Plasma Chemical Vapor Deposition of Nanocrystalline Diamond

Encyclopedia of Nanoscience and Nanotechnology

www.aspbs.com/enn

Plasma Chemical Vapor Deposition of Nanocrystalline Diamond Katsuyuki Okada National Institute for Materials Science, Tsukuba, Ibaraki, Japan

CONTENTS 1. Introduction 2. Growth of Nanocrystalline Diamond 3. Spectroscopic Characterization 4. Plasma Diagnostics Glossary References

1. INTRODUCTION Covalently bonded disordered thin-film materials have been of considerable interest from fundamental and applied perspectives in the last twenty years, since the chemical vapor deposition (CVD) of diamond was developed, followed by fullerenes and carbon nanotubes [1, 2]. Amorphous and nanostructured carbon films are being extensively studied for electron emitters, cold-cathode sources, hard lowfriction coatings, etc. From fundamental perspectives, on the other hand, the structure of these materials contains both three-fold coordinated (sp2 -bonded) and four-fold coordinated (sp3 -bonded) carbon atoms. Nanocrystalline diamond films have also drawn remarkable attention [3] because they have a low coefficient of friction and low electron emission threshold voltage. The small grain size (approximately 5–100 nm) gives films valuable tribology and fieldemission properties, compared with those of conventional polycrystalline diamond films. Furthermore, it has been proposed that applications for microelectromechanical systems (MEMS) devices, metal-semiconductor field effect transistors MESFETs, electrochemical electrodes, and biochemical devices can be created by taking advantage of these excellent properties [4–6]. Recent plasma dry processes (e.g., deposition, coating, etching) require a wide area and high density plasma at low pressures (<1 Torr) [7]. An electron cyclotron resonance (ECR) plasma was first developed to meet these conditions. Subsequently, a helicon-wave excited plasma was employed. ISBN: 1-58883-064-0/$35.00 Copyright © 2004 by American Scientific Publishers All rights of reproduction in any form reserved.

It was found [8] that the density of an inductively coupled plasma (ICP) becomes high at low pressures. Amorim et al. [9] reported that the plasma density of ICP after the transition from a low-density E-discharge to a high-density H-discharge reaches 1012 cm−3 . It is necessary in plasmaenhanced chemical vapor deposition (PE-CVD) to get a sufficient radical flux for deposition. Thus the ICP is thought to be a promising method for PE-CVD at low pressures. In this review article, the synthesis of nanocrystalline diamond by low-pressure ICP-CVD and its spectroscopic characterizations are mainly described. Plasma diagnostics with a Langmuir probe are also reviewed.

2. GROWTH OF NANOCRYSTALLINE DIAMOND 2.1. Conventional Microwave Plasma CVD A microwave plasma is most commonly used for the PE-CVD of diamond films [10], in which the conventional pressure of deposition is on the order of a few 10 Torr. A CH4 /H2 mixture, in which the CH4 content ([CH4 ]) is usually less than 5%, leads to micrometer-size polycrystalline diamond films. One methodology for synthesizing nanocrystalline diamond films is to increase the [CH4 ] to 10% while the substrate temperature is kept constant [11–13]. The resultant morphology changes from faceted microcrystals to ball-shaped nanocrystals. Another methodology is to substitute a noble gas for hydrogen while the [CH4 ] is kept constant. Gruen et al. [14, 15] demonstrated the synthesis of nanocrystalline diamond films in a CH4 /Ar or C60 /Ar microwave discharge, without the addition of hydrogen. Extensive characterizations with X-ray diffraction (XRD), transmission electron microscopy (TEM), and electron energy loss spectroscopy (EELS) showed [16] that the films consist of a pure crystalline diamond phase with a grain size ranging from 3 to 20 nm. They also proposed from the plasma diagnostics [17] and the theoretical calculations [18] that the C2 dimer could Encyclopedia of Nanoscience and Nanotechnology Edited by H. S. Nalwa Volume 8: Pages (691–702)

692 be responsible for the growth of the very fine-grained diamond films, whereas the CH3 radical is generally believed to be a precursor for the diamond growth. Recently Butler et al. [4] reported the deposition of nanocrystalline diamond films with the conventional deposition conditions for micrometer-size polycrystalline diamond films. The substrate pretreatment by the deposition of a thin H-terminated a-C film, followed by the seeding of nanodiamond powder, increased the nucleation densities to more than 1012 /cm2 on a Si substrate. The resultant films were grown to thicknesses ranging from 100 nm to 5 m, and the thermal conductivity ranged from 2.5 to 12 W/cm K. In addition to microwave plasma, direct current (dc) plasma [19], hot-filament [20], magnetron sputtering [21], and radiofrequency (rf) [22–24] plasmas were utilized for nanocrystalline diamond deposition. Amaratunga et al. [23, 24], using CH4 /Ar rf plasma, reported that single-crystal diffraction patterns obtained from nanocrystalline diamond grains all show {111} twinning. From a theoretical point of view, the stability of nanocrystalline diamond was discussed by several authors. Badziag et al. [25] pointed out that, according to semi-empirical quantum chemistry calculations, sufficiently small nanocrystalline diamond (3–5 nm in diameter) may be more stable than graphite by forming C-H bonds at the growing surface. Barnard et al. [26] performed the ab initio calculations on nanocrystalline diamond up to approximately 1 nm in diameter. The results revealed that the surfaces of cubic crystals exhibit reconstruction and relaxations comparable to those of bulk diamond, and the surfaces of the octahedral and cubooctahedral crystals show the transition from sp3 to sp2 bonding.

Plasma CVD of Nanocrystalline Diamond

by-product O atoms with H2 and CH4 are highly responsible for the enhanced diamond growth. A 13.56-MHz low-pressure ICP system [31–33] has been utilized as a radical source for PE-CVD and applied to the preparation of diamond films from a CH4 /H2 plasma. CO has been added to a CH4 /H2 plasma to clarify the effect of oxygen-containing radicals on the diamond growth. A schematic view of the apparatus is given in Figure 1. It consisted of a water-cooled quartz tube surrounded by a Teflon tube and a stainless-steel growth chamber. An inductively coupled plasma was generated by applying 13.56MHz rf powers of 1 kW to a three-turn helical coil wound around the Teflon tube. The chamber was pre-evacuated to 5 × 10−5 Torr by a turbo molecular pump (200 liters/s). A Faraday shield was inserted between the rf coil and the quartz tube to rule out electrostatic coupling from the rf coil to the plasma. The quartz tube was wrapped with a grounded copper plate slitted at regular intervals along the azimuthal direction. A substrate holder was manipulated from the top of the chamber. Molybdenum plates and silicon (100) wafers ( 10 mm) were used as a substrate. The substrate was heated by a tungsten filament. The substrate temperature (Ts ) was monitored with a sheathed thermocouple and kept at a prescribed temperature within ±1  C by a proportional integral derivative (PID) controller. The deposition conditions were as follows. The flow rates of CH4 and H2 were kept at 4.5 and 75 sccm, respectively, whereas

