www.elsevier.com/locate/tsf

Thin Solid Films 47

Role of atomic hydrogen density and energy in low power

chemical vapor deposition synthesis of diamond films

Randell Mills*, Jayasree Sankar, Andreas Voigt, Jiliang He, Paresh Ray, Bala Dhandapani

BlackLight Power, Inc., 493 Old Trenton Road, Cranbury, NJ 08512, USA

Received 8 December 2003; received in revised form 7 October 2004; accepted 7 October 2004

Available online 23 November 2004

Abstract

Polycrystalline diamond films were synthesized on silicon substrates without diamond seeding by a very low power (~40–80 W)

microwave plasma continuous vapor deposition reaction of a mixture of helium–hydrogen–methane (48.2/48.2/3.6%) or argon–hydrogen–

methane (17.5/80/2.5%). However, predominantly graphitic carbon films or no films formed when neon, krypton, or xenon was substituted

for helium or argon. The films were characterized by time of flight secondary ion mass spectroscopy, X-ray photoelectron spectroscopy,

Raman spectroscopy, scanning electron microscopy, and X-ray diffraction. It is proposed that each of He+ and Ar+ served as a catalyst with

atomic hydrogen to form an energetic plasma since only plasmas having these ions in the presence of atomic hydrogen showed significantly

broadened H a lines corresponding to an average hydrogen atom temperature of N100 eV as reported previously. It was found that not only

the energy, but also the H density uniquely increases in He–H2 and Ar–H2 plasmas. Bombardment of the carbon surface by highly energetic

hydrogen formed by the catalysis reaction may play a role in the formation of diamond. Then, by this novel pathway, the relevance of the CO

tie line is eliminated along with other stringent conditions and complicated and inefficient techniques which limit broad application of the

versatility and superiority of diamond thin film technology.

D 2004 Elsevier B.V. All rights reserved.

Keywords: Polycrystalline diamond films; Chemical vapor deposition; Helium or argon–hydrogen–methane microwave plasma; Spectroscopic

characterization; Energetic hydrogen mechanism

1. Introduction

Diamond has some of the most extreme physical

properties of any material such as outstanding mechanical

strength, optical transparency, high thermal conductivity,

high electron mobility, and unique chemical properties [1].

Thus, a variety of possible applications are envisioned for

diamond materials. Yet, its practical use in applications has

been limited due to its scarcity, expense, and immalle-

ability. The development of techniques for depositing thin

films of synthetic diamonds on a variety of substrates has

enabled the exploitation of diamond’s superlative proper-

ties in many new and exciting applications. These include

0040-6090/$ - see front matter D 2004 Elsevier B.V. All rights reserved.

doi:10.1016/j.tsf.2004.10.009

* Corresponding author. Tel.: +1 609 490 1090; fax: +1 609 490 1066.

E-mail address: rmills@blacklightpower.com (R. Mills).

cutting tools, thermal management of integrated circuits,

optical windows, high temperature electronics, surface

acoustic wave filters, field emission displays, electro-

chemical sensors, composite reinforcement, microchemical

devices and sensors, and particle detectors [1]. However,

the fundamental impediment facing the technology at the

present is insufficient growth rate of high-quality diamond

under milder conditions (low temperature and power

density).

Synthetic diamond was initially commercially produced

as single crystals using the so-called high-pressure high-

temperature (HPHT) growth technique [1] wherein graph-

ite is compressed in a hydraulic press to tens of thousands

of atmospheres, heated to over 2000 K, and left until

diamond crystallizes. Recent novel HPHT methods which

have been largely unsuccessful, except for the production

of nanocrystals by Orwa et. al. [2], are based on attempts

8 (2005) 77–90

R. Mills et al. / Thin Solid Films 478 (2005) 77–9078

to use high energy ion implantation to bury carbon deep in

metals or fused silica to take advantage of the large

confining pressures there. More versatile thin films have

been produced by an addition-of-one-atom-at-a-time

approach using chemical vapor deposition (CVD) techni-

ques. All CVD techniques for producing diamond films

require activation of the gaseous carbon-containing pre-

cursor molecules. To promote diamond over graphite

growth, the precursor gas is usually CH4 that is diluted

in excess hydrogen that is typically 99% of the reactant

mixture, and the temperature of the substrate is usually

maintained in excess of 700 8C. Activation may be

achieved thermally using a hot filament, gas discharge

such as DC, radio frequency (RF), or microwave dis-

charges, or a combustion flame such as an oxyacetylene

torch [1].

Although the mechanism of diamond growth on a seed of

diamond is still not well understood, it is believed to be

based on the extraction of H of a CH terminal bond to form

a dangling carbon center to which CH3 reacts. A carbon–

carbon bond forms between adjacent methyl groups, and the

hydrogen is gradually extracted, probably by H forming H2.

The further preferential etching of graphitic carbon over

diamond carbon by hydrogen permits diamond growth [1].

H may also be required to decrease the concentration of gas

phase unsaturated hydrocarbons.

More recent advances of diamond formation have been

towards developing methods to grow diamond at low

temperatures (b500 8C rather than 700 8C) such that

diamond films can be grown on a wider range of substrate

materials of commercial importance with low melting

points such as plastics, aluminum, some glasses, nickel,

steel, and electronic materials such as GaAs. Many gas

mixtures have been investigated to achieve this goal

including ones containing some halogens, presumably

substituting for the role played by H [3]. More common

mixtures have different combinations of H2, CH4, O2,

CO2, and CO [3]. Quite successful diamond film growth

has been achieved at temperatures as low as 180 8C using

gas mixtures of CH4 mixed with CO2 or CO in microwave

plasma deposition reactors wherein an optimal rate is

obtained when the gas ratio is about 50/50%. Although the

concentration of H2 in the activated gas mixture is

approximately half that seen in the CH4–H2 mixtures [4],

the CO2–CH4 and CO–CH4 systems are unique in that

hydrogen is low compared to the excess needed in other

systems presumably because oxygen species such as O2,

O, and OH in the CO2–CH4 and CO–CH4 system plasmas

perform the same role as H in the CH4–H2 system

plasmas. Recent molecular beam mass spectroscopy

investigations of the CO2–CH4 system indicate the

incorporation of CH3 at a dangling carbon bond is the

most probable mechanism as in the case of the CH4–H2

system. However, the species that extracts H may not be

an oxygen species. Rather, CO may activate the surface by

extracting a terminating H [5].

Empirically, it is known that only a narrow set of ratios of

O, C, and H result in diamond formation. Using the

combined data from over 70 diamond deposition experi-

ments, Bachmann et al. [6,7] produced a C–H–O phase

diagram for diamond deposition which showed that low

pressure diamond synthesis is only possible within a very

narrow well-defined domain centered on a line called the

CO tie line. A consequence of this analysis was that the

exact nature of the plasma gases was unimportant for most

CVD processes; rather, the relative ratios of O, C, and H

controlled the deposition.

