Growth Process.

Contrary to the growth process for MCD and NCD films described above, the UNCD films, discussed in this chapter, were first produced by the MPCVD process, but in recent years the HFCVD process was also developed to produce high-quality UNCD films, as discussed in this chapter. The original MPCVD process used to grow UNCD films was based on gas mixture flowing into an air evacuated chamber, and involved – and still involves – a novel Ar-rich chemistry (Ar [99%]/CH4 [1%]) with no H₂ gas flow in the system [Reference Jiao, Sumant and Kirk20–Reference Shenderova and Gruen23, Reference Krauss, Ding and Auciello50] and in more recent years including also extremely small H₂ gas flow (≤1%). In both cases, it was determined, using in situ/real-time optical emission spectroscopy imaging of the MPCVD plasma, generated with the Ar/CH₄ chemistry, that carbon dimers (C₂) are produced in the plasma (inducing a green color to the plasma from the light emission from the excited C₂ dimers) generated by the MPCVD process, from methane decomposition, via reactions (1.1) and (1.2):

$2{\mathrm{CH}}_4\to {\mathrm{C}}_2{\mathrm{H}}_2+3{\mathrm{H}}_2,$(1.1)

${\mathrm{C}}_2{\mathrm{H}}_2\to {\mathrm{C}}_2+{\mathrm{H}}_2.$(1.2)

Although the Ar-rich/CH₄ plasma induces the formation of a complex mixture of carbon (C₂ dimers) and hydrocarbon molecules (CH_x, with x = 1, 2, and 3), the C₂ dimers have been proposed and demonstrated to play a critical role in the UNCD nucleation and growth process (see review in [Reference Auciello and Sumant22]). Calculations predict that C₂ dimers have low activation energy (~6 kcal/mol) for insertion into the surface of substrates to induce nucleation and subsequent growth of UNCD films, thus establishing the unique growth process of the UNCD films. However, recent modeling indicated that while the C₂ population in the plasma is high, the population near the surface may be lower, and other hydrocarbon radicals (e.g., C₂H₂) may also contribute to the growth of UNCD films [Reference May, Harvey, Smith and Mankelevich65, Reference May, Allan, Ashfold, Richley and Mankelevich66]. However, the model [Reference May, Allan, Ashfold, Richley and Mankelevich66] is related to the growth of diamond films produced by MPCVD using a mixture of Ar∕CH₄∕H₂ gases, which, as shown in Figure 1.2, does not produce the unique UNCD structure (with 3–5 nm grains) grown by MPCVD with the Ar/CH₄ chemistry [Reference Naguib, Birrell, Elam, Carlisle and Auciello21–Reference Shenderova and Gruen23]. In addition, the model could not fully explain the low-temperature growth of UNCD films as demonstrated for UNCD film growth at ~400 °C [Reference Xiao, Birrell, Gerbi, Auciello and Carlisle39, Reference Carlisle, Gruen, Auciello and Xiao40]. Clearly, more experimental and modeling studies are needed. Regardless of the mechanism, the distinctive characteristic of the UNCD film growth process is that the plasma contains very small quantities of hydrogen, which arise mainly from the thermal decomposition of methane to acetylene in the plasma (about 1.5%) and eventual addition of an extremely small amount of H₂ (≤1%). The MPCVD process is implemented in small research and, most importantly, industrial-type systems, like the ones operating in Auciello’s group laboratory at UTD and the company OBI-México (Figure 1.3a), which grows UNCD, NCD, and MCD films on up to 200 mm diameter substrates (Figure 1.3b) with outstanding uniformity in thickness (≤1%) and nanostructure.

Figure 1.3

(a) Industrial MPCVD system (IPLAS-Germany); (b) Si wafer (200 mm diameter) coated with outstanding uniform UNCD film.