2.2. Low-Pressure Plasma CVD Whereas a microwave plasma is most commonly used for the PE-CVD of diamond films, an ECR is the only plasma that is used for diamond deposition below 1 Torr [27–29]. Although Bozeman et al. [30] reported diamond deposition at 4 Torr with the use of a planar ICP, there have been a few reports that describe the synthesis of diamond by lowpressure ICP. Okada et al. [31–33] first reported the synthesis of nanocrystalline diamond particles in a low-pressure CH4 /CO/H2 ICP, followed by Teii and Yoshida [34], with the same gas-phase chemistry. In the PE-CVD of diamond films, the addition of oxygen to a CH4 /H2 system was intended to get a high growth rate and/or an improvement of the quality of resultant films [35–38]. Several authors proposed the role of oxygen in the diamond growth. Mucha et al. [37] pointed out that oxygen additions to a CH4 /H2 system increase the concentration of atomic hydrogen; correspondingly, amorphous and graphitic carbon, which would otherwise inhibit diamond growth, is better suppressed. From a detailed chemical kinetics model that describes both gas-phase and surface processes occurring in diamond CVD, Frenklach and Wang [38] postulated that OH radicals and atomic oxygen are considered to gasify sp2 carbon and to suppress the formation of aromatic species in the gas phase. Teii et al. [39] pointed out that the OH radicals resulting predominantly from loss reactions of the

Figure 1. Schematic description of the low-pressure inductively coupled rf plasma CVD system. Adapted with permission from [33], K. Okada et al., J. Mater. Res. 14, 578 (1999). © 1999, Materials Research Society.

Plasma CVD of Nanocrystalline Diamond

693

the flow rate of CO ([CO]) was varied between 0, 1.0, 5.0, and 10 sccm, respectively. The total gas pressure (Pr) was varied from 45 to 50 mTorr. The Ts was kept at 900  C. The deposition duration was 2 h.

3. SPECTROSCOPIC CHARACTERIZATION 3.1. SEM and TEM Figure 2 presents the SEM photographs of the resultant deposits on a Si(100) substrate. Figure 2a, b, c, and d correspond to [CO] = 0, 1.0, 5.0, and 10 sccm, which are labeled as samples A, B, C, and D, respectively. The morphology of sample A was scale-like as shown in Figure 2a, and crystal facets were not clearly seen. When CO was added to a CH4 /H2 plasma, particles 200–300 nm in diameter with scale-like deposits appeared as shown in Figure 2b. With the increase in [CO], only particles were deposited as shown in Figure 2c and d. The background of the particles was the substrate. The diameter of the particles was 200–700 nm. The detailed observation reveals that the particles consist of small particles about several tens of nanometers in diameter, and that the sizes are almost the same regardless of increasing [CO]. It is therefore speculated that increasing [CO] makes the supersaturation degree of carbon large, and then the number of encountered particles is increased. The TEM image of sample B is shown in Figure 3a. Figure 3b shows the transmission electron diffraction (TED) pattern corresponding to Figure 3a. In the TEM image, a scale-like deposit was mainly observed. The ring pattern in the TED indicates that the deposit is composed of a large number of microcrystallites. Figure 4a presents the TEM image of sample C. The corresponding TED pattern is presented in Figure 4b. Only particles 200–500 nm in diameter were observed in the TEM image. The TED pattern was spotty compared with that of Figure 3b. The TEM images clearly show that each particle is composed of small particles about several tens of nanometers in diameter. The interplanar spacings (d values) obtained from the TED patterns are summarized in Table 1, where they are compared with the experimental values reported for diamond. The agreement between the values obtained and those for diamond is within the experimental error. No reflections due to nondiamond planes were observed in the patterns of sample C, whereas the reflections of nondiamond planes were observed in the pattern of sample B. Although these reflections are not always assigned to graphite, it is speculated that they are attributed to disordered microcrystalline graphite. From the TEM image of sample B, the size of the disordered microcrystalline graphite was estimated to be about several hundred nanometers. It is therefore concluded that scalelike deposits consist of disordered microcrystalline graphite, whereas particles are composed of only diamond microcrystallites.

3.2. XRD The X-ray diffraction (XRD) pattern for sample D is shown in Figure 5. D(111) and D(220) denote the diffraction peaks of the diamond plane. The diffraction peaks of the silicon

Figure 2. SEM micrographs of the obtained deposits. (a) [CH4 ]/ [CO] = 4.5/0 sccm. (b) [CH4 ]/[CO] = 4.5/1.0 sccm. (c) [CH4 ]/[CO] = 4.5/5.0 sccm. (d) [CH4 ]/[CO] = 4.5/10 sccm. Reprinted with permission from [33], K. Okada et al., J. Mater. Res. 14, 578 (1999). © 1999, Materials Research Society.

substrate also appeared. Although the incident X-ray angle was quite shallow, it reached the substrate because of the imperfect coverage of the deposit as shown in Figure 2d. The peaks corresponding to graphite were not observed. Although no clear crystal facets can be seen in the SEM observations, the diffraction pattern clearly reveals that the deposit is neither amorphous carbon nor graphite, but diamond. The crystallite size was estimated to be approximately

694

Plasma CVD of Nanocrystalline Diamond Table 1. Interplanar spacings from the TED patterns and the reported values. Sample B 0.207 0.124 0.114 0.106 0.090 0.081 0.072 0.061

Sample C

Diamonda

(hkl)

0207 0124

0206 0126

(111) (220)

0106

010754 008916 008182 007280b 006864b

(311) (400) (331) (422) (511), (333)

a Joint Committee on Powder Diffraction Standards (JCPDS) Powder Diffraction File, 6-675. b Values are calculated from a = 03567 nm.

Figure 3. (a) TEM micrograph of sample B ([CH4 ]/[CO] = 4.5/ 1.0 sccm). (b) TED pattern corresponding to (a). Reprinted with permission from [33], K. Okada et al., J. Mater. Res. 14, 578 (1999). © 1999, Materials Research Society.

20 nm from the full width at half-maximum (FWHM) of the diamond peaks with the use of Scherrer’s equation [40]. It accordingly implies that the particles that are several hundred nanometers in diameter consist of smaller particles several tens of nanometers in diameter. This is consistent with the TEM observation. The addition of CO to CH4 /H2 plasmas is considered to produce oxygen-containing radicals, such as atomic oxygen, the OH radical, and the CO radical itself, in plasmas. As mentioned previously, the morphological change from a scale-like deposit to a particle took place with the CO additive. Furthermore, the number of particles encountered increased with an increase in [CO]. According to the TED and XRD patterns, nondiamond carbon was effectively removed with an increase in [CO]. We therefore presume that oxygen-containing radicals produced by the addition of CO play an effective role in the removal of nondiamond carbon in the diamond growth conditions and that the CO

Figure 4. (a) TEM micrograph of sample C ([CH4 ]/[CO] = 4.5/ 5.0 sccm). (b) TED pattern corresponding to (a). Reprinted with permission from [33], K. Okada et al., J. Mater. Res. 14, 578 (1999). © 1999, Materials Research Society.

additive makes the supersaturation degree of carbon large. This is consistent with the previously reported hypotheses [37, 38] that OH radicals and atomic oxygen gasify sp2 carbon and that they suppress the formation of amorphous carbon and graphitic carbon.

