed laser stimulated plasma using a

solid gold target and 1% methane in helium. Each experiment in the

volatility study is described below.

Sublimation Experiments

The two sublimation experiments that were performed used the

reaction products formed in a 75% He/25% CH4 plasma and isolated in the

product recovery system. The first experiment used the (195 K) U-tube

cold trap from 861018. This trap had been positioned after a 300 K

cold finger during the plasma reaction experiment. The second

sublimation experiment utilized the first cold finger trap (195 K), and

the glass wool plug contained in its outlet, from 861023. Both of

these cold traps contained milligram quantities of the dark-colored

plasma reaction products.

Both sublimation experiments were performed by attaching the

plasma containing cold traps (which had been sealed and stored at 195

K) to a fresh U-tube cold trap kept at 195 K. This trap was connected

to the mechanical vacuum pump, via a metering valve and a 195 K pump

trap, and the system was evacuated slowly to < 1 Pascal. In the first

test the products were warmed from 195-373 K over a period of four

hours. A second collection trap was exchanged for the first one when

the products reached 300 K. In the second test the plasma products

were warmed from 300 to 900 K over a period of 90 minutes. Only one

collection trap was used in this test. All of the collection traps,

and both of the product recovery traps, were sealed in air and heated

to 900 K in a bunsen burner. Color changes in the trap contents,

which indicated the presence of gold films, were observed visually.

Mass Spectrometer Experiments

Six experiments were performed using quadrupole or ICR mass

spectrometers to detect compounds in the plasma reaction products. All

of the instruments were located in the Department of Chemistry at the

University of Florida (U.F.). A synopsis of these experiments is

presented below in Table 5.

Table 5. Volatility Study Mass Spectrometer Experiments

Expt. Plasma Expt. Sample MS Type Sample Introduction
No. (Gas Mix) Type Method

1 860704 target residue quadrupole solids probe,
(75% He/25% CH4) heated to 670 K

2 870121 chamber wall GC/quad.MS injected acetone
(93% H2/8% C2H2) residue suspension/solution

3 870819 U-tube cold quadrupole connected to gas
(75% H2/25% CH4) trap inlet, heated to
680 K
4 860722 cold trap FTICR connected to gas
(75% He/25% CH4) after GWP inlet, heated to
380 K
5 870819 coated glass FTICR placed adjacent to
(75% H2/25% CH4) plate ICR cell; e-beam &
laser desorption
6 870819 residue from FTICR solids probe,
(75% H2/25% CH4) coated glass heated to 450 K,
plate CID reactions

The first experiment was done in the laboratory of Prof. R.A. Yost

by Mr. M. Hail using a Finnigan/MAT 4515 quadrupole MS. Reaction

products were recovered from the gold target used in test 860704 (75%

He/25% CH4) by scraping the tar-like residue with a teflon spatula.

The sample was placed in a solids probe and heated to 673 K while

observing the mass response from 0-494 atomic mass units (a.m.u.). The

instrument's mass range was calibrated using perfluorotributylene.

The second experiment was done in the department's Analytical

Service Laboratory by Dr. R.W. King using a Finnigan/MAT 4500 gas

chromatograph/quadrupole mass spectrometer (GC/MS). A sample was

recovered from the plasma chamber sidewall after test 870121 (93% H2/8%

C2H2) by rinsing the sidewall with spectral grade acetone. The
extract/suspension was injected onto a 5 ft. x 1/4 in. glass

chromatography column packed with an inert support coated with 3%

SP2100 (an organo-silane polymer commonly used for hydrocarbon

separations). Helium carrier gas flow rate through the column was 30

ml/min. The column temperature was increased from 318-573 K at 10

degrees per minute while observing the mass response from 0-300 a.m.u.

This instrument was also calibrated using a standard hydrocarbon

mixture (FC 43).

The third experiment was performed in the laboratory of Prof. M.T.