A critical outcome of the nucleation and growth processes for UNCD films is that they are the only diamond films that have been demonstrated to grow at temperatures as low as 350–400 °C [Reference Xiao, Birrell, Gerbi, Auciello and Carlisle39, Reference Carlisle, Gruen, Auciello and Xiao40], as determined not only by in situ substrate temperature measurements during growth [Reference Auciello and Sumant22, Reference Xiao, Birrell, Gerbi, Auciello and Carlisle39, Reference Carlisle, Gruen, Auciello and Xiao40], but most importantly by the demonstration that CMOS devices exhibit practically the same performance before and after growing UNCD film on them [Reference Xiao, Wang and Carlisle28, Reference Sumant, Auciello and Yuan67]. Similar demonstration of low-temperature NCD films integration with CMOS devices has not been published in the open literature yet. The demonstrated integration of UNCD films with CMOS devices is paving the way for the integration with CMOS for the development of monolithically integrated UNCD-MEMS/NEMS/CMOS devices, as recently demonstrated [Reference Sumant, Auciello and Yuan67], and for encapsulation of Si microchips implantable inside the eye [Reference Xiao, Wang and Carlisle28] or other parts of the human body.

Structure Characterization.

The nucleation and growth process described above produce the UNCD films with the name based on the equiaxed 3–5 nm grains dimension and 0.4 nm wide grain boundaries for undoped insulating UNCD films grown both at 800 °C (Figure 1.4a) and 400 °C (Figure 1.4b), which exhibit extremely smooth as-grown surfaces (~3–5 nm) (Figure 1.4d) when using optimized seeding techniques.

Figure 1.4

High-resolution transmission electron microscopy (HRTEM) images of UNCD films grown by MPCVD at 800 °C (a) and 400 °C (b), showing the unique grain structure (3–5 nm dimensions) of UNCD produced at both temperatures, confirmed by statistical grain size measurements (c). (d) Cross-section SEM images of an Si tip coated with UNCD films grown by the MPCVD process described above, with thicknesses from 0.1 µm to 2.4 µm, showing that the surface roughness remains in the range 3–5 nm independently of the film thickness.

(reproduced from J. Appl. Phys., vol.88 (11), p. 2958, 2001 (Fig. 2) in [50] with permission from AIP Publisher)

Chemical Characterization.

The UNCD films grown by MPCVD were and are characterized by Raman analysis, to provide information on the chemical bonds of the C atoms in the grains and in the grain boundaries. Raman spectroscopy is a nondestructive chemical analysis technique that provides detailed information about the chemical bonds of atoms in a material, distinguishing the atomic bonds in a crystal-type structure such as crystalline diamond and other structures such as the noncrystalline structure of grain boundaries in polycrystalline diamond films. Raman analysis is based on the interaction of light with the chemical bonds of molecules within a material, such that light induces excitation of electrons in the electronic cloud around the atomic nuclei in molecules, producing displacement of the negatively charged electrons with respect to the positively charged nucleus, inducing molecular vibration and causing a “change in polarizability” of the molecule, such that the induced dipole emits or scatters light at the optical frequency of the incident light wave, and the scattered light detected in a detector provides the information on the chemical bonds of the atoms (see a recent review in [Reference Weber and Merlin69]).