Recently, bombardment of the carbon surface with

energetic hydrogen atoms has been shown to play a key

role in the formation of diamond [8]. The model for

diamond nucleation by energetic species involves (1)

spontaneous bulk nucleation of a diamond embryo cluster

in a dense, amorphous carbon hydrogenated matrix, (2)

stabilization of the cluster by favorable boundary con-

ditions of nucleation sites and hydrogen termination, and

(3) ion bombardment-induced growth through a preferen-

tial displacement of sp2 bonded C atoms over sp3 bonded

atoms since the displacement energy of latter is consid-

erably lower. The bombardment with H of a few 100 eV

leaves the diamond atoms intact but displaces the graphitic

atoms, which can then occupy diamond positions. The

acceleration of charged particles in a high field was used to

produce fast H. Mills et al. [9,10] have shown that it is

possible to form fast H in certain plasmas by a nonfield-

acceleration mechanism. From the width of the 656.3 nm

Balmer a line emitted from microwave and glow discharge

plasmas, it was found that a strontium–hydrogen micro-

wave plasma showed a broadening similar to that observed

in the glow discharge cell of 27–33 eV, whereas, in both

sources, no broadening was observed for magnesium–

hydrogen. Microwave helium–hydrogen and argon–hydro-

gen plasmas showed extraordinary broadening correspond-

ing to an average hydrogen atom temperature of 180–210

and 110–130 eV, respectively. The corresponding results

from the glow discharge plasmas were 33–38 and 30–35

eV, respectively, compared to c3 eV for plasmas of pure

hydrogen, neon–hydrogen, krypton–hydrogen, and xenon–

hydrogen maintained in either source. External Stark

broadening or acceleration of charged species due to high

fields cannot explain the microwave results since no high

field was present, and the electron density was orders of

magnitude too low for the corresponding Stark effect.

Rather, a proposed resonant energy transfer mechanism

explains these results [9,10]. In the case of helium or

argon-mixed plasmas, He+ or Ar+, respectively, serves as a

catalyst to resonantly accept energy equivalent to the

potential energy of atomic hydrogen, Eh=27.2 eV where

Eh is one hartree. The theory has been given previously

[11–13].

In the quest for low temperature diamond synthesis,

methane as a carbon source was mixed with helium– or

argon–hydrogen plasmas having documented fast H as

R. Mills et al. / Thin Solid Films 478 (2005) 77–90 79

well with matching control plasmas wherein the noble gas

was neon, krypton, or xenon. In this paper, we report the

deposition of polycrystalline diamond films on silicon

wafers only by helium–hydrogen–methane (48.2/48.2/

3.6%) or argon–hydrogen–methane (17.5/80/2.5%) micro-

wave plasmas maintained with an Evenson cavity, whereas

predominately hydrocarbon or graphitic carbon films

formed upon replacement of these noble gases with neon,

krypton, or xenon. The films formed under very mild

conditions without diamond seeding or abrasion that

provides seed crystals [14]. After each plasma processing

reaction, the surface was characterized by time of flight

secondary ion mass spectroscopy (ToF-SIMS), X-ray

photoelectron spectroscopy (XPS), Raman spectroscopy,

scanning electron microscopy (SEM), and X-ray diffrac-

tion (XRD). In order to understand and confirm the role of

fast H, the helium–hydrogen–methane plasma was char-

acterized by recording the line broadening of the 656.3 nm

Balmer a line to determine the excited hydrogen atom

energy. The effect on the H densities of the addition of

helium or argon versus xenon to hydrogen plasmas was

also studied.

2. Experimental

2.1. Synthesis

Diamond films were grown on silicon wafer substrates

by their exposure to a low pressure noble gas (NG)–H2–

CH4 microwave plasma. The experimental setup compris-

ing a microwave discharge cell operated under flow

Fig. 1. The experimental setup comprising a microwav

conditions is shown in Fig. 1. A silicon wafer substrate

(0.5�0.5�0.05 cm, Silicon Quest International, silicon

(100), boron-doped) cleaned by heating to 700 8C under

vacuum was placed axially about 2 cm off center inside of

a quartz tube (1.2 cm in diameter�25 cm long) with

vacuum valves at both ends. The tube was center-fitted

with an Opthos coaxial microwave cavity (Evenson cavity)

and connected to the gas/vacuum line. The quartz tube and

vacuum line were evacuated for 2 h at a base pressure of 2

mTorr to remove any trace moisture or oxygen and

residual gases. Premixed NG–H2 (50/50%) where NG

was a noble gas: He, Ne, Kr, and Xe was further mixed

with CH4 such that a NG–H2–CH4 (48.2/48.2/3.6%) gas

mixture was introduced through the quartz tube reactor at a

total pressure of 3 Torr as monitored by an absolute

pressure gauge. The corresponding gas flow rates con-

trolled by mass flow controllers were maintained at 60 and

2.25 sccm for NG–H2 and CH4, respectively. In separate

experiments, the NG–hydrogen premixed gas was varied

from (90/10%) to (50/50%). Since the best diamond film

results were obtained with the (50/50%) mixture, only

these results will be presented.

In the case that NG was Ar, precursor reactant gases

CH4, H2, and Ar were introduced through the quartz tube

reactor at a total pressure of 2.5 Torr as monitored by an

absolute pressure gauge. The corresponding gas flow

rates controlled by mass flow controllers were maintained

at 2, 65, and 15 sccm for CH4, H2, and Ar, respectively.

The microwave generator shown in Fig. 1 was an

Opthos model MPG-4M generator (frequency: 2450 MHz).

The microwave plasma was maintained with a 40 W

(forward)/2 W (reflected) power for about 12–16 h. The

e discharge cell operated under flow conditions.

R. Mills et al. / Thin Solid Films 478 (2005) 77–9080

substrate was at the cool edge of the plasma glow region.

The wall temperature at this position measured with a

contacting thermocouple was about 300 8C. No films were

formed when the noble gas was neon; thus, no analyses

were performed for the corresponding samples. A thick

(~10–100 Am) crystalline, shiny coating formed on the

substrate and the wall of the quartz reactor in the case of

He– and Ar–H2–CH4 plasmas. A thick (~100 Am) trans-

lucent, golden-yellow, shiny coating formed on the

substrate and the wall of the quartz reactor in the case

of Kr– and Xe–H2–CH4 plasmas.