3.3. Raman Spectroscopy Raman spectroscopy is a nondestructive method for the study of the vibrational band structure of materials. It has been extensively used for the characterization of diamond [41–44], graphite [45], and diamond-like carbon (DLC) [42, 43, 46–49]. So far Raman scattering is the most popular technique for identifying sp3 bonding in diamond and sp2 bonding in graphite and DLC. Although Raman scattering should represent the phonon density of state (PDOS), weighted by a coupling parameter, Raman spectroscopy of sp2 -bonded carbon with excitation wavelength in the visible range (514, 488 nm, etc.) does not provide a good representation of the PDOS [50]. Since the local sp2 -bonded carbon energy gap of ∼2 eV is comparable to the energy of visible Raman excitation, the sp2 -bonded carbon network exhibits the electronic - ∗ transition resonance enhancement in the Raman cross section [43, 48–50]. On the other

Figure 5. XRD pattern of sample D ([CH4 ]/[CO] = 4.5/10 sccm). Reprinted with permission from [57], K. Okada et al., J. Appl. Phys. 88, 1674 (2000). © 2000, American Institute of Physics.

Plasma CVD of Nanocrystalline Diamond

hand, the sp3 -bonded carbon does not exhibit such a resonance effect because of the higher local gap of ∼5.5 eV. As a result, the Raman spectra obtained with visible excitation are completely dominated by the sp2 -bonded carbon. However, Raman spectroscopy with ultraviolet (UV) excitation has recently been used to characterize DLC [50–53]. The advantage of UV Raman spectroscopy is to eliminate the resonance effect of sp2 -bonded carbon. Excitation of 244 nm (5.1 eV) is far from the energy of previously mentioned sp2 -bonded carbon resonance. It is accordingly expected that the resonance Raman scattering from sp2 -bonded carbon is suppressed and that the signal from sp3 -bonded carbon is possibly increased. UV Raman spectra of tetrahedral amorphous carbon (t-aC) films containing sp3 bonding dominantly exhibit a peak at ∼1150 cm−1 attributed to an sp3 -bonded carbon network [50–52]. Okada et al. [54] first applied UV Raman spectroscopy to the characterization of bonding structures of nanocrystalline diamond. Sun et al. [55] reported UV Raman characteristics of nanocrystalline diamond films with different grain sizes from 120 to 28 nm. They pointed out that the downshift and broadening of the diamond peak with decreasing grain size are consistent with the phonon confinement model. On the other hand, Prawer et al. [56] reported the visible Raman spectra of nanocrystalline diamond powder with a size of ∼5 nm. The peaks at 500, 1140, 1132, and 1630 cm−1 are compared with the calculated VDOS of diamond, and the origin of features in the vibrational spectrum from nanocrystalline diamond is suggested. In Raman measurements [57], the 514-nm line of an Ar+ laser, the 325-nm line of a He-Cd laser, and the 244-nm line of an intracavity frequency-doubled Ar+ laser were employed. The incident laser beam was directed onto the sample surface under the back-scattering geometry, and the samples were kept at room temperature. In the 514-nm excitation, the scattered light was collected and dispersed in a SPEX 1403 double monochromator and detected with a photomultiplier. The laser output power was 300 mW. In the 325- and 244-nm excitations, the scattered light was collected with fused silica optics and was analyzed with a UV-enhanced CCD camera, using a Renishaw microRaman system 1000 spectrometer modified for use at 325 and 244 nm, respectively. A laser output of 10 mW was used, which resulted in an incident power at the sample of approximately 1.5 mW. The spectral resolution was approximately 2 cm−1 . That no photoalteration of the samples occurred during the UV laser irradiation was ensured by confirming that the visible Raman spectra were unaltered after the UV Raman measurements. Figure 6 shows the 514-nm excited Raman spectra. Figure 6a, b, and c corresponds to sample A, sample B, and sample D, respectively. The Raman spectrum of sample A exhibits two peaks at ∼1355 cm−1 (D peak) and ∼1580 cm−1 (G peak) assigned to sp2 bonding [45]. New peaks appear at ∼1150 cm−1 and ∼1480 cm−1 in sample B. The former peak is due to sp3 bonding. Nemanich et al. [42, 44] assigned the peak at ∼1150 cm−1 to microcrystalline or amorphous diamond. Prawer and Nugent [58] attributed the peak at 1100– 1150 cm−1 to the surface phonon mode of diamond from the EELS studies of the nanophase diamond powder. On the

695

Figure 6. 514-nm excited Raman spectra of the deposits obtained with different gas mixtures. (a) [CH4 ]/[CO] = 4.5/0 sccm. (b) [CH4 ]/ [CO] = 4.5/1.0 sccm. (c) [CH4 ]/[CO] = 4.5/10 sccm. Reprinted with permission from [57], K. Okada et al., J. Appl. Phys. 88, 1674 (2000). © 2000, American Institute of Physics.

other hand, Ferrari and Robertson [59] have recently proposed another interpretation, that the peak at ∼1150 cm−1 is not attributable to nanocrystalline diamond but to sp2 bondings of a trans-polyacetylene oligomer of a given conjugation length. The peak at ∼1480 cm−1 is probably related to amorphous sp2 structures [44], though it is downshifted from the G peak. Both the D and G peaks become relatively small compared with those of sample A. A peak at ∼950 cm−1 is derived from scattering by two transverseoptical (TO) phonons of silicon substrate [43]. The incident laser beam reached the substrate because the substrate was not fully covered with the deposit as shown in Figure 2b. In the Raman spectrum of sample D, the G peak turns into a shoulder and the D peak becomes faint, whereas the peak at ∼1480 cm−1 has a maximum intensity and the peak at ∼1150 cm−1 still remains. However, the three spectra in Figure 6 do not clearly exhibit the 1332 cm−1 diamond peak. These features of Raman spectra imply that the deposit from CH4 /H2 predominantly contains sp2 bonding; the addition of CO to CH4 /H2 produces a mixture of sp2 and sp3 bondings; the increase in [CO] brings about the increase in sp3 bonding. These tendencies are consistent with the TEM and XRD observations, which reveal that the increase in [CO] improves the crystallinity of the deposit. Figure 7 shows a sequence of Raman spectra from sample D with different excitation wavelengths. Figure 7a, b, and c corresponds to 514 nm, 325 nm, and 244 nm, respectively. Compared with the spectrum of 514-nm excitation, the 325-nm excited Raman spectrum exhibits a clear peak at 1332 cm−1 and the remarkable enhancement of the

696

Plasma CVD of Nanocrystalline Diamond

enhancement of the Raman cross section due to sp2 -bonded carbon still remains and that the - ∗ transition in both sp2 and sp3 -bonded carbon is possibly induced. The remarkable enhancement of the peak at 1332 cm−1 and the diminution of the peak at ∼1580 cm−1 in the 244-nm excitation reveals that the resonance Raman effect due to sp2 -bonded carbon is suppressed, whereas the resonant enhancement due to the - ∗ transition of sp3 -bonded carbon is predominantly attained. The 244-nm excited Raman spectra of t-aC films exhibit the appearance of the peak at ∼1150 cm and the increase in the intensity proportional to the amount of sp3 bonding in the films [50, 51]. However, the diamond peak at 1332 cm−1 is enhanced in this study because the deposit obtained is not amorphous carbon but nanocrystalline diamond. The peak at ∼1150 cm−1 is probably disappearing because of the striking enhancement of the diamond peak at 1332 cm−1 .