Raman spectroscopy is the most widely used technique to confirm the sp³ chemical bonds of C atoms in the diamond crystalline lattice and sp² C atoms bonds in the grain boundaries of the films. It is a nondestructive technique, often used to determine the nanostructure of carbon thin films due to its ability to discriminate between the presence of sp²- and sp³-bonded carbon, as well as the local bonding environment of the carbon itself in some cases. Difficulty arises when Raman spectroscopy is used to examine the structure of carbon films with a mixture of sp²- and sp³-bonded carbon in a number of different bonding configurations that possess different short- and long-range order, as is found in nanocrystalline diamond and amorphous carbon films. When a laser with a wavelength in the visible region is used (e.g., 632 nm, 535 nm), the energy of the incident photons is much lower than the energy of the band gap for sp³-bonded carbon, resulting in a much larger Raman scattering cross-section for sp²-bonded carbon than for sp³-bonded carbon due to the well-known resonant Raman effect [Reference Weber and Merlin69]. The spectra observed are thus completely dominated by Raman scattering from the sp²-bonded C atoms, even when a significant amount of the C in the sample, like in UNCD films, is sp³-C bonded in the small diamond crystalline grains, but with a large network of grain boundaries with sp²-C atom bonds (Figure 1.5a,b, with a wide peak around ~1560 cm^–1) [Reference Birrell, Gerbi and Auciello70]. For polycrystalline diamond films grown with less than 10% H₂ in the plasma (Figure 1.5a), the fingerprint peak of diamond at 1332 cm^–1 practically disappears, and the visible Raman signal is dominated by the sp²-bonded C atoms at the grain boundaries. Several distinct broad peaks can be seen for the films: at 1140, 1330, 1450, and 1560 cm^–1. The peak at 1560 cm^–1 corresponds to the G mode Raman peak, which arises from the in-plane stretching modes of the sp²-bonded C atoms at the grain boundaries of UNCD [Reference Birrell, Gerbi and Auciello70]. On the other hand, the 1330 cm^–1 broaden peak is attributed to D band-type stretching of C atom bonds in the grain boundaries. In disordered sp²-bonded C atoms, D band stretching arises from the breathing modes in small aromatic clusters. The argument for the assignment of the broad peak at 1330 cm^–1 to the sp²-bonded C atoms’ D band is based on the fact that the peak is absent from the UV Raman spectra (Figure 1.5b). The D peak is believed to arise from a double resonant Raman process (incident photon wave vector k = 0.5q), where the intensity is highest for low-energy excitation and decreases strongly with increasing excitation energy. In order to see the diamond peak at 1332 cm^–1, even for UNCD films, UV laser needs to be used for the Raman analysis (Figure 1.5b). The peaks observed in the UV Raman spectra at 1550 cm^–1 and 1560 cm^–1 are mainly due to sp²-C atoms bonding in the grain boundaries [Reference Birrell, Gerbi and Auciello70].

Figure 1.5

(a) Raman spectra obtained with a laser beam wavelength in the visible range, showing the diamond peak at 1332 cm^–1 only for NCD films grown with H₂ flow ≥5%. (b) Raman spectra obtained using a laser beam with UV wavelength, for the same polycrystalline diamond films analyzed with the visible laser (a), such that the UV laser enables seeing the diamond peak at 1332 cm^–1, even for the UNCD film grown with H₂ flow of 0%.

(reprinted from Diam. Relat. Mater., vol. 14, p. 86, 2005 (Fig. 4) in [70] with permission from Elsevier Publisher)

The confirmation that the dominant sp²-C atoms bonds are not from an impurity graphite phase (i.e., parallel planes formed by hexagonal unit lattice with C atoms on the corner) is provided by near-edge X-ray absorption fine structure spectroscopy (NEXAFS) done on UNCD, MCD, and a crystal diamond gem. Core-level photo absorption has been used to characterize the empty electronic states of a wide variety of materials [Reference Stohr71]. Specifically, the near-edge electronic energy band region of the photo-absorption process has been used to determine the relative quantity of sp²- or sp³-C atoms bonding in several powders and thin films [Reference Stohr71]. The technique is known by several acronyms, including NEXAFS and XANES (X-ray absorption near-edge spectroscopy). Its sensitivity to the local atomic bond order in a material arises from the dipole-like electronic transitions from core atoms’ electronic states, which have well-defined orbital angular momenta, into empty electronic (e.g., antibonding) states. The symmetry of the final state can be determined, and thus the difference between sp² (n-like) or sp³ (a-like) bonding can be readily observed in covalent, low-z materials such as diamond.

The NEXAFS spectra show the characteristic spectrum of crystalline diamond for UNCD, MCD, and a diamond gem (Figure 1.6a), which is completely different from the NEXAFS spectrum done on a pure graphite film (Figure 1.6a) [Reference Zuiker, Krauss and Gruen72].

Figure 1.6

(a) NEXAFS spectra from UNCD, MCD, and crystal diamond, showing that they are all identical and do not have graphite impurity, demonstrated by no graphite peak as shown in the graphite NEXAFS spectrum reference. (b) NEXAFS spectra from several UNCD films grown at different temperatures in the range 250–800 °C, showing that all UNCD films have the same chemistry and structure.