Vala by the author using a Finnigan 3500 quadrupole MS. The U-tube

cold trap used in test 870819 (75% H2/25% CH4) was connected to the

instrument's solids probe inlet and heated to 680 K in a tube furnace

while observing the mass response from 0-500 a.m.u. The instrument

response was tested by placing a sample of triethylphosphine-gold

chloride (synthesized by Dr. E. Sczlick in the laboratory of G.J.

Palenik at U.F.) in the solids probe and heating it to 570 K.

The fourth, fifth, and sixth experiments were performed in the

laboratory of Prof. J.R. Eyler by Dr. M. Moini and the author. The

fourth one used a sample from test 860722 (75% He/25% CH4). The sample

was collected in a 195 K cold trap positioned after a glass wool plug

during the plasma experiment. This cold trap was connected to the gas

inlet of an ICRMS (Nicolet FTMS-1000 equipped with a 3.0 Tesla

superconducting magnet and described fully in the literature).65 The

sample trap was then evacuated to 0.1 Pascals while still immersed in

dry ice. The valve to the MS was then opened and the mass response

from 0-1000 a.m.u. was observed using Fourier transform techniques

while the sample trap was slowly warmed to 380 K. The instrument

response was tested by placing a sample of triethylphosphine-gold

chloride in a solids probe and warming it to 450 K.

The fifth experiment used an inhouse fabricated ICRMS (equipped

with a Nicolet FTICR MS data station and a Varian Fieldial 1.0 Tesla

electromagnet) to examine plasma reaction products coated on a glass

plate in run 870819 (75% H2/25% CH4). The sample was mounted on a

solids probe and positioned in the vacuum chamber adjacent to the ICR

cell. The mass response from 0-400 a.m.u. was monitored using Fourier

transform methodology while the sample was exposed to a 500 nanoamp

electron beam for one hour. The sample was then irradiated with an

unfocused frequency doubled, 532 nm pulsed Nd:YAG laser (Quantel Model

YG 580) at power levels of 50-500 mJ/pulse in an attempt to desorb

compounds that would have then been ionized by co-desorbed potassium

ions. (The K+ ions came from the materials used to construct the ICR

cell. This was considered a "soft" ionization technique.) The mass

response was monitored for the next twenty minutes while the laser and

e-beam were on, and then for another ten minutes with the laser on and

the e-beam off.

The sixth experiment was performed in the Nicolet FTMS-1000 ICRMS

using a sample generated in run 870819 (75% H2/25% CH4). The sample

was placed in a capillary tube that was then inserted into the solids

probe and warmed to 450 K while the mass response from 0-1000 was

observed. Collision-induced ion dissociation (CID) experiments were

performed on compounds with mass numbers 44, 149, 195, 255, 334, and


Time of Flight MS Plasma Experiment

This experiment was performed in the laboratory of Prof. P.J.

Brucat in the Department of Chemistry at U.F. by Mr. D.E. Lessen and

the author. This laboratory was equipped with a cluster beam apparatus

linked to a time of flight mass spectrometer (see Figure 11) that is

described fully elsewhere.66 The cluster beam was produced by

irradiating a solid gold target with a focused pulsed Nd:YAG laser

(Quantel Model YG-580) beam at a power level of 1-2 mJ/pulse at target

impact (7 nsec pulses at 10 hz). Helium or helium/1% methane gas was

introduced to the target at a pressure of 1 atm. using a 1 msec

pulsed valve. The resultant plasma exited the target holder through a

2 mm dia. hole and was hypersonically expanded into a large vacuum

chamber maintained at a pressure of 0.01 Pascals. A skimmer in direct

Figure 11. Cluster Beam/TOFMS Apparatus

laser 4


- 0


micro channel
plate detector



Au source
target rod


line of sight with the plume exit allowed undeflected clusters to enter

a repeller grid where they were redirected down a three meter TOF tube

and into a multi-stage detector. A variable voltage horizontal

deflector (HD) located at the beginning of the TOF tube optimized the

transfer efficiency of various mass ranges of clusters.