3.4. HREELS

Figure 7. Raman spectra of sample D ([CH4 ]/[CO] = 4.5/10 sccm) with different excitation wavelengths. (a) 514 nm. (b) 325 nm. (c) 244 nm. Reprinted with permission from [57], K. Okada et al., J. Appl. Phys. 88, 1674 (2000). © 2000, American Institute of Physics.

peak at ∼1580 cm−1 , while the peak at ∼1150 cm−1 turns into a shoulder. In a 244-nm excited Raman spectrum, the peak at 1332 cm−1 is only enhanced, whereas the peak at ∼1580 cm−1 is weakened. Neither a peak nor a shoulder can be recognized at ∼1150 cm−1 . The Raman spectra in Figure 7a are superimposed on a broad photoluminescence background. The luminescence background diminishes with increasing excitation photon energy as shown in Figure 7b and c. The TED and XRD patterns revealed that the deposit is not amorphous carbon but nanocrystalline diamond. Nonetheless, the 514-nm excited Raman spectra do not exhibit a clear diamond peak at 1332 cm−1 , though the peak due to the sp3 -bonded carbon network appears at ∼1150 cm−1 . The Raman cross section of the sp2 -bonded carbon network with visible excitation is resonantly enhanced [43, 48–50]. It consequently makes the 1332 cm−1 diamond peak overlap with the peaks due to sp2 -bonded carbon. The suppression of the dominance of the resonance Raman effect in the 514-nm excitation and the possible increase in the signal from sp3 -bonded carbon are expected to occur in higher photon energies. Wagner et al. [43] mentioned that a resonance enhancement of scattering by sp3 -bonded carbon is expected at an incident photon energy of 4.8 eV, which is close to the onset of the - ∗ transition in both sp2 - and sp3 -bonded carbon. Gilkes et al. [51] stated that the 244-nm photon energy is sufficient to excite the - ∗ transition of both sp2 and sp3 sites. The enhancement of the peaks at both 1332 cm−1 and ∼1580 cm−1 in the 325-nm excitation suggests that the resonance

High-resolution electron energy loss spectroscopy (HREELS) is a method for the study of the vibrational motion of atoms and molecules on and near the surface by the analysis of the energy spectrum of low-energy electrons back-scattered from it [60]. With developments in modern ultra-high-vacuum (UHV) technology and increasing interest in the physical properties of clean and adsorbate-covered surfaces and of the chemical phenomena occurring on these surfaces, scientific interest in HREELS has increased dramatically, and the method has been widely used in recent years [60–65]. HREELS experiments [66] were performed in a UHV chamber. The chamber was pre-evacuated by polyphenylether-oil diffusion pump; the base pressure reached ∼2 × 10−8 Torr. The HREELS spectrometer consisted of a double-pass electrostatic cylindrical-deflectortype monochromator and the same type of analyzer. The energy resolution of the spectrometer is 4–6 meV (32–48 cm−1 ). A sample was transferred from the ICP growth chamber to the HREELS chamber in the atmosphere. It was clipped by a small tantalum plate, which was suspended by tantalum wires. The sample was radiatively heated in vacuum by a tungsten filament placed at the rear. The sample temperature was measured by an infrared ( = 20 m) optical pyrometer. All HREELS measurements were taken at room temperature. The electron incident and detection angles were each 72 to the surface normal. The primary electron energy was 15 eV. Figure 8 presents the HREELS spectra of heated films. Figure 8a, b, and c corresponds to sample A, sample B, and sample D, respectively. In Figure 8a without CO additive, the spectrum has a faint peak at ∼1500 cm−1 derived from the C C stretching mode of threefold bonded carbon atoms [60]. The whole spectrum is similar to that of a single crystal graphite (0001) surface [63]. With CO additive as shown in Figure 8b, one can see a peak at ∼1100 cm−1 , which is assignable to the C-C stretching mode of fourfold bonded carbon atoms [42, 60]. The peak at ∼1500 cm−1 disappeared. With increasing [CO] as shown in Figure 8c, the intensity of the peak at ∼1100 cm−1 became strong. In addition, a shoulder centered at ∼700 cm−1 appeared. These features of the VDOS are consistent with the theoretical results

697

Plasma CVD of Nanocrystalline Diamond

Figure 8. HREELS spectra of the nanocrystalline diamond and diamond-like carbon films with various [CH4 ]/[CO]. (a) [CH4 ]/[CO] = 4.5.0/0. (b) [CH4 ]/[CO] = 4.5/1.0. (c) [CH4 ]/[CO] = 4.5/10 sccm. The elastic peak for (c), reduced by a factor of 25, is shown for comparison. Reprinted with permission from [66], K. Okada et al., Diamond Relat. Mater. 10, 1991 (2001). © 2001, Elsevier Science.

for random network models of t-aC. Tight binding molecular dynamics (TBMD) simulations for a sample with 75% fourfold atoms predict two major broad peaks at ∼700 cm−1 and ∼1100 cm−1 , respectively, corresponding to the bending and stretching vibrations in the network [67]. A rather similar form of the PDOS is obtained in ab initio calculations with a 90% sp3 fraction [68]. According to the characterizations by TEM and XRD, the sample prepared from a CH4 /H2 plasma was composed of nanocrystalline diamond and disordered microcrystalline graphite. Then nondiamond carbon was effectively removed with an increase in [CO]. It is therefore concluded that the VDOS of the nanocrystalline diamond and DLC films extracted from the HREELS data is in qualitative agreement with the characterizations of TEM and XRD. Although the HREELS probes only the region near the surface, the agreement suggests that the surface dynamics do not differ dramatically from those of the bulk.

3.5. EELS Mapping EELS in TEM has been demonstrated to be a powerful technique for performing microanalysis and studying the electronic structure of materials [69]. The energy loss near edge structures (ELNES) is sensitive to the crystal structure. The C-K edge of diamond and that of graphite are typical examples. For trigonal sp2 -bonded carbon, the spectrum within the first 30 eV of the edge can be separated into two broad features, corresponding to the * states between 282 and 288 eV and the  ∗ states between 290 and 320 eV, whereas for tetrahedral sp3 -bonded carbon only the  ∗ peak is observed between 289 and 320 eV [70]. Therefore, one can easily distinguish between diamond and diamond-like carbon with the help of ELNES. Muller et al.