The quartz tube was removed and transferred to a

drybox with the samples inside by closing the vacuum

valves at both ends and detaching the tube from the

vacuum/gas line. The coated silicon wafer substrate was

mounted on XPS and ToF-SIMS sample holders under an

argon atmosphere in order to prepare samples for the

corresponding analyses. Controls for XPS analysis com-

prised a cleaned commercial silicon wafer (Silicon Quest

International, silicon (100), boron-doped) and known

standards: (a) single crystal diamond, (b) diamond film,

(c) glassy carbon, (d) pyrolytic graphite, (e) mineral

graphite, and (f) hydrogenated diamond-like carbon

(HDLC). The control for ToF-SIMS analysis comprised

a cleaned commercial silicon wafer (Silicon Quest Interna-

tional, silicon (100), boron-doped). The coated substrate

was also sent for Raman analysis (Charles Evans &

Associates, Sunnyvale, CA), SEM characterization

(S.S.W., University of Western Ontario, Canada and

Material Testing Laboratory, Pennington, NJ), and XRD

analysis (IC Laboratories, Amawalk, NY).

2.2. ToF-SIMS characterization

A cleaned commercial silicon wafer before and after

plasma treatment to form a film coating were characterized

using a Physical Electronics TRIFT ToF-SIMS instrument.

A diamond powder was also used as a control. The primary

ion source was a pulsed 69Ga+ liquid metal source operated

at 15 keV. The secondary ions were extracted by a F3 keV

(according to the mode) voltage. Three electrostatic

analyzers (Triple-Focusing-Time-of-Flight) deflect them in

order to compensate for the initial energy dispersion of ions

of the same mass. The 400 pA dc current was pulsed at a 5

kHz repetition rate with a 7 ns pulse width. The analyzed

area was 60�60 Am and the mass range was 0–1000 AMU.

The total ion dose was 7�1011 ions/cm2, ensuring static

conditions. Charge compensation was performed with a

pulsed electron gun operated at 20 eV electron energy. In

order to remove surface contaminants and expose a fresh

surface for analysis, the samples were sputter-cleaned for 30

s using a 80�80 Am raster, with 600 pA current, resulting in

a total ion dose of 1015 ions/cm2. Three different regions on

each sample of 60�60 Am were selected using a microscope

and analyzed. The positive and negative SIMS spectra were

acquired. Representative post sputtering data is reported.

The ToF-SIMS data were treated using dCadenceT software(Physical Electronics), which calculates the mass calibration

from well-defined reference peaks.

It has been shown by Sanders et al. [15] that the signal

intensity of Cm� fragments decreases exponentially with

cluster size in the negative ToF-SIMS spectrum. The

absolute value of the slope of the regression line through

the plot of the logarithm of the signal intensity versus cluster

size, called the bfragmentation factorQ, F, is empirically

related to the coordination of the carbon at the surface of the

source of the ions. The smaller the value of F, the higher the

coordination (i.e. sp3 versus sp2 content). He– and Ar–H2–

CH4 plasma-formed carbon films gave high-order clusters,

and F for each of the film samples were compared to that of

standards after Sanders et al. [15].

2.3. XPS characterization

A series of XPS analyses were made on the samples

using a Scienta 300 XPS Spectrometer at Lehigh University,

Bethlehem, PA. The fixed analyzer transmission mode and

the sweep acquisition mode were used. The aluminum X-

ray incidence angle was 158. The step energy in the survey

scan was 0.5 eV, and the step energy in the high resolution

scan was 0.15 eV. In the survey scan, the time per step was

0.4 s, and the number of sweeps was 4. In the high

resolution scan, the time per step was 0.3 s, and the number

of sweeps was 30. C 1s at 284.6 eV was used as the internal

standard.

2.4. Raman spectroscopy

Experimental and control samples were analyzed by

Charles Evans & Associates. Raman spectra were obtained

with a LABRAM spectrometer (Dilor of Jobin Yvon) with a

Spectrum One charge coupled device (CCD) detector (Spex

and Jobin Yvon) that was air and Peltier cooled. An

Omnichrome HeNe laser (Melles Griot) with the light

wavelength of 632.817 nm was used as the excitation

source. The spectra were taken at ambient conditions and

the samples were placed under a Raman microscope

(Olympus BX40). Spectra of the film samples were acquired

using the following condition: the laser power at the sample

was 4–8 mW, the slit width of the monochromator was 100

Am which corresponds to a resolution of 3 cm�1, the

detector exposure time was 20 min, and three scans were

averaged.

Experimental and control samples were also analyzed by

Jobin Yvon, Edison, NJ. Raman spectra were obtained with

a LabRam HR system (Jobin Yvon) with a built-in micro-

scope. The spectrometer had two interchangeable gratings

with 1800 gr/mm, blazed at 500 nm, and was equipped with

a liquid nitrogen-cooled CCD detector (1024�256, BI, UV

coated). A helium–cadmium laser with the light wavelength

of 325 nm was used as the excitation source. The spectra

were taken at ambient conditions, and the samples were

R. Mills et al. / Thin Solid Films 478 (2005) 77–90 81

mounted under the Raman microscope. Spectra of the film

samples were acquired at 3-s integration using the following

condition: the laser power at the sample was 15 mW and the

slit width of the monochromator was 500 Am, which

corresponds to a resolution of 2.5 cm�1.

2.5. SEM microstructure analysis

The surface microstructure of the films was determined

using a Hitachi-S-4500 Field Emission SEM at S.S.W.,

University of Western Ontario, Canada and an Amray 1000

A SEM at Material Testing Laboratory.

2.6. Characterization by XRD

The XRD patterns were obtained by IC Laboratories

using a Phillips 547 Diffractometer tuned for CuKa

(1.540590 2) radiation generated at 45 kV and 35 mA.

The sample was scanned from 10 to 100 two-theta (2h) witha step size of 0.028 and 1 s per step.

2.7. Visible spectroscopy and Balmer line broadening

measurements

The width of the 656.3 nm Balmer a line, 486.1 nm

Balmer h line, 434.0 nm Balmer g line, and 410.2 nm

Balmer y line emitted from hydrogen alone, xenon–

hydrogen mixture (90/10%), helium–hydrogen mixture

(90/10%), hydrogen–methane mixture (10–50/90–50%),

and helium–hydrogen–methane mixture (48.2/48.2/3.6%)

microwave discharge plasmas maintained in the micro-

wave discharge cell shown in Fig. 1 was measured with a

high resolution visible spectrometer capable of a reso-

lution of F0.006 nm according to methods given

previously [9,16]. The spectra were recorded over the

regions (410.05–410.35 nm, 433.85–434.35, 485.90–

486.40, and 656.0–657.0 nm). The total flow rate was

controlled at 62.25 sccm using one mass flow controller

for hydrogen alone and two in the case of a mixture of

two gases. In the case of the helium–hydrogen–methane

mixture, the methane flow at one mass flow controller

was 2.25 sccm and the helium–hydrogen flow of 60 sccm

was controlled by a second mass flow controller from a

50/50% mixture. The total pressure was 3 Torr, and the

input power to the plasma was set at 40 W. The 667.816

nm He I line width was also recorded with the high

resolution (F0.006 nm) visible spectrometer for helium–

hydrogen (90/10%) and helium microwave discharge

plasmas. The visible spectrum was also recorded on the

helium–hydrogen–methane mixture (48.2/48.2/3.6%)

microwave discharge plasma.