During the course of the experiment the HD voltage was varied

between 50-120 volts. This allowed the observation of signals from

ionized gold clusters and cluster compounds from Au(1) to Au(7). At

each voltage setting the signals from positive ion clusters produced in

pure helium and in helium/1% methane plasmas were recorded.

Gold Thin Film Characterization
Six different techniques were utilized to characterize the plasma

reaction products formed in the 3:1 hydrogen/methane-gold plasmas.

Samples of the ultrafine particle product and coated substrates (both

annealed and unannealed) were subjected to analyses. Most samples were

in the form of thin films deposited on glass plates, ceramic

feedthroughs, and CVD tungsten or aluminum on silicon chips. A major

objective of these analyses was the determination of the relative

abundance, and the physical state, of the gold present in the soot-like

plasma reaction products. Film stoichiometry, structure, and

electrical properties were also investigated. The analytical

techniques utilized in this part of the study were

1) visible absorption spectroscopy,

2) optical microreflectometry (OMR),

3) x-ray photoelectron spectroscopy (XPS),

4) scanning electron microscopy (SEM) with

5) energy dispersive spectroscopy (EDS), and

6) electrical resistance (R)

The principle features of each technique are described briefly along

with the relevant parameters employed to analyze the sample films.

Detailed lists of the samples analyzed using each technique are

presented in Chapter III along with the respective analytical results

and a discussion of their meaning.

Visible Absorption Spectroscopy

A Hewlett-Packard Model 8450 diode array spectrometer was used by

the author to measure the absorption spectra (300-800 nm) of acetone

suspension/solutions of several samples recovered from the plasma

chamber sidewalls. Samples from CH4, He/CH4, and He/CO plasma

experiments were subjected to solvent dissolution tests using hexane,

toluene, acetone, diethyl-ether, 1:1 solutions of the preceding

solvents, water, hydrochloric acid, nitric acid, and aqua regia before

determining that pure acetone yielded the best results (discussed in

the next chapter). The absorption spectra of 10 ppm and 10,000 ppm Au

standards in acetone/water were also measured. These standards were

prepared by diluting 1 ml of the 1000 ppm and 10 ml of the 100,000 ppm

AA standards, respectively, to 100 ml with spectral grade acetone. All

of the sample and standard spectra were measured against an acetone

reference cell. All measurements were made in standard matched 1 cm

wide quartz cuvets.

The major objective of these analyses was the determination of the

presence or absence of the metallic absorption band for gold in the

550-600 nm region of the spectrum. This indicated the presence or

absence of gold particles large enough to exhibit metallic behavior.

Optical Microreflectometry

An optical microreflectometer located in the laboratory of Prof.

P. Holloway in the Material Science Department at U.F. was used by the

author to measure the reflectance spectra of several samples. The

instrument consisted of a quartz halogen light source and focusing lens

assembly, a 1200 I/mm grating monochromator, a binocular microscope

with internal mirrors for directing reflected light to a GaAs

photomultiplier tube, and a Keithley autoranging picoammeter (Model

485). Reflectance curves were determined by comparing the sample

intensities to the intensities of a standard with known reflectivity in

air, R(std). The reflectance intensity of a SiC standard, I(std), was

measured at 10 nm intervals between 400 and 900 nm just prior to

measuring the intensity of each sample's reflectance, I(samp), at the

same wavelengths. The reflectance value for each sample point,

R(samp), was then calculated using the following equation:

R[samp] = [I(samp) DC] x [R(std)]/[I(std) DC]

where DC was the photomultiplier dark current. R(std) values for SiC

are listed in Appendix C.

Sample films were examined visually to determine a representative

area for reflectance measurements. The instrument analyzed an area 20

microns in diameter. The same spot on the SiC standard was measured

each time by aligning two small defects on the crystal surface with

points on the microscope measurement grid. Reflectance curves were

determined for annealed and unannealed films deposited on glass plate

and on CVD tungsten on Si chips, as-deposited films on the two

polymeric tapes, and the Harris Corp. standard sputtered Au film.