[71] acquired the mapping of sp2 and sp3 states across the silicon-diamond interface by scanning transmission electron microscopy (STEM) and parallel acquisition electron energy loss spectroscopy (PEELS). Bruley’s two-window technique [72], by integrating intensities of the  ∗ and  ∗ peaks, has demonstrated the quantitative analysis of the sp2 content in the samples. Bursill et al. [73–75] carefully performed the structural analysis on nanocrystalline diamond powder and studied the surface and bulk plasmon response derived from the low-loss spectra by using high-resolution TEM (HRTEM)/PEELS and STEM/PEELS. Prawer et al. [56, 76] also reported the EEL spectra of nanocrystalline diamond synthesized by detonation and ion implantation. The HR-TEM observations [77] were performed on a Hitachi HF-3000 at 297 keV, and the EELS measurements [77] were made with a post-column energy filter (GATAN, GIF2002). The vacuum of the microscope was less than 12 × 10−6 Pa, in which the samples were not contaminated with carbon during TEM observations. Two-dimensional arrays of charge-coupled devices (CCDs) were used for the digital recording of TEM images, EEL spectra, and chemical maps. The typical CCD readout times were 5 s for the acquisition of EEL spectra and 50 s for chemical mappings, respectively. The energy resolution of the instrument was approximately 0.5 eV measured as the zero-loss FWHM. Figure 9a shows a HR-TEM image of an outer part of a nanocrystalline diamond particle. Lattice fringes clearly appeared in the enlargement of the left-hand side of the sample as shown in Figure 9b. The d values were 0.206 nm, corresponding to (111) spacings of diamond. The diffractogram derived from the Fourier transformation of the HR-TEM image exhibited a ring pattern, which means the observed sample is polycrystalline. Furthermore, careful observation reveals that the particle consists of smaller subgrains approximately 20–50 nm in diameter. Thus the boundaries are not surface steps of a single crystal. The size of subgrains is comparable to the grain sizes of nanocrystalline diamond films previously reported by several authors [15, 23, 29, 78], in which the grain sizes were found to be in the range of 3–50 nm. The EEL spectrum corresponding to Figure 9a is shown in Figure 10. It exhibits a peak at 290 eV due to  ∗ states, and a slight peak appears at ∼285 eV, corresponding to  ∗ states. The ELNES above 290 eV is similar to that of diamond [70] and is obviously different from that of graphite or sp3 -rich a

b

0.206 nm

10 nm Figure 9. (a) High-resolution transmission electron microscope image of an outer part of a nanocrystalline diamond particle and (b) enlargement of the left-hand side of (a).

698

Plasma CVD of Nanocrystalline Diamond

the ELNES. The width of the sp2 bondings is estimated to be approximately 1 nm from the  ∗ image in Figure 11a. Fallon and Brown reported [79] the presence of amorphous carbon at the grain boundaries of CVD diamond films by TEM observation and EELS analysis. The amorphous carbon is shown to contain almost complete sp2 bonding and to be nonhydrogenated. It was also demonstrated from a theoretical point of view [80] that sp2 bondings are energetically stable in grain boundaries of nanocrystalline diamond. It is consequently considered that the sp2 bondings are possibly localized in the grain boundaries of 20–50-nm subgrains.

Figure 10. Electron energy loss spectrum corresponding to Figure 9a. a1 , a2 , b, and c indicate the four energy windows: 272–277 eV, 277–282 eV, 282–287 eV, and 287–292 eV, respectively.

tetrahedral amorphous carbon [72]. The intensity of the  ∗ peak is much lower than that of the  ∗ peak. Although the  ∗ peak in general includes contributions from both sp2 and sp3 bondings, the  ∗ peak of the EEL spectrum in Figure 10 is considered to be mainly due to sp3 bondings. The use of a narrow energy window positioned on the ELNES signal allows the mapping of the variation in intensity as a function of position within the microstructure. The conventional socalled three-window method [69] was employed to remove the background contribution. The two pre-edge images (272– 277 eV and 277–282 eV) (indicated by a1 and a2 in Fig. 10) were used to obtain an extrapolated background image. The subtraction of the extrapolated background image from the post-edge images (282–287 eV and 287–292 eV) (indicated by b and c) produces the  ∗ image and the  ∗ image, respectively, with removal of the background contribution. The  ∗ image and the  ∗ image corresponding to the TEM image of Figure 9a are shown in Figure 11a and b, respectively. Since the intensity of the  ∗ image was weak compared with that of the  ∗ image, the former was multiplied by 5. The  ∗ image reveals that the intensity is strong around the subgrains, whereas the  ∗ image shows that the intensity is strong within the subgrains. Although sp2 -bonded graphitic layers do not appear clearly with the limitation of resolution of the HR-TEM image, these energy-filtered  ∗ and  ∗ images imply that sp2 bondings are localized around 20–50-nm subgrains. The sp2 bondings around the subgrains are considered to contribute the slight peak at ∼285 eV in a

b

10 nm

π*image

10 nm

σ*image

Figure 11. (a)  ∗ image corresponding to Figure 9a. (b)  ∗ image corresponds to Figure 9a.

4. PLASMA DIAGNOSTICS A Langmuir probe has been utilized to measure plasma parameters (plasma potential (Vp ), electron temperature (Te ), electron density (Ne ), and electron energy distribution function (EEDF)) [81–84]. Although there has been much controversy about the interpretation of Langmuir probe characteristics of rf plasmas, the recent development of a sophisticated Langmuir probe system [85] enables one to obtain reproducible plasma parameters precisely. However, only a few works have been carried out on actual process plasmas used for etching or deposition, because it is difficult to maintain the initial condition of probe tips and/or chamber walls for the measurement reproducibility. Cali et al. [86] reported the Langmuir probe characterization of a low-pressure capacitively coupled CH4 /H2 rf plasma. The structure of an EEDF can help one to understand the electron distribution and the heating mechanism in a rf plasma. The EEDF of a low-pressure rf plasma generally consists of two Maxwellian distributions, fast and slow electrons [87]. This feature is due to the stochastic electron heating on the oscillating plasma-sheath boundary. A Langmuir probe was inserted [88] through one of the side flanges of the chamber as shown in Figure 1. The probe was a cylindrical tungsten wire, which was 0.19 mm in radius and 10 mm in length. The probe tip was supported by a ceramic tube, which was 2.5 mm in diameter and was enclosed by a stainless-steel casing capacitively coupled to the probe tip. Since the mean free path of an electron is large compared with the diameter of the ceramic tube for all investigated pressures, the collisionless sheath approximation can be applied [89]. When the plasma was in the high-density regime (1011 to 1012 cm−3 ), the tip length was shortened to 4 mm to prevent the probe tip from glowing red. The rf fluctuations of the plasma potential were compensated for by self-resonant inductors [85]. The reference probe, which was a stainless-steel casing mounted on the probe shaft, automatically compensated for the plasmaground sheath impedance, fluctuations in the plasma sheath impedance, plasma potential shifts, and low-frequency noise [85]. The plasma-ground sheath impedance can be significant when nonconducting layers (e.g., carbon layers in this study) coat the chamber wall. A single I–V characteristic curve was taken, averaging 120 samples per data point. Between scans the probe tip was cleaned by electron bombardment by pulse biasing it to +150 V. Before every probe measurement, the chamber was cleaned with an O2 plasma, which removed contamination of thin carbon layers on the chamber wall. This probe and chamber cleaning made the