To measure the absolute intensity, the high resolution

visible spectrometer and detection system were calibrated

[17] with 546.08, 576.96, and 696.54 nm light from a Hg–

Ar lamp (Ocean Optics, model HG-1) that was calibrated

with a National Institute of Standards and Technology

(NIST) certified silicon photodiode. The population density

of the n=3 hydrogen excited state N3 was determined from

the absolute intensity of the Balmer a (656.28 nm) line

measured using the calibrated spectrometer. The spectrom-

eter response was determined to be approximately flat in the

400–700 nm region by ion etching and with a tungsten

intensity calibrated lamp [9].

To further characterize the energetic mixed plasmas, the

effect of the addition of helium and argon versus xenon to

hydrogen plasmas was studied. The extreme ultraviolet

(EUV) spectrum of the region of the Lyman a line was

recorded before and after the addition of 5% of each of

these noble gases to a hydrogen plasma according to the

methods described in Ref. [11]. Hydrogen or the hydro-

gen–noble gas mixture (95/5%) was flowed through the

1.2 cm diameter quartz tube of the microwave cavity cell

at 1 Torr and 30 sccm. The input power to the plasma was

set at 85 W. For spectral measurement, the light emission

from the microwave plasmas were introduced to a normal

incidence McPherson 0.2 meter monochromator (Model

302) equipped with a 1200 lines/mm holographic grating

with a platinum coating. The wavelength region covered

by the monochromator was 5–560 nm. The UV spectrum

(90–180 nm) of the cell emission was recorded with a

channel electron multiplier (CEM). The wavelength reso-

lution was about 1 nm full width half maximum (FWHM)

with an entrance and exit slit width of 300 Am. The

increment was 0.2 nm and the dwell time was 500 ms. The

spectrometer was calibrated between 40–200 nm with a

standard discharge light source using He, Ne, Ar, Kr, and

Xe lines: He I (58.4 nm), He II (30.4 nm), Ne I (73.5 nm),

Ne II (46.07 nm), Ar I (104.8 nm), Ar II (93.2 nm), Kr II

(96.4 nm), Xe I (129.56 nm), Xe II (104.13 nm), Xe II

(110.043 nm). The wavelength and intensity ratios

matched those given by Kelly [18]. The spectrometer

response was determined to be approximately flat in the

100–130 nm region.

The electron density was determined using a Langmuir

probe according to the method given previously [19].

3. Results

3.1. ToF-SIMS

The positive ion ToF-SIMS spectra of a cleaned

commercial silicon wafer before and after being coated

with a carbon film from He– and Ar–H2–CH4 plasmas are

shown in Fig. 2a–c, respectively. The positive ion spectrum

of the silicon substrate was dominated by Si+, oxides SixOy+,

and hydroxides Six(OH)y+, whereas that of the carbon film

samples contained no silicon-containing fragments. Rather,

a large H+ peak and hydrocarbon fragments CxHy+ were

observed from each carbon film.

The negative ion ToF-SIMS spectra of a cleaned

commercial silicon wafer before and after being coated

Fig. 3. Negative ToF-SIMS characterization of the substrate and carbon

films. (a) The negative ion ToF-SIMS spectrum (m/e=0–100) of a

noncoated cleaned commercial silicon wafer. (b) The negative ion ToF-

SIMS spectrum (m/e=0–100) of a cleaned commercial silicon wafer coated

by reaction of a He–H2 plasma with CH4 as the source of C. (c) The

negative ion ToF-SIMS spectrum (m/e=0–100) of a cleaned commercial

silicon wafer coated by reaction of an Ar–H2 plasma with CH4 as the source

of C.

Fig. 4. A plot of the linear regression of the signal intensity of the negative

ToF-SIMS of the He–H2–CH4 plasma carbon film versus cluster Cm�.

Fig. 2. Positive ToF-SIMS characterization of the substrate and carbon

films. (a) The positive ion ToF-SIMS spectrum (m/e=0–100) of a

noncoated cleaned commercial silicon wafer. (b) The positive ion ToF-

SIMS spectrum (m/e=0–100) of a cleaned commercial silicon wafer coated

by reaction of a He–H2 plasma with CH4 as the source of C. (c) The

positive ion ToF-SIMS spectrum (m/e=0–100) of a cleaned commercial

silicon wafer coated by reaction of an Ar–H2 plasma with CH4 as the source

of C.

R. Mills et al. / Thin Solid Films 478 (2005) 77–9082

with a carbon film from He– and Ar–H2–CH4 plasmas are

shown in Fig. 3a–c, respectively. The silicon substrate

spectrum was dominated by oxide (O�) and hydroxide

(OH�), whereas spectra of the He– and Ar–H2–CH4

plasma-formed carbon films were dominated by hydride

ion (H�) and carbon ion (C�). Very little oxide (O�) or

hydroxide (OH�) was observed, whereas large O� and

hydroxide OH� peaks were observed in the case of the

Kr– and Xe–H2–CH4 plasma-formed carbon films. Also,

higher-order multimeric carbon clusters Cx� at m/e=12, 24,

36, 48, 60, 72, 84, 96, 108, 120 were observed from He–

and Ar–H2–CH4 plasma-formed films that were also seen

in the diamond powder control.

A plot of the linear regression of the signal intensity of

the negative ToF-SIMS of the He–H2–CH4 plasma carbon

film versus cluster Cm� is shown in Fig. 4. The absolute

value of the slope B is the F factor of Sanders [15]. The

empirical values of F that are similar in trend to those

measured by Sanders et al. [15] for various forms of carbon

are given in Table 1. Based on a match with the empirical

trend of F with carbon coordination mode (i.e. sp3 versus

sp2 content), the coordination trend also shown in Table 1 is

He–H2–CH4 plasma carbon film NAr–H2–CH4 plasma

carbon film.

Table 1

Fragmentation factor, F, for glassy carbon, graphite, DLC, diamond, and

films from the plasma reactions

Sample Fragmentation factor,

F (�102)

Glassy carbon 50.2

Graphite 48.2

DLC 47.3

Natural diamond 37.1

He–H2–CH4 plasma carbon film 42.3

Ar–H2–CH4 plasma carbon film 45.9

R. Mills et al. / Thin Solid Films 478 (2005) 77–90 83

3.2. XPS

A survey spectrum was obtained over the region Eb=0 to

1200 eV. The primary element peaks allowed for the

determination of all of the elements present. The XPS

survey scan of a cleaned commercial silicon wafer before

and after being coated with a carbon film from Xe, He–, and

Ar–H2–CH4 plasmas are shown in Fig. 5a–d, respectively.

The major peaks identified in the XPS spectrum of the

silicon substrate sample were O 1s at 533.0 eV, trace C 1s at

284.6 eV, dominant Si 2s at 152.4 eV, and Si 2p3/2 at 101.9

Fig. 5. XPS survey scan characterization of the substrate and carbon films.