(reprinted from Mat. Res. Soc. Proc., vol. 437, p. 211, 1996 (Fig. 2) in [72] with permission from Cambridge Publisher)

In addition to the NEXAS analysis, electron diffraction from HRTEM done on UNCD films grown at 800 °C and 400 °C (see Figure 1.4a,b) confirm that the UNCD films are made of crystalline diamond nanograins, and no impurity graphite phase is present. Relevant information given by the NEXAFS analysis is that UNCD films grown at different temperatures in the range 250–800 °C exhibit the same chemistry and nanostructure (Figure 1.6b). It is important to note that Figure 1.6b shows that UNCD films may be grown by MPCVD at ~250 °C, which may enable coating polymers used in biomedical devices such as glaucoma valves used to drain humor of the eye from people suffering from glaucoma, to keep eye pressure stable, or in polymer soft contact lenses. Both of these case require hydrophobic (liquid-rejecting) surfaces to minimize or eliminate protein and other biomolecule adsorption on normally hydrophilic (liquid-adsorbing) polymer surfaces, to avoid undesirable effects such as clogging of the humor’s eye-draining tubes in glaucoma valves, or the need for frequent cleaning of polymer contact lenses’ surfaces due protein adsorption (see review in [Reference Auciello59]). On the other hand, it has been demonstrated that hydrophilic UNCD film surface provides an excellent scaffold for embryonic cell growth and differentiation for developmental biology [Reference Shi, Jin, Chen and Auciello58]. In relation to hydrophilic vs. super-hydrophobic UNCD films’ surfaces, recent R&D has demonstrated that the surface of UNCD films can be tailored from super-hydrophilic surfaces, for scaffolds for developmental biology, to super-hydrophobic surfaces for medical devices, as just described.

To finish this section reviewing the growth of UNCD films using the conventional wet chemical seeding plus MPCVD processes, it is relevant to give a brief review of recent research [Reference Tirado, Alcantar-Peña, de Obaldia, Garcia and Auciello73], focused on exploring, in a systematic series of studies, the combined effect of total gas pressure (20–80 mbar), precursor gas chemistry (Ar/CH₄/H₂), and microwave power on the grain size, atoms’ chemical bonding, and growth rate of polycrystalline diamond films grown by MPCVD [Reference Tirado, Alcantar-Peña, de Obaldia, Garcia and Auciello73], with a particular focus on determining the conditions that can produce UNCD films even with the inclusion of H₂ gas flow in the gas chemistry. Figure 1.7 shows a series of SEM images and Raman analysis spectra which provide very revealing information about the mechanism for growing UNCD to MCD diamond films using a tailored combination of gas chemistry and flows, total pressure and power in the MPCVD process. For a gas chemistry of H₂ (98%)/CH₄ (2%), the film grown at 20 mbar total pressure for 2 h was only ~60 nm thick (Figure 1.7a – right) and the surface morphology was characteristic of NCD film structure, with grains ~14 nm, as calculated from the Debye–Scherrer equation using parameters from XRD measurements (Figure 1.7a – left), mainly due to the fact that at low pressure the amount of C-based species is much lower than for the higher pressures, resulting in a film with initial nucleation nanoscale grains. SEM imaging of films grown at 40 mbar pressure (not shown here – see figure 3b in [Reference Tirado, Alcantar-Peña, de Obaldia, Garcia and Auciello73]), also for 2 h, showed a thickness of ~850 nm and grain size also of ~14 nm. Figure 1.7b,c shows that films grown at 60 mbar and 80 mbar pressure, with the same H₂ (98%)/CH₄ (2%) chemistry, for 2 h, exhibit the highest thickness (~2500 nm) and grain size of ~230 nm, and 1500 nm thickness and ~230 nm grain size, respectively. The Raman spectra of the films grown at 20 and 40 mbar show the typical low end of NCD (~14 nm) structure, close to UNCD (3–7 nm [Reference Auciello and Sumant22]) (Figure 1.7d), while the Raman spectra of the films grown at 60 and 80 mbar (Figure 1.7d) show a small, sharp peak at 1332 cm^–1, indicative of the more noticeable sp³-type C atomic bond from diamond, indicative of grain sizes in the hundreds of nanometer range, as described above (see figures 3 and 4 in [Reference Tirado, Alcantar-Peña, de Obaldia, Garcia and Auciello73]). XRD analysis of the polycrystalline diamond films (Figure 2b in [Reference Tirado, Alcantar-Peña, de Obaldia, Garcia and Auciello73]) reveal only peaks corresponding to diamond, indicating that no graphite was formed in any of the grown films. The intensity of the peaks increased with increasing pressure, indicative of thicker films with larger grains due to grain growth with film thickness.