Plasma CVD of Nanocrystalline Diamond

obtained I–V characteristics reproducible, without the need for time-dependent hysteresis. Figure 12a–c shows the variation in the Vp , the Te , and the Ne of a CH4 /H2 plasma with pressure. The Vp and Ne decrease with increasing pressure up to 40 mTorr and inversely increase at 50 mTorr. The Ne increases with increasing pressure up to 40 mTorr and inversely decreases at 50 mTorr. The pressure dependence of the Vp , the Te , and the Ne of a CH4 /H2 plasma is the same as that of an Ar plasma up to 40 mTorr. Therefore, these features in a CH4 /H2 plasma up to 40 mTorr can be explained by the same hypotheses of Ar plasma [83, 84], whereas the opposite feature at 50 mTorr should let us assume different phenomena in a CH4 /H2 plasma. Figure 13a–e shows the EEDFs of a CH4 /H2 plasma as a function of pressure. The shape of the EEDF at 50 mTorr corresponding to Figure 13a is different from that of an Ar plasma at the same pressure. There is a hump at ∼6 eV. The hump gradually disappears with decreasing pressure. The shape of the EEDF at 30 mTorr in Figure 13c is almost a straight line, which means a Maxwellian distribution. The shape of the EEDF at 20 mTorr in Figure 13d deviates from a straight line and comes close to a Druyveysten distribution. It still keeps a Druyveysten distribution at 10 mTorr in Figure 13e. Accordingly, the transition from a Maxwellian distribution to a Druyveysten distribution occurs at 20 mTorr in a CH4 /H2 plasma. Turner and Hopkins [90] previously reported an unusual structure of the EEDF. They found a dip at ∼4 eV in the EEDF of a N2 plasma. They interpreted the dip as the electric absorption of a N2 molecule corresponding to the resonant peak of the vibrational excitation cross section.

Figure 12. Variation of the plasma parameters of a CH4 /H2 plasma with pressure. (a) Plasma potential. (b) Electron temperature. (c) Electron density. Reprinted with permission from [88], K. Okada et al., J. Vac. Sci. Technol., A 17, 721 (1999). © 1999, American Institute of Physics.

699

Figure 13. Electron energy distribution functions of a CH4 /H2 plasma as a function of pressure. (a) 50 mTorr. (b) 40 mTorr. (c) 30 mTorr. (d) 20 mTorr. (e) 10 mTorr. Reprinted with permission from [88], K. Okada et al., J. Vac. Sci. Technol., A 17, 721 (1999). © 1999, American Institute of Physics.

Cali et al. [86] reported a similar dip at ∼6 eV in the EEDF of a CH4 /H2 plasma. In the CH4 vibrational excitation cross sections, v (1, 3) and v (2, 4) have a peak at ∼6 eV [91]. The dip can be attributed to the electric absorption of a CH4

Figure 14. Variation of the plasma parameters of a CH4 /CO/H2 plasma with [CO] content. (a) Plasma potential. (b) Electron temperature. (c) Electron density.

700

Figure 15. Electron energy distribution functions of a CH4 /CO/H2 plasma as a function of [CO] content. (a) [CO] = 0 sccm. (b) [CO] = 1.0 sccm. (c) [CO] = 5.0 sccm.

molecule analogous to the vibrational absorption of ∼4 eV electrons of a N2 molecule. Although the unusual structure in the EEDF of a CH4 /H2 plasma at 50 mTorr is not a clear dip but a hump, it is presumably equivalent to the energy corresponding to the resonant peak of the vibrational excitation cross section of a CH4 molecule. Since the electron mean free path becomes longer with decreasing pressure and the EEDF is expected to be relatively insensitive to the electron impact cross sections [90], the hump gradually disappears with decreasing pressure. Thus the hump is considered to bring about the opposite feature of the Vp , the Te , and the Ne of a CH4 /H2 plasma at 50 mTorr. Figure 14a–c shows the variation in the Vp , the Te , and the Ne of a CH4 /CO/H2 plasma with [CO]. While the Vp and Te increase with increasing [CO], the Ne decreases with increasing [CO]. Figure 15a–c shows the EEDFs of a CH4 /CO/H2 plasma as a function of [CO]. The hump at ∼6 eV still appears with the addition of CO, and the spectrum shape retains almost the same feature. The inclinations below 6 eV become slow with increasing [CO], which means that the Te increases with increasing [CO]. It is consistent with the result of Figure 14b. The electron impact dissociation of CO produces O− (CO + e → C + O− ) in a CH4 /CO/H2 plasma. It is therefore assumed that the increase in negative ions reduces the Ne relatively to satisfy the charge balance in a plasma, and that the decrease in Ne leads to the increase in Vp and Te .

GLOSSARY Electron energy distribution function The distribution function of electrons in a plasma. That of a low-pressure radiofrequency plasma generally consists of two Maxwellian distributions, that is, fast and slow electrons.

Plasma CVD of Nanocrystalline Diamond

Electron energy loss spectroscopy An analytical technique used to characterize the chemistry, bonding, and electronic structure of thin samples of materials. It is normally performed in a transmission electron microscope. The inelastically scattered electron beams are spectroscopically analyzed to give the energy spectrum of electrons after the interaction. Inductively coupled plasma Plasmas generated by application of radiofrequency power to a nonresonant inductive coil and maintained by an inductive electromagnetic field. Lowpressure ICP is a high-density plasma source. Langmuir probe One of the most useful tools for diagnosing a plasma, in which a metal probe is inserted in a discharge and is biased positively or negatively to draw electron or ion current. The plasma potential, electron temperature, electron density, and electron energy distribution function are obtained from the current-voltage characteristics. Nanocrystalline diamond Diamond crystallites or films in which the crystal grains range between several nm and several hundred nanometers in diameter. Plasma CVD Plasma chemical vapor deposition. Technique for synthesizing materials in which chemical components in vapor phase excited by plasma react to form a solid film at some surface. Raman spectroscopy A nondestructive method for the study of the vibrational band structure of materials, which has been extensively used for the characterization of diamond, graphite, and diamond-like carbon. Raman spectroscopy is so far the most popular technique for identifying sp3 bonding in diamond and sp2 bonding in graphite and diamond-like carbon.