(a) The XPS survey scan of a cleaned commercial silicon wafer. (b) The

XPS survey scan of a cleaned commercial silicon wafer coated by reaction

of a Xe–H2 plasma with CH4 as the source of C. (c) The XPS survey scan

of a cleaned commercial silicon wafer coated by reaction of a He–H2

plasma with CH4 as the source of C. (d) The XPS survey scan of a cleaned

commercial silicon wafer coated by reaction of an Ar–H2 plasma with CH4

as the source of C.

eV. Only carbon and trace silicon and oxygen contamination

were observed from the He– and Ar–H2–CH4 plasma

carbon films as indicated by the trace O 1s peak at 532.9

eV, the trace Si 2s at 153.2 eV and Si 2p3/2 at 102.2 eV, and

the dominant C 1s peak at 284.6 eV, whereas a dominant

oxide as well as a carbon peak were observed for the Xe–

H2–CH4 plasma-formed film.

The high resolution XPS spectra (0–35 eV) of the

valence band region of (a) single crystal diamond, (b)

diamond film, (c) glassy carbon, (d) pyrolytic graphite, (e)

mineral graphite, and (f) HDLC are shown in Fig. 6 [20].

The corresponding XPS spectrum of the carbon film

samples from Xe–, He–, and Ar–H2–CH4 plasmas are

shown in Fig. 7a–c, respectively. The films from the Kr–

and Xe–H2–CH4 plasmas matched graphitic carbon or

HDLC as exemplified by Fig. 7a. An O 2s peak was also

observed at 23 eV, and valence band features of Si–C–O

were observed. The films from the He– and Ar–H2–CH4

plasmas each had a sharp peak at 13.2 eV and a broad

peak at 17.4 eV which matched the peak energies of single

crystal diamond rather than that of the other forms of

carbon which were observed at higher binding energies.

No O 2s peak was also observed in the region of 23 eV as

shown in Fig. 7b and c.

The high resolution XPS spectra (280–340 eV) of the

C 1s energy loss region of (a) single crystal diamond, (b)

diamond film, (c) glassy carbon, (d) pyrolytic graphite, (e)

mineral graphite, and (f) HDLC are shown in Fig. 8 [20].

The corresponding XPS spectrum of the carbon film

samples from Xe–, He–, and Ar–H2–CH4 plasmas are

Fig. 6. High resolution XPS spectra (0–35 eV) of the valence band region of

(a) single crystal diamond, (b) diamond film, (c) glassy carbon, (d)

pyrolytic graphite, (e) mineral graphite, and (f) HDLC.

Fig. 8. High resolution XPS spectra (280–340 eV) of the C 1s energy loss

region of (a) single crystal diamond, (b) diamond film, (c) glassy carbon,

(d) pyrolytic graphite, (e) mineral graphite, and (f) HDLC.

Fig. 7. High resolution XPS spectra (0–35 eV) of the valence band region of

cleaned commercial silicon wafers coated by reaction of noble gas–

hydrogen plasmas with CH4 as the source of C. (a) Valence band region of a

film formed by a Xe–H2–CH4 plasma. (b) Valence band region of a film

formed by a He–H2–CH4 plasma. (c) Valence band region of a film formed

by an Ar–H2–CH4 plasma.

R. Mills et al. / Thin Solid Films 478 (2005) 77–9084

shown in Fig. 9a–c, respectively. Single crystal diamond,

diamond film, and HDLC have an energy loss feature

which begins at about 290 eV which is at a higher energy

than that of the other possible forms of carbon as shown

in Fig. 8. The films from the Kr– and Xe–H2–CH4

plasmas showed features that matched graphitic carbon or

HDLC as exemplified by Fig. 9a. The closest match to the

shape of the energy loss feature of the carbon film from

He– and Ar–H2–CH4 plasmas is single crystal diamond to

which each film was assigned as a polycrystalline form

based on the Raman and SEM characterizations given

below.

3.3. Raman

The Raman spectra of corresponding regions of each

diamond film from Xe–, He–, and Ar–H2–CH4 plasmas

that were analyzed by ToF-SIMS and XPS are shown in

Fig. 10a–c, respectively. Extensive curve fit analysis was

performed. The peak positions, FWHM, and peak areas

were calculated by Gaussian curve fitting the baseline

corrected spectrum. The Raman spectra of the films from

the Xe–H2–CH4 plasmas showed mostly hydrocarbons

indicated by their bending vibration at 1446 cm�1 as

Fig. 9. High resolution XPS spectra (280–340 eV) of the C 1s energy loss

region of cleaned commercial silicon wafers coated by reaction of noble

gas–hydrogen plasmas with CH4 as the source of C. (a) C 1s energy loss

region of a film formed by a Xe–H2–CH4 plasma. (b) C 1s energy loss

region of a film formed by a He–H2–CH4 plasma. (c) C 1s energy loss

region of a film formed by an Ar–H2–CH4 plasma.

Fig. 11. The Raman spectrum of a second diamond film formed by

He–H2–CH4 microwave discharge plasma CVD.

Fig. 10. The Raman spectra of films from Xe–, He–, and Ar–H2–CH4

plasmas. (a) The Raman spectrum of the film from a Xe–H2–CH4 plasma.

(b) The Raman spectrum of the film from a He–H2–CH4 plasma. (c) The

Raman spectrum of the film from an Ar–H2–CH4 plasma.

R. Mills et al. / Thin Solid Films 478 (2005) 77–90 85

shown in Fig. 10a. Rarely, some spots of carbon bands,

amorphous sp3 at 1179 cm�1, D-band of diamond-like

carbon (DLC) at 1336 cm�1, and G-band of graphite at

1611 cm�1, were observed. No film formed when neon

replaced xenon, and results similar to Xe–H2–CH4 were

obtained with Kr–H2–CH4.

For the film formed from the He–H2–CH4 plasma, the

diamond band was observed at 1323.5 cm�1, with a FWHM

of 19.6 cm�1 indicative of the polycrystalline nature of the

diamond film [21,22]. In addition to the diamond band, the

D-band, G-band of DLC, and G-band of graphite were

observed at 1327.0 cm�1 with a FWHM of 76 cm�1, 1484.0

cm�1 with a FWHM of 130.2 cm�1, and 1591.6 cm�1 with

a FWHM of 46.5 cm�1, respectively.

The ratio of the areas of the diamond peak to G-band of

graphite, ID/IG, is considered an indirect measure of carbon

sp3/sp2 bonding ratio. The ratio ID/IG was 0.73. The Raman

spectrum confirmed the XPS results that the film comprised

diamond. Based on quantitative studies [23 24], we estimate

that the diamond composition of the films was well over

50%.

The Raman spectrum was repeatable as shown in Fig. 11.

A diamond band was observed at 1332 cm�1 for a second

diamond film formed by He–H2–CH4 microwave discharge

plasma CVD. In addition, the G-band of graphite was

observed at 1580.