The SEM images of films grown with different chemistries, as indicated in Figures 1.7e–g and the corresponding Raman spectra (Figure 1.7h) indicate that films grown with H₂ concentrations of 0–12% (Ar 98–86%) show the typical UNCD/small NCD structures with grain sizes in the range 2–10 nm [Reference Auciello and Sumant22, Reference Shenderova and Gruen23]. On the other hand, in the Raman spectra of the films grown with H₂ concentrations of 73–98% (Ar 25–0%), in addition to the bands discussed above, a peak appears at 1332 cm^–1, attributed to sp³-C bonds of diamond [Reference Fuentes-Fernandez, Alcantar-Peña and Lee74, Reference Filik75], indicating that the grain size increased significantly in the diamond films by increasing the H₂ concentration during growth. XRD analysis shows only peaks attributed to diamond.

Figure 1.7

SEM top view and cross-section images of MPCVD-processed polycrystalline diamond (PCD) films grown at (a) 20, (b) 60, and (c) 80 mbar using a gas mixture composition of H₂ (98%)/2% CH₄ and microwave power of 1900 W. (d) Raman spectra corresponding to the PCD films, for which SEM images are shown in (a)–(c). Top view and cross-section SEM images of PCD films grown with different Ar/CH₄/H₂ compositions as (e) 0%/2%/98%, (f) 86%/2%/12%, (g) 98%/2%/0%, with total pressure of 60 mbar and 1900 W of microwave power. (h) Raman spectra corresponding to the PCD films, for which SEM images are shown in (e)–(g). Top view and cross-section SEM images of PCD films grown at (i) 1900 W, (j) 2200 W, and (k) 2500 W using gas composition of H₂ (98%)/CH₄ (2%) at a total pressure of 60 mbar. (l) Raman spectra corresponding to the PCD films, for which SEM images are shown in (i)–(k).

(reprinted from IEEE 7th International Engineering, Sciences and Technology Conference, p. 85, 2019 (Figs. 1.7 (a, b, c) – reprint of Figs. 3 (a, c, d) in [73], respectively; Figs. 1.7 (e, f, g) – reprint of Figs. 5 (a, b, d) in [73], respectively; Figs. 1.7 (I, j, k) – reprint of 2nd Figs. 5 (a, b, d) in [73], respectively; Figs. 1.7 (d), 1.7 (h), 1.7 (l) – reprint of Figs. 2 (a), 4 (a), 6 (a), respectively, with permission from IEEE Publisher)

The data shown in Figure 1.7 indicate that films with structure close to UNCD can be grown by MPCVD, even when using H₂ gas flow under specific percentages in combination with Ar and CH₄ gases, but only for very thin films, correlated with the initial nucleation of NCD to MCD films, while for films grown with substantially higher thickness the MCD structure is developed. This statement is confirmed by diamond films grown with H₂ (98%)/CH₄ (2%) at 60 mbar and microwave power of 2500 W, but increasing the growth time to 6 h, since it has been reported in the literature that the film grain size increases with growth when using the HFCVD technique, which demonstrated MCD films [Reference Fuentes-Fernandez, Alcantar-Peña and Lee74]. Figure 1.8a shows a Raman spectrum of a large grain size MCD film, while Figure 1.8b shows the top surface SEM image of the same MCD film, grown by the MPCVD process for 6 h using a gas mixture of H₂ (98%)/CH₄ (2%) at 60 mbar pressure and 2500 W microwave power. Both figures show clearly an MCD structure.

Figure 1.8

(a) Raman spectrum and (b) SEM top view image of an MCD film grown using the MPCVD process with a gas mixture of H₂ (98%)/CH₄ (2%), 60 mbar pressure, 2500 W microwave power, 6 h growth time [Reference Tirado, Alcantar-Peña, de Obaldia, Garcia and Auciello73].