ACKNOWLEDGMENTS The author thanks Drs. S. Komatsu, H. Kanda, K. Kimoto, T. Aizawa, R. Souda, and S. Matsumoto of Advanced Materials Laboratory/National Institute for Materials Science (AML/NIMS) for their cooperation in experiments and the fruitful discussion of the results.

REFERENCES 1. J. Robertson, Prog. Solid State Chem. 21, 199 (1991). 2. M. P. Siegal, W. I. Milne, and J. E. Jaskie, Eds., Mater. Res. Soc. Symp. Proc. 498 (1998). 3. D. M. Gruen, Annu. Rev. Mater. Sci. 29, 211 (1999). 4. J. E. Butler, D. S. Y. Hsu, B. H. Houston, X. Lui, J. Vignola, T. Feygelson, J. Wang, and C. T.-C. Nyguen, “8th International Conference on New Diamond Science and Technology,” Melbourne, 2002, p. 116. 5. J. A. Carlisle, J. Birrell, J. E. Gerbi, O. Auciello, J. M. Gibson, and D. M. Gruen, “8th International Conference on New Diamond Science and Technology,” Melbourne, 2002, p. 129. 6. G. M. Swain, A. B. Anderson, and J. C. Angus, MRS Bull. 56 (1998). 7. M. A. Lieberman and A. J. Lichtenberg, “Principles of Plasma Discharges and Materials Processing.” Wiley, New York, 1994. 8. J. Hopwood, Plasma Sources Sci. Technol. 1, 109 (1992). 9. J. Amorim, H. S. Maciel, and J. P. Sudano, J. Vac. Sci. Technol. B9, 362 (1991). 10. M. Kamo, Y. Sato, S. Matsumoto, and N. Setaka, J. Cryst. Growth 62, 642 (1983).

Plasma CVD of Nanocrystalline Diamond 11. R. Erz, W. Dotter, D. Jung, and H. Ehrhardt, Diamond Relat. Mater. 2, 449 (1993). 12. J. Lee, B. Hong, R. Messier, and R. W. Collins, Appl. Phys. Lett. 69, 1716 (1996). 13. T. Sharda, M. M. Rahaman, Y. Nukaya, T. Soga, T. Jimbo, and M. Umeno, Diamond Relat. Mater. 10, 561 (2001). 14. D. M. Gruen, S. Lui, A. R. Krauss, J. Luo, and X. Pan, Appl. Phys. Lett. 64, 1502 (1994). 15. D. Zhou, T. G. McCauley, L. C. Qin, A. R. Krauss, and D. M. Gruen, J. Appl. Phys. 83, 540 (1998). 16. D. Zhou, D. M. Gruen, L. C. Qin, T. G. McCauley, and A. R. Krauss, J. Appl. Phys. 84, 1981 (1998). 17. A. N. Goyette, J. E. Lawler, L. W. Anderson, D. M. Gruen, T. G. McCauley, D. Zhou, and A. R. Krauss, J. Phys. D: Appl. Phys. 31, 1975 (1998). 18. P. C. Redfern, D. A. Horner, L. A. Curtiss, and D. M. Gruen, J. Phys. Chem. 100, 11654 (1996). 19. V. I. Konov, A. A. Smolin, V. G. Ralchenko, S. M. Pimenov, and E. D. Obraztsova, Diamond Relat. Mater. 4, 1073 (1995). 20. A. V. Khomich, V. I. Polyakov, P. I. Perov, V. P. Varnin, and I. G. Terenmetskaya, Mater. Res. Soc. Symp. Proc. 423, 723 (1996). 21. S. N. Kundu, M. Basu, A. B. Maity, S. Chaudhuri, and A. K. Pal, Mater. Lett. 31, 303 (1997). 22. D. V. Fedoseev, V. L. Bukhovets, I. G. Varshavskaya, and Y. N. Tolmachev, Russ. J. Phys. Chem. 70, 1594 (1996). 23. G. Amaratunga, A. Putnis, K. Clay, and W. Milne, Appl. Phys. Lett. 55, 634 (1989). 24. G. A. J. Amaratunga, S. R. P. Silva, and D. A. McKenzie, J. Appl. Phys. 70, 5374 (1991). 25. P. Badziag, W. S. Verwoerd, W. P. Ellis, and N. R. Greiner, Nature 343, 244 (1990). 26. A. S. Barnard, S. P. Russo, and I. K. Snook, Diamond Relat. Mater., in press. 27. J. Wei, H. Kawarada, J. Suzuki, and A. Hiraki, J. Cryst. Growth 99, 1201 (1990). 28. C. R. Eddy, D. L. Youchison, B. D. Sartwell, and K. S. Grabowski, J. Mater. Res. 7, 3255 (1992). 29. M. Zarrabian, N. Fourches-Coulon, G. Turban, C. Marhic, and M. Lancin, Appl. Phys. Lett. 70, 2535 (1997). 30. S. P. Bozeman, D. A. Tucker, B. R. Stoner, J. T. Glass, and W. M. Hooke, Appl. Phys. Lett. 66, 3579 (1995). 31. K. Okada, S. Komatsu, T. Ishigaki, and S. Matsumoto, Mater. Res. Soc. Symp. Proc. 363, 157 (1995). 32. K. Okada, S. Komatsu, T. Ishigaki, and S. Matsumoto, “Proceedings of the 12th International Symposium on Plasma Chemistry,” Minnesota, 1995, p. 2261. 33. K. Okada, S. Komatsu, and S. Matsumoto, J. Mater. Res. 14, 578 (1999). 34. K. Teii and T. Yoshida, J. Appl. Phys. 85, 1864 (1999). 35. Y. Hirose and Y. Terasawa, Jpn. J. Appl. Phys. 25, L519 (1986). 36. T. Kawato and K. Kondo, Jpn. J. Appl. Phys. 26, 1429 (1987). 37. J. A. Mucha, D. L. Flamm, and D. E. Ibbotson, J. Appl. Phys. 65, 3448 (1989). 38. M. Frenklach and H. Wang, Phys. Rev. B 43, 1520 (1991). 39. K. Teii, H. Ito, M. Hori, T. Takeo, and T. Goto, J. Appl. Phys. 87, 4572 (2000). 40. M. von Laue, Z. Kristallogr. 64, 115 (1926). 41. S. A. Solin and A. K. Ramdas, Phys. Rev. B 1, 1687 (1970). 42. R. J. Nemanich, J. T. Glass, G. Lucovsky, and R. E. Shroder, J. Vac. Sci. Technol., A 6, 1783 (1988). 43. J. Wagner, M. Ramsteiner, Ch. Wild, and P. Koidl, Phys. Rev. B 40, 1817 (1989). 44. R. E. Shroder and R. J. Nemanich, Phys. Rev. B 41, 3738 (1990). 45. F. Tuinstra and J. L. Koenig, J. Chem. Phys. 53, 1126 (1970). 46. A. C. Ferrari and J. Robertson, Phys. Rev. B 61, 14095 (2000). 47. M. Nakamizo, R. Kammereck, and P. L. Walker, Jr., Carbon 12, 259 (1974).