The results for the film formed with the Ar–H2–CH4

plasmas were similar to those from the He–H2–CH4

plasmas. As shown in Fig. 10c, the diamond film has a

distinguished sharp Lorentzian peak at 1329 cm�1 with a

FWHM of 22.6 cm�1. This peak is typical of polycrystalline

diamond and demonstrates the high quality of the diamond

in the sample [21,22]. The overlapping D-band of DLC, G-

bands of DLC, and graphite were also observed at 1330

(125 FWHM), 1490 (104 FWHM), and 1594 (58 FWHM)

cm�1, respectively. The presence of a shoulder at 1190

cm�1 is possibly due to the nanocrystalline nature of the

deposited film. This feature has been discussed recently as

arising from the amorphous sp3 carbon which is formed as a

precursor in the initial stage of diamond deposition [21–23].

The peak values observed for two samples developed under

the same conditions are given in Table 2. The Raman

spectrum confirmed the XPS results that the film comprised

diamond.

Table 2

The curve fit analysis of Raman spectra recorded on two polycrystalline diamond films synthesized on silicon substrates by a low power (~80 W) microwave

plasma chemical vapor deposition (MPCVD) reaction of a mixture of argon–hydrogen–methane (17.5/80/2.5%)

Diamond band Amorphous sp3 carbon D-band G-band of DLC G-band of graphite

Position

(cm�1)

FWHM

(cm�1)

Position

(cm�1)

FWHM

(cm�1)

Position

(cm�1)

FWHM

(cm�1)

Position

(cm�1)

FWHM

(cm�1)

Position

(cm�1)

FWHM

(cm�1)

1328.2 15.5 1218.0 140.7 1332.1 85.9 1486.0 141.3 1597.1 44.9

1329.5 17.7 1226.7 115.9 1335.6 91.6 1489.9 133.2 1599.7 48.3

R. Mills et al. / Thin Solid Films 478 (2005) 77–9086

3.4. SEM microstructure

Scanning electron microscopic images showed that the

diamond films from the He–H2–CH4 plasmas were poly-

crystalline. Individual 2–3 Am diameter clusters comprising

6–10 grains of diamond each having a grain size of 15–300

nm were observed as shown in Fig. 12. In some areas of the

surface, the clusters were embedded in an amorphous

matrix. With increasing reaction time, the clusters increased

in size, agglomerated, and then fused to form a thick

uniform film as shown in Fig. 13. The cross-sectional view

of the film formed after 48 h showed a typical columnar

growth pattern. The film thickness was in the range of 100–

150 Am per 12–16 h of deposition time.

Scanning electron microscopic images of the films from

the Ar–H2–CH4 plasmas are shown in Fig. 14a and b. The

polycrystalline films grown over a period of 15 h consists of

600 nm particles agglomerated and fused together to form a

continuous thick film. Columnar growth pattern was

observed in the film and the thickness was in the range of

10–12 Am. A growth rate of 0.6–0.8 Am/hr was observed

under the experimental conditions.

3.5. XRD

The XRD pattern of the He–H2–CH4 plasma-formed

diamond film for 2h=40–808 is shown in Fig. 15a. Since the

X-rays penetrated the film, despite complete coverage, the

reflections from the sample were dominated by those from

the silicon substrate. Diamond peaks were observed at

2h=42.28 (rhombohedral (104)), 43.98 (cubic (111)), and

75.38 (cubic (220)) [25,26]. The strongest diamond line

showed very strong cubic (111) orientation. The peaks at

61.88, 63.08, and 76.28 are assigned to SiO2/SiC, SiO2, and

Fig. 12. The surface SEM image of the He–H2–CH4 plasma-formed film at

500X.

Si, respectively. The broad peak at 69.48 belongs to the Si

substrate plane and SiC which is formed as a transition layer

[25–28].

The XRD pattern of the Ar–H2–CH4 plasma-formed

diamond film for 2h=40–808 is shown in Fig. 15b. Diamond

peaks were observed at 2h=43.98 (111) and 75.68 (220)

[25,26]. The strongest diamond line showed very strong

cubic (111) orientation. The broad peak at 708 belongs to theSi substrate plane [25–28].

3.6. Visible spectrum and line broadening

The visible spectrum (275–700 nm) recorded on a He–

H2–CH4 (48.2/48.2/3.6%) microwave discharge plasma is

shown in Fig. 16. The H a, h, g, and y lines observed at

656.3, 486.2, 434.1, and 410.2 nm, respectively, were

significantly broadened as discussed infra. He I emission

lines were observed at 667.8 and 468.57 nm. A progression

of CH vibrational peaks was observed at 389 and 431 nm.

The C3 band was observed at 405.4 nm, and the C2 Swan

band (A3jgYX3jA), Dv=�1 was observed at 468.2 nm.

The H:C2, CH:C2, and CH:C3 peak intensity ratios were

similar to those reported by Elliott et al. [3] that were

correlated with the optimal conditions for diamond film

quality and growth rate during diamond CVD using CH4–

CO2 plasmas.

The Doppler-broadened line shape for atomic hydrogen

has been studied on many sources such as hollow cathode

[29,30] and RF [31,32] discharges. The energetic hydrogen

atom densities and energies were calculated from the

intensities and widths of the 656.3 nm Balmer a line

Fig. 13. The cross-sectional SEM of the He–H2–CH4 plasma-formed

diamond film at 200X.

Fig. 14. SEM images of the Ar–H2–CH4 plasma-formed diamond film. (a)

Surface image at 1000X. (b) The cross-sectional SEM of the Ar–H2–CH4

plasma-formed diamond film at 5000X.

R. Mills et al. / Thin Solid Films 478 (2005) 77–90 87

emitted from microwave discharge plasmas of H2 compared

with each of Xe–H2 (90/10%) and He–H2 (90/10%) as

shown previously [9,10,16]. The average He–H2 Doppler

half-width was not appreciably changed with pressure. The

corresponding energy of 180–210 eV and the number

density of 5�1014F20% atoms/cm3, depending on the

pressure, were significant compared to only c3 eV and

7�1013F20% atoms/cm3 for pure hydrogen, even though

Fig. 15. The X-ray diffraction (XRD) patterns of diamond films. (a) The XRD pat

XRD pattern of an Ar–H2–CH4 plasma-formed diamond film for 2h=40–808.

10 times more hydrogen was present. Xe did not serve as a

catalyst, and the plasma was much less energetic. Xe–H2

showed no excessive broadening corresponding to an

average hydrogen atom temperature of c3 eV, and the

atom density was also low, 3�1013F20% atoms/cm3.

Significant broadening was also observed for the He–H2–

CH4 (48.2/48.2/3.6%) plasma. The average hydrogen atom

temperature and density were high, 120–140 eV and

3�1014F20% atoms/cm3, respectively, compared to c3

eV and 2�1013F20% atoms/cm3 for a H2–CH4 plasma. The

percentage of hydrogen was varied from 10% to 50% with

CH4 comprising the other component of the mixture. No

broadening was observed with any of these plasmas.