(reprinted from IEEE 7th International Engineering, Sciences and Technology Conference, p. 85, 2019 (Figs. 8 (a, b) in [73] with permission from IEEE Publisher)

Figure 1.8 confirms the hypothesis that the growth time, using H₂-rich/CH₄ gas chemistry, induces the development of an MCD film structure due to grain growth.

Growth of N-UNCD Films via the MPCVD Plus Chemical Seeding Process

The first attempt at producing N-UNCD films was done by Ding/Krauss/Auciello et al. [Reference Ding, Krauss and Auciello79], developing N-UNCD films for field emission devices [Reference Ding, Krauss and Auciello79]. N atoms grain boundary incorporated UNCD (named N-UNCD) films were grown using the MPCVD BEN-BEG process on high-conductivity Si substrates; this work will be described in the next section.

This section focuses on the R&D performed after the demonstration of the BEN-BEG synthesis of N-UNCD films. The subsequent work focused on using the conventional chemical seeding plus MPCVD growth process. The initial R&D reviewed here is described in detail in [Reference Bhattacharyya, Auciello and Birrell29]. That work showed that when N₂ gas is inserted in the Ar/CH₄ gas mixture, the MPCVD-created plasma contains C₂ dimers and CN radicals due to C–N atoms’ chemical reaction in the gas phase, such that the CN content increases substantially as N₂ gas is added to the Ar/CH₄ gas mixture (see figure 1a in [Reference Bhattacharyya, Auciello and Birrell29]). Small additions of N₂ (1–5%) resulted in nanostructure similar to the UNCD films (3–7 nm grains and ~0.4 nm grain boundaries) (Figure 1.15a,b,d,e), while for N₂ (20%) the films transform in what was originally defined as N-UNCD films (Figure 2a in [Reference Bhattacharyya, Auciello and Birrell29], patent in [Reference Gruen, Krauss, Auciello and Carlisle30]) with grains of 7–16 nm dimensions and grain boundaries with 1–2 nm width (Figure 1.15c,d) [Reference Bhattacharyya, Auciello and Birrell29, Reference Birrell, Carlisle, Auciello, Gruen and Gibson110].

Figure 1.15

HRTEM images of N-UNCD films grown with Ar/CH₄ gas mixture with N₂ (0%) (a) low resolution, (d) high-resolution; N₂ (5%) (b) low resolution, (e) high resolution; N₂ (20%) (c) low resolution, (f) high resolution (this shows the initial characterization of N-UNCD films, revealing the enlargement of grains and grain boundaries with respect to UNCD films revealed in the original first demonstration of MPCVD-grown N-UNCD films).

(reprinted from App. Phys. Lett., vol. 81 (12), p. 2235, 2002 (Fig. 1) in [110] with permission from AIP Publisher)

Raman spectroscopy has been used by many groups as the first easy/fast technique to characterize the atomic chemical bonds in the N-UNCD films, as described for the UNCD films in Section 1.2.1.2. However, as for the case of characterization of the UNCD films, although Raman spectroscopy is sensitive enough to detect the small amounts of nondiamond carbon in largely tetrahedrally C-bonded atoms such as in the UNCD and N-UNCD films, two major drawbacks, already described for the case of UNCD, exist in Raman analysis of N-UNCD films: (1) there is still a large difference in the cross-section for Raman scattering between sp²- and sp³-bonded C atoms [Reference Coffman, Cao and Pianetta111]; and (2) Raman scattering still has a strong dependence on the diamond crystallite size [Reference Nemanich, Glass and Lucovsky112] for N-UNCD, with larger grain size (~10 nm) than UNCD (3–5 nm). This dependence is very difficult to quantify and is especially challenging when measuring amorphous or nanocrystalline carbon films, including UNCD. The Raman spectra, obtained with a visible laser beam (He–Ne laser at a wavelength of 632 nm and spot size of 6 µm), from analysis of N-UNCD films grown with N₂ gas contents from 0 to 20% are shown in Figure 1.16a. The spectra reveal the characteristic D band peak at 1340 cm^–1 and G band peak at 1556 cm^–1, both attributed to sp²-bonded carbon, and a shallow shoulder at 1140 cm^–1, attributed to the nanocrystalline diamond structure [Reference Birrell, Gerbi and Auciello70, Reference Coffman, Cao and Pianetta111, Reference Nemanich, Glass and Lucovsky112] The Raman spectra show an increase of the G band peak relative to the D band peak as the N atom content in the plasma increases, with a slight shift of the peak to a higher wavenumber, indicating a slight change in the bonding configuration of the sp²-bonded carbon at the grain boundaries