X-Ray Photoelectron Spectroscopy

X-ray photoelectron spectroscopy studies were carried out on

several CH4/H2/Au plasma reaction product coated and annealed W on Si

substrates. The XPS analyses had three major objectives: 1)

determination of the elemental surface composition of the film, 2)

determination of the binding energy (BE) of the Au 4f5/2 and 4f7/2

peaks and the C Is peak, and 3) determination of the gold and carbon

concentrations in the bulk of the film. The first objective was

achieved through the use of elemental survey scans from 0-1000 eV BE.

The second objective was accomplished using narrow scans (-10 eV) of

the respective Au and C peaks and observing the shift in BE and the

peak shapes and widths at half height (whh). The determination of

carbon and gold concentrations in the bulk of the films was

accomplished using narrow scans with variable photoelectron takeoff

angles and Ar+ ion etching. Spectra were recorded in a fixed analyzer

transmission mode, which allowed direct comparison of adjusted peak

areas for semiquantitative analyses. A detailed list of all samples

subjected to XPS analyses, along with the respective analytical

conditions and results, is presented in Chapter III together with a

detailed discussion of the results.

Two instruments were used to perform the analyses in these

studies. One instrument, a Kratos XSAM 800, was located in the

laboratory of Prof. V.Y. Young in the Department of Chemistry at U.F.

and was operated by Mr. P.C. McCaslin and Mr. J.G. Davis, Jr. The

instrument used a Mg anode (1254.6 eV Ka line) x-ray source operated at

15 kV accelerating potential. Samples were mounted to a grounded

holder using stainless steel screws that contacted the surface of the

sample films and the underlying W substrate near the edges. This

arrangement prevented charging of the thin film while it was being

exposed to the x-radiation and undergoing electron loss. Spectra were

recorded at photoelectron takeoff angles of 10 and 70 degrees. This

technique allowed observation and comparison of the signals produced

predominantly by near surface species (-1.5 nm sampling depth at 10

degree takeoff angle) and those produced predominantly by bulk species

(-7.5 nm, or three times the mean free path of 1.2 keV electrons in

gold,67,68 sampling depth at 70 degree takeoff angle). One sample was

also etched with Ar* ions at 2.5 keV accelerating potential and 0.25 mA

sample current for a total of 60 minutes. Spectra were recorded before

the etch, after 15 minutes of etching and after 60 minutes of etching.

The second XPS instrument was a GCA/McPherson ESCA 36 located in

the laboratory of Prof. A.T. D'Agostino in the Department of Chemistry

at the University of South Florida and was operated by Mr. C. Biver.

This instrument used an Al anode (1487.6 eV Ka line) x-ray source

operated at 15 kV potential. An initial set of samples that was

analyzed in this instrument was improperly grounded and thus did not

yield valid BE data. However, since the charging effect reached an


equilibrium status within the first few seconds of exposure to the

constant intensity x-ray beam,69 the total areas of the Au and C peaks

and the elemental survey scans provided useful information. One sample

film and one Harris Corp. standard sputter deposited Au film were

subjected to depth profiling using Ar1 ion etching (13 AiA ion current

and 6.3 mA arc current). These two samples were grounded by pressing

copper foil over them, with an open area for analysis, and attaching

the foil securely to the grounded sample stand. All spectra were

recorded at a photoelectron takeoff angle of 75 degrees. Peak areas

were adjusted for elemental composition determinations using Scofield's

photoionization cross sections70 at 1254.6 and 1487.6 eV, respectively,

for the two instruments. Since only gold and carbon were detectable in

the films (the photoionization cross section for hydrogen is extremely

small, 0.0002 relative to carbon, and was not detectable with these

instruments), and the Scofield cross section value for the C Is peak is

taken as 1.00 relative (22,000 barns at 1254.6 eV, and 13,600 barns at

1487.6 eV), only the combined areas (XAu) of the Au 4f peaks were

adjusted. This was done using the following equation for the Mg Ka

source (1254.6 eV):