701 48. M. Yoshikawa, G. Katagiri, H. Ishida, A. Ishitani, and T. Akamatsu, Appl. Phys. Lett. 52, 1639 (1988); J. Appl. Phys. 64, 6464 (1988). 49. M. Yoshikawa, N. Nagai, M. Matsuki, H. Fukuda, G. Katagiri, H. Ishida, and A. Ishitani, Phys. Rev. B 46, 7169 (1992). 50. V. I. Merkulov, J. S. Lannin, C. H. Munro, S. A. Asher, V. S. Veerasamy, and W. I. Milne, Phys. Rev. Lett. 78, 4869 (1997). 51. K. W. R. Gilkes, H. S. Sands, D. N. Batchelder, J. Robertson, and W. I. Milne, Appl. Phys. Lett. 70, 1980 (1997). 52. K. W. R. Gilkes, S. Prawer, J. Robertson, and H. S. Sands, Mater. Res. Soc. Symp. Proc. 498, 49 (1998). 53. D. R. Tallant, T. A. Friedmann, N. A. Missert, M. P. Siegal, and J. P. Sullivan, Mater. Res. Soc. Symp. Proc. 498, 37 (1998). 54. K. Okada, H. Kanda, S. Komatsu, and S. Matsumoto, Mater. Res. Soc. Symp. Proc. 593, 459 (2000). 55. Z. Sun, J. R. Shi, B. K. Tay, and S. P. Lau, Diamond Relat. Mater. 9, 1979 (2000). 56. S. Prawer, K. W. Nugent, D. N. Jamieson, J. O. Orwa, L. A. Bursill, and J. L. Peng, Chem. Phys. Lett. 332, 93 (2000). 57. K. Okada, H. Kanda, S. Komatsu, and S. Matsumoto, J. Appl. Phys. 88, 1674 (2000). 58. S. Prawer and K. W. Nugent, in “Amorphous Carbon: State of the Art” (S. R. P. Silva, J. Robertson, and G. A. Amaratunga, Eds.), p. 199. World Scientific, Singapore, 1998. 59. A. C. Ferrari and J. Robertson, Phys. Rev. B 63, 121405 (2001). 60. H. Ibach and D. L. Mills, “Electron Energy Loss Spectroscopy and Surface Vibrations.” Academic Press, London, 1982. 61. B. J. Waclawski, D. T. Pierce, N. Swanson, and R. J. Celotta, J. Vac. Sci. Technol. 21, 368 (1982). 62. C. Oshima, R. Souda, M. Aono, and Y. Ishizawa, Rev. Sci. Instrum. 56, 227 (1985). 63. C. Oshima, T. Aizawa, R. Souda, and Y. Ishizawa, Solid State Commun. 65, 1601 (1988). 64. T. Aizawa, R. Souda, S. Otani, Y. Ishizawa, and C. Oshima, Phys. Rev. Lett. 64, 768 (1989). 65. T. Aizawa, T. Ando, K. Yamamoto, M. Kamo, and Y. Sato, Diamond Relat. Mater. 4, 600 (1995). 66. K. Okada, T. Aizawa, R. Souda, S. Komatsu, and S. Matsumoto, Diamond Relat. Mater. 10, 1991 (2001). 67. C. Z. Wang and K. M. Ho, Phys. Rev. Lett. 71, 1184 (1993). 68. D. A. Drabold, P. A. Fedders, and P. Stumm, Phys. Rev. B 49, 16415 (1994). 69. R. Brydson, “Electron Energy Loss Spectroscopy.” Springer-Verlag, New York, 2001. 70. R. F. Egerton and M. J. Whelan, J. Electron Spectrosc. Relat. Phenom. 3, 232 (1974). 71. D. A. Muller, Y. Tzou, R. Raj, and J. Silcox, Nature 366, 725 (1993). 72. J. Bruley, D. B. Williams, J. J. Cuomo, and D. P. Pappas, J. Microsc. 180, 22 (1995). 73. L. A. Bursill, J. L. Peng, and S. Prawer, Philos. Mag. A 76, 769 (1997). 74. R. P. Fehlhaber and L. A. Bursill, Philos. Mag. B 79, 477 (1999); Phys. Rev. B 60, 14147 (1999); Phys. Rev. B 62, 17094 (2000). 75. J. L. Peng, R. P. Fehlhaber, L. A. Bursill, and D. G. McCulloch, J. Appl. Phys. 89, 6204 (2001). 76. S. Prawer, J. L. Peng, J. O. Orwa, J. C. McCallum, D. N. Jamieson, and L. A. Bursill, Phys. Rev. B 62, 16360 (2000). 77. K. Okada, K. Kimoto, S. Komatsu, and S. Matsumoto, “8th International Conference on New Diamond Science and Technology,” Melbourne, 2002, p. 187. 78. R. Erz, W. Dotter, D. Jung, and H. Ehrhardt, Diamond Relat. Mater. 2, 449 (1993). 79. P. J. Fallon and L. M. Brown, Diamond Relat. Mater. 2, 1004 (1993). 80. P. Keblinski, D. Wolf, S. R. Phillpot, and H. Gleiter, J. Mater. Res. 13, 2077 (1998). 81. J. Hopwood, C. R. Guarnieri, S. J. Whitehair, and J. J. Cuomo, J. Vac. Sci. Technol., A 11, 152 (1993). 82. L. J. Mahoney, A. E. Wendt, E. Barrios, C. J. Richards, and J. L. Shohet, J. Appl. Phys. 76, 2041 (1994).

702 83. P. A. Miller, G. A. Hebner, K. E. Greenberg, P. D. Pochan, and B. P. Aragon, J. Res. Natl. Inst. Stand. Technol. 100, 427 (1995). 84. A. Schwabedissen, E. C. Benck, and J. R. Roberts, Phys. Rev. E 55, 3450 (1997). 85. M. B. Hopkins, J. Res. Natl. Inst. Stand. Technol. 100, 415 (1995). 86. F. A. Cali, P. A. F. Herbert, and W. M. Kelly, J. Vac. Sci. Technol., A 13, 2920 (1995).

Plasma CVD of Nanocrystalline Diamond 87. V. A. Godyak and R. B. Piejak, Phys. Rev. Lett. 65, 996 (1990). 88. K. Okada, S. Komatsu, and S. Matsumoto, J. Vac. Sci. Technol., A 17, 721 (1999). 89. L. Schott, in “Plasma Diagnostics” (W. Lochte-Holtgreven, Ed.). North-Holland, Amsterdam, 1968. 90. M. M. Turner and M. B. Hopkins, Phys. Rev. Lett. 69, 3511 (1992). 91. W. L. Morgan, Plasma Chem. Plasma Process. 12, 477 (1992).