However, Raman analysis does not provide all the information needed to understand how the chemical bonding in grain boundaries of N-UNCD films produces the high electrical conductivity, as proposed by molecular dynamic simulations [Reference Zapol, Sternberg, Curtiss, Frauenhein and Gruen113] that indicated that N atoms bonding to C atoms, with sp² bonding available in grain boundaries, plays a critical role in inducing the high electrical conductivity of N-UNCD films. Therefore, small-spot HRTEM-based EELS were performed to obtain information to establish correlations between the bonding structure in the diamond grains and the grain boundaries in a series of films grown with increasing amounts of nitrogen in the plasma [Reference Birrell, Carlisle, Auciello, Gruen and Gibson110]. Figure 1.16b shows a representative EELS spectrum with the HRTEM electron beam focused on a UNCD grain, indicating that it is phase pure diamond for all nitrogen doping levels, as evidenced by the near-edge electronic structure of the peak [Reference Birrell, Carlisle, Auciello, Gruen and Gibson110]. The well-defined absorption edge at 289.5 eV and the presence of the second band gap feature, revealed at ~302 eV, clearly shows that the diamond grains are sp³-bonded carbon with a high degree of short-range ordering [Reference Birrell, Carlisle, Auciello, Gruen and Gibson110]. This effect does not change as nitrogen is added to the plasma, indicating that the electronic structure within the grains does not change. The grain boundary EELS spectra, shown in Figure 1.16c, reveals that the π* peak, at 285 eV, increases slightly when N₂ gas increases from 0% to 20% during N-UNCD film growth, inducing N atom insertion into the N-UNCD films’ grain boundaries. The σ* peak at 289 eV, associated with sp³ bonding in the grains, is unchanged. The relative intensities of the sp² vs. sp³ peaks in the EELS spectra in Figure 1.16c indicate that the amount of sp² bonding increases only slightly within the grain boundaries when N₂ flows, thus N^o atoms and N⁺ ions increase in the MPCVD plasma process [Reference Birrell, Carlisle, Auciello, Gruen and Gibson110].

Figure 1.16

(a) Spectra from Raman analysis of N-UNCD films grown with Ar/CH₄ gas flux with N₂ gas added from 0% to 20% for the five different N-UNCD films. (b) HRTEM-EELS analysis within diamond grains of the same N-UNCD films for which Raman spectra are shown in (a). (c) HRTEM-EELS analysis in the grain boundaries of the same N-UNCD films for which Raman spectra are shown in (a).

(reprinted from App. Phys. Lett., vol. 81 (12), p. 2235, 2002 (Fig. 2) in [110] with permission from AIP Publisher)

Another key analysis on the N-UNCD films was done using the atomic scale resolution technique “atom probe analysis” to determine whether N atoms were inserted in the grain boundaries. In this study, N-UNCD films with the hypothesized N atoms inserted in the N-UNCD films’ grain boundaries, as confirmed later, were grown on atomically sharp tips that were biased with a high positive potential with respect to the ground potential, while picosecond laser pulses, impinging on the specimen’s microtip along the axis in a high-vacuum system, triggered electric field-induced evaporation of ions from the surface of the N-UNCD-coated tip, such that field-evaporated ions were accelerated along diverging electric field lines to project a highly magnified image of the microtip’s surface onto a position-sensitive time-of-flight detector. The time of flight of the ions was used to identify their mass-to-charge-state ratio and hence their chemical identities (see figures 2, 3, and 4 in [Reference Heck, Staderman and Isheim114] for a detailed explanation of the technique). Three-dimensional tomographic atom probe tomography reconstructions of nano-diamonds provided information about the N and C atoms in the grain boundaries of N-UNCD films (see figure 6 in [Reference Heck, Staderman and Isheim114], showing maps of single-atom distributions).