Only the hydrogen lines were broadened. For example,

the 667.816 nm He I line width was also recorded with the

high resolution (F0.006 nm) visible spectrometer on He–H2

(90/10%) and He microwave discharge plasmas. No broad-

ening was observed in either case. Whereas, in each case,

the broadening of the H h, g, and y lines were equivalent tothat of the H a line.

We have assumed that Doppler broadening due to

thermal motion was the dominant source to the extent that

other sources may be neglected. This assumption was

confirmed when each source was considered. In general,

the experimental profile is a convolution of a Doppler

profile, an instrumental profile, the natural (lifetime) profile,

a Stark profile, a van der Waals profile, a resonance profile,

and fine structure. The electron density recorded with a

Langmuir probe was five orders of magnitude too low for

detectable Stark broadening, and the contribution from each

remaining source was determined to be below the limit of

detection [9,10,16].

tern of a He–H2–CH4 plasma-formed diamond film for 2h=40–808. (b) The

Fig. 16. The visible spectrum (275–700 nm) recorded on a He–H2–CH4 microwave discharge plasma.

R. Mills et al. / Thin Solid Films 478 (2005) 77–9088

3.7. EUV spectra

EUV spectra in the Lyman a region of the cell emission

from the hydrogen plasma (dotted line) and the hydrogen

plasma to which 5% helium or argon was added (solid line)

is shown in Fig. 17a and b, respectively. Upon the addition

Fig. 17. EUV spectra of hydrogen alone and with the addition of helium or

argon. (a) The EUV spectrum (110–180 nm) of the cell emission from the

hydrogen plasma (dotted line) and the hydrogen plasma to which 5%

helium was added (solid line). (b) The EUV spectrum (90–127 nm) of the

cell emission from the hydrogen plasma (dotted line) and the hydrogen

plasma to which 5% argon was added (solid line).

of 5% helium or argon, the hydrogen Lyman a emission

intensity was observed to increase by about an order of

magnitude. A concomitant decrease in the intensity of

molecular hydrogen was observed. In contrast, essentially

no effect was observed for the addition of 5% xenon to the

hydrogen plasma. Thus, not only the energy, but also the H

density uniquely increases in He–H2 and Ar–H2 plasmas.

4. Discussion

In the previously developed CH4–H2–system and varia-

tions thereof, diamond formation occurs within a small

domain about the CO tie line. Stringent conditions of a large

excess of hydrogen, diamond seeding, and an elevated

temperature are required. Similarly, in the CO2–CH4 system,

diamond only formed within a range of a few percent from a

50/50% mixture. We observed that diamond was very

reproducibly formed from a CH4 carbon source with a He–

or Ar–H2 plasma without the requirements of diamond

seeding, an elevated temperature, high power, or an excess

of hydrogen, or any particular former set of stringent

conditions. Thus, a potential advancement in thin film

diamond deposition has been shown.

It was reported previously that microwave He–H2 and

Ar–H2 plasmas showed extraordinary broadening corre-

sponding to an average hydrogen atom temperature of 180–

210 and 110–130 eV, respectively, compared to c3 eV for

plasmas of pure H2, Ne–H2, Kr–H2, and Xe–H2 [9,10]. The

energy of H formed in the He–H2–CH4 microwave plasma

was also found to be very high, 120–140 eV, as shown

previously [9,10]. It was found that noble gas–hydrogen

plasmas that were energetic (He–H2 and Ar–H2 plasmas)

formed diamond films when CH4, a carbon source, was

present. Whereas, plasmas that did not form fast H (Ne–H2,

R. Mills et al. / Thin Solid Films 478 (2005) 77–90 89

Kr–H2, and Xe–H2 plasmas) formed no carbon film or

predominately graphite films. It was also found that upon

the addition of 5% helium or argon to a hydrogen plasma,

the Lyman a emission was observed to increase by about an

order of magnitude, whereas xenon control had no effect.

Thus, not only the energy, but also the H density uniquely

increases in He–H2 and Ar–H2 plasmas. This indicates that

the mechanism may be based on energetic hydrogen formed

in the plasma reaction. Diamond and DLC are metastable

materials; thus, continuous bombardment of the surface with

energetic species that produce thermal and pressure spikes at

the growth surface is required for deposition of diamond,

DLC, and related films [33]. By quenching a beam of C+

ions accelerated in an ultrahigh vacuum to a negatively

biased substrate, Aisenberg and Chabot [34] were able to

deposit DLC films for the first time. Rather than resulting in

commercially useful processes, subsequently developed

beam-type and sputtering production methods are essen-

tially used for research. Exemplary methods discussed by

Grill and Meyerson [35] are single low-energy beams of

carbon ions, dual ion beams of carbon and argon, ion

plating, RF sputtering or ion-beam sputtering from carbon/

graphite targets, vacuum-arc discharges, and laser ablation.

Using sputter deposition, amorphous DLC coatings can be

prepared at low temperature due to high ion bombardment

during the deposition of carbon. The absence of ion

bombardment during carbon deposition leads to soft,

conductive carbon films with no diamond-like properties.

It has been shown that films with diamond-like properties

are produced at ion energies of about 100 eV [36,37].

Bombardment of the depositing carbon film by energetic H

that produce a bthermal spikeQ has recently been confirmed

[8] as the primary mechanism for the formation of diamond

wherein the fast H was formed by charge acceleration in a

high field. Here, an H atom-energy transfer to carbon atoms

of a few 100 eV was found to be optimal for the transition of

carbon films to diamond.

5. Conclusion

Polycrystalline diamond films were synthesized on

silicon substrates without diamond seeding by a very

low power (~40–80 W) microwave plasma continuous

vapor deposition (MPCVD) reaction of a mixture of He–

H2–CH4 (48.2/48.2/3.6%) or Ar–H2–CH4 (17.5/80/2.5%).

However, predominately graphitic carbon films or no films

formed when Ne, Kr, or Xe was substituted for He or Ar.

The ToF-SIMS of the films from He–H2–CH4 and Ar–H2–

CH4 plasmas showed essentially no oxide and more

extensive carbon clusters. The XPS of the valence band

region and C 1s energy loss region, the Raman diamond

peak at 1323.5 cm�1 with a FWHM of less than 20 cm�1,

and the SEM characterization showed that the films was

polycrystalline diamond with a high growth rate of about

10 Am/h for He–H2–CH4.

Diamond is proposed to form from CH4 by the catalytic

reaction of He+ or Ar+ with atomic hydrogen to form

energetic plasmas which have extraordinarily fast H,

whereas Ne–H2, Kr–H2, and Xe–H2 plasmas do not have

fast H and do not produce diamond. Fast H formed in the

He–H2–CH4 microwave plasma showed extraordinary

Balmer a line broadening corresponding to an average

hydrogen atom temperature of 120–140 eV. It was also

found that the H density uniquely increases in He–H2 and

Ar–H2 plasmas. Bombardment of depositing carbon to

cause a shift in equilibrium may be the basis of the

formation of the diamond film from CH4. Without diamond

seeding, production of polycrystalline diamond films on

heterogeneous substrates was achieved under relatively low-

temperature, low-power, mild conditions using the uniquely

energetic He– and Ar–H2 plasmas with CH4 as a carbon

source.