(XAu)/(7.68 + 9.79) = adjusted Au area

For the Al Ka source (1487.6 eV), the following equation was used:

(XAu)/(7.54 + 9.58) = adjusted Au area

Once the adjusted Au areas (adjAu) were calculated they were compared

directly to the corresponding C Is peak areas to yield relative percent

atomic composition using the following equations:

(100)(adjAU)/(adjAu + C Is) = relative atomic % Au

(100)(C is)/(adjAu + C Is) = relative atomic % C

Relative percent mass concentrations were then calculated using the

following equations, where ra%Au and ra%C are the relative atomic

percent concentrations of gold and carbon, respectively:

(ra%Au)(197)/[(ra%Au)(197) + (ra%C)(12.0)] = relative mass % Au

(ra%C)(12.0)/[(ra%Au)(197) + (ra%C)(12.0)] = relative mass % C

In some discussions of the results in the following chapter references

are made to the ratio of gold to carbon and vice versa. In all

instances these discussions are referring to the atomic ratios of the

two elements.

Scanning Electron Microscopy with Energy Dispersive Spectroscopy

The SEM/EDS analyses were carried out in order to determine 1) the

average grain or particle size and shape of the species comprising the

as deposited sample films and 2) the change in film structure

associated with various annealing regimes. These analyses were

performed in a JEOL SEM equipped with an EDS system located in the

laboratory of Prof. M. Ammons in the College of Engineering Center for

Electron Microscopy at the University of South Florida. The instrument

was operated by Prof. Ammons and Ms. A.S. Slater.

The optics in the EDS system precuded the observation of low Z

elements, thus hydrogen and carbon were not observable. The EDS was

used primarily to determine qualitatively the relative abundance of

gold in the surface being scanned by the SEM. By varying the electron

energy, and thus the mean free path in the sample film, a rough depth

profile could be observed. This facilitated the interpretation of the

signals produced by the underlying substrate. Plasma reaction products

coated on glass chips, W on sc and pc SI chips and Scotch tape were

examined along with the Harris Corp. standard sputtered Au film. Both

annealed and as-deposited samples were examined. A detailed list of

the samples analyzed is presented in the next chapter.

Electrical Resistance Measurements

Two methods of observing the electrical resistance properties of

the sample films were employed. The first technique consisted of

observing the resistance across a 3 mm wide ceramic junction in an

electrical feedthrough that had been heavily coated with plasma

reaction products. The feedthrough was inserted into the substrate

anneaiing chamber described earlier along with a thermocouple junction.

The resistance was monitored as the temperature was slowly increased

from 300-750 K. In one experiment hydrogen gas was flowed through the

system at 200 ml/min. In a second experiment the system was left open

to the ambient atmosphere. The purpose of these two tests was to

observe any minima in the resistance values versus the temperature as


the reaction product film connecting the electrodes was heated under

hydrogen and under air.

The second resistance measurement method used a 4-point probe

technique, described in detail by L.B. Valdes,71 to determine the sheet

resistance, Rs, in ohms/square, of sample films coated on nonconductors

(glass, quartz, Scotch tapes). An Alessi CPA/1 Resistivity Test

Fixture with a 4-point osmium head was available for use in the

laboratory of Prof. P. Holloway. Since the sample films were unevenly

coated, and thickness determinations of the delicate films was not

obtainable with available instrumentation, the main objective of these

measurements was the determination of relative sheet resistance values

for various areas on the samples.

Measurements were made by adjusting the input current to an

appropriate level, measured in amps, and reading the voltage drop

between the two outer probes and the two inner probes. The sheet

resistance was then calculated using the relationship:

Rs = (V/I)(4.2)

The correction factor, 4.2, is based on a probe spacing equal to -10%

of the average specimen diameter. Since the probe spacing was 1 mm and

the average sample film dimensions were not discernible due to the

microscopic discontinuities inherent in the films, the sheet resistance

values can be viewed only in relation to each other and to the standard

sputtered Au film sheet resistance.