Acknowledgments

Thanks to A. Miller of Lehigh University for XPS

analysis and very useful discussions, and V. Pajcini of

Charles Evans & Associates and O. Klueva of Jobin Yvon

for Raman analysis and useful discussions.

References

[1] P.W. May, Philos. Trans. R. Soc. Lond., A 358 (2000) 473.

[2] J.O. Orwa, S. Prawer, D.N. Jamieson, J.L. Peng, J.C. McCallum, K.W.

Nugent, Y.J. Li, L.A. Bursill, Appl. Phys. 90 (2001) 3007.

[3] M.A. Elliott, P.W. May, J. Petherbridge, S.M. Leeds, M.N.R. Ashfold,

W.N. Wang, Diamond Relat. Mater. 9 (2000) 311.

[4] J.R. Petherbridge, P.W. May, S.R.J. Pearce, K.N. Rosser, M.N.R.

Ashfold, J. Appl. Phys. 89 (2001) 1484.

[5] J. Petherbridge, P.W. May, S.R.J. Pearce, K.N. Rosser, M.N.R.

Ashfold, Diamond Relat. Mater. 10 (2001) 393.

[6] P.K. Bachmann, D. Leers, H. Lydtin, D.U. Wiechert, Diamond Relat.

Mater. 1 (1991) 1.

[7] P.K. Bachmann, H.G. Hagemann, H. Lade, D. Leers, F. Picht, D.U.

Wiechert, Mater. Res. Soc. Proc. 339 (1994) 267.

[8] Y. Lifshitz, Th. Kfhler, Th. Frauenheim, I. Guzmann, A. Hoffman,

R.Q. Zhang, X.T. Zhou, S.T. Lee, Science 297 (2002) 1531.

[9] R.L. Mills, P. Ray, B. Dhandapani, R.M. Mayo, J. He, J. Appl. Phys.

92 (2002) 7008.

[10] R.L. Mills, P. Ray, B. Dhandapani, J. He, IEEE Trans. Plasma Sci. 31

(2003) 338.

[11] R.L. Mills, P. Ray, B. Dhandapani, J. He, J. Phys., D, Appl. Phys. 36

(2003) 1535.

[12] R.L. Mills, P. Ray, B. Dhandapani, M. Nansteel, X. Chen, J. He, J.

Mol. Struct. 643/1–3 (2002) 43.

[13] R.L. Mills, P. Ray, J. Dong, M. Nansteel, B. Dhandapani, J. He, Vib.

Spectrosc. 31/2 (2003) 195.

[14] J.H.D. Rebello, D.L. Straub, V.V. Subramaniam, J. Appl. Phys. 72

(1992) 1133.

[15] P. Sander, M. Altebockwinkel, W. Storm, L. Wiedmann, A.

Benninghoven, J. Vac. Sci. Technol., B 7 (1989) 517.

[16] R.L. Mills, P. Ray, New J. Phys. 4 (2002) 22.1 (http://www.njp.org).

[17] J. Tadic, I. Juranic, G.K. Moortgat, J. Photochem. Photobiol., A

Chem. 143 (2000) 169.

R. Mills et al. / Thin Solid Films 478 (2005) 77–9090

[18] R.L. Kelly, J. Phys. Chem. Ref. Data 16 (Suppl. 1) (1987) (Part III).

[19] F.F. Chen, in: R.H. Huddleston, S.L. Leonard (Eds.), Plasma

Diagnostic Techniques, Academic Press, New York, NY, 1965.

[20] Provided by A. Miller, Zettlemoyer Center for Surface Studies,

Sinclair Laboratory, Lehigh University Bethlehem, PA.

[21] J. Wagner, Ch. Wild, P. Koidl, Appl. Phys. Lett. 59 (1991) 779.

[22] M.S. Dresselhans, G. Dresselhans, M.A. Pimenta, P.C. Eklund, in:

M.J. Pelletier (Ed.), Raman Scattering in Carbon Materials in

Analytical Applications of Raman Spectroscopy, Blackwell Science,

Oxford, 1999, p. 389.

[23] S.M. Leeds, T.J. Davis, P.W. May, C.D.O. Pickard, M.N.R. Ashfold,

Diamond Relat. Mater. 7 (1998) 233.

[24] K.W.R. Gilkes, S. Prawer, K.W. Nugent, J. Robertson, H.S. Sands, Y.

Lifshitz, X. Shi, J. Appl. Phys. 87 (2000) 7283.

[25] P.D. Ownby, X. Yang, J. Liu, J. Am. Ceram. Soc. 75 (1992) 1876.

[26] J. Thewlis, A.R. Davey, Philos. Mag., Part B 1 (1956) 409.

[27] Y.X. Wang, J. Wen, Z. Guo, Y.Q. Tang, H.G. Tang, J.X. Wu, Thin

Solid Films 338 (1999) 93.

[28] M. Kitabatake, M. Deguchi, T. Hirao, J. Appl. Phys. 74 (1993) 4438.

[29] I.R. Videnovic, N. Konjevic, M.M. Kuraica, Spectrochim. Acta, Part

B: Atom. Spectrosc. 51 (1996) 1707.

[30] S. Alexiou, E. Leboucher-Dalimier, Phys. Rev., E 60 (1999) 3436.

[31] S. Djurovic, J.R. Roberts, J. Appl. Phys. 74 (1993) 6558.

[32] S.B. Radovanov, K. Dzierzega, J.R. Roberts, J.K. Olthoff, Appl. Phys.

Lett. 66 (1995) 2637.

[33] C. Weissmantel, in: K.J. Klabunde (Ed.), Thin Films from Free Atoms

and Molecules, Academic Press, New York, NY, 1985, p. 153.

[34] S. Aisenberg, R. Chabot, J. Appl. Phys. 42 (1971) 2953.

[35] A. Grill, B. Meyerson, in: K.E. Spear, J.P. Dismukes (Eds.), Synthetic

Diamond: Emerging CVD Science and Technology, John Wiley and

Sons, New York, NY, 1994, p. 91.

[36] J. Ishikawa, K. Ogawa, K. Miyata, T. Takagi, Nucl. Instrum. Methods

B21 (1987) 205.

[37] F. Rossi, B. Andre, Proceedings of 8th International Conference on

Ion and Plasma Assisted Techniques (IPAT ’91), Brussels, Belgium,

May 21–24, CEP Consultants, Edinburgh, UK, 1991, p. 43.