The initial theory [Reference Zapol, Sternberg, Curtiss, Frauenhein and Gruen113] indicated that when nitrogen is incorporated in the N-UNCD or other PCD films’ grain boundaries, new electronic states are produced via electron release when N atoms bond chemically to sp² open C atom bonds in the grain boundaries. It is important to note that this mechanism of electronic conductivity change in N-UNCD films is not due to the conventional N atom substitution of C atoms in the diamond crystal lattice of the grains producing electron donor doping, such as in Si. Complementary nanostructure and chemical characterization of the N-UNCD films, performed by independent groups [Reference Coffman, Cao and Pianetta111, Reference Gicquel, Silva and Hassouni115] to compare with theory, provided valuable information for understanding the UNCD transformation into N-UNCD films and the electrical conductivity.

In addition to the structural and chemical bonding characterization of N-UNCD films, an important characterization is related to measuring the electrical conductivity of the N-UNCD films. In this sense, Figure 1.17a shows SIMS data for the total nitrogen content in the films as a function of the percentage of N₂ gas added to the plasma. Along with these data is a plot of the room-temperature conductivities for the same films (see Figure 1.17b), which reveals the high electrical conductivity of the N-UNCD, critical to developing key implantable biocompatible electrically conductive electrodes and other electronic devices for neural stimulation and other medical applications.

Figure 1.17

(a) Total N atoms content (left axis) and room-temperature conductivity (right axis) as a function of N₂ gas content in the MPCVD plasma. (b) Arrhenius plot of conductivity data obtained in the temperature range 300–4.2 K for a series of films synthesized using different N₂ gas concentrations in the plasma during N-UNCD film growth.

(reprinted from App. Phys. Lett., vol. 79, p. 1441, 2001 (Fig. 3) in [29] with permission from AIP Publisher)

The other key property demonstrated for the N-UNCD films developed by Auciello et al., using chemical seeding plus MPCVD growth [Reference Getty, Auciello, Sumant, George, Islam and Dutta53] and BEN-BEG processes [Reference Ding, Krauss and Auciello79], is sustainable EFE, which is critical for a wide variety of applications. Figure 1.18 shows information on the structural form and electron emission properties demonstrated for N-UNCD film-based electron emission devices. Three important properties were demonstrated for the N-UNCD EFE films described in [Reference Getty, Auciello, Sumant, George, Islam and Dutta53]: (1) N-UNCD films exhibit one of the lowest electric fields (~2 V/µm – Figure 1.18e) for emission onset demonstrated today, for both the conventional sharp field emitter tips (Figure 1.18a,b) and (2) flat surfaces (Figure 1.18c,d), the most recent providing a much lower cost for the electron emission manufacturing process than the more complex and costly fabrication of field emitter tips; and (3) the N-UNCD films reported in [Reference Getty, Auciello, Sumant, George, Islam and Dutta53] were the field emitters with the longest stable emission (1000 h) demonstrated today (Figure 1.18f).

Figure 1.18

SEM images of sharp N-UNCD-coated Si field emitter tips (a) and (b) and planar UNCD-coated field emitter devices (c) and (d). (e) Emission current vs. cathode voltage and electric field emission for N-UNCD field emitter devices. (f) Electron emission current vs. emission time, showing the longest time (1000 h) field emission demonstrated today [Reference Getty, Auciello, Sumant, George, Islam and Dutta53].

(reprinted from Proc. SPIE. vol. 7679, p. 76791N-1, 2010 in [53] with permission from SPIE Publisher)

The R&D performed on the synthesis and characterization of the properties of N-UNCD films by Auciello’s group has been replicated by several groups worldwide, as briefly discussed in the following.

Electrical Conductivity

1. 1. The temperature dependence of Hall (electron) mobility measurements on N-UNCD films, together with an enhanced electron density, was used to interpret the unusual magneto-transport features, correlated with delocalized electronic transport in N-UNCD films, which can be described as low-dimensional superlattice structures. N-UNCD films grown with 10–20% N₂ gas in the MPCVD process showed temperature dependences of conductivity as well as magnetoresistance (MR), contributions from weak localization (WL), and also from variable range hopping (VRH) transport (strong localization) to the net conductivity [Reference Choy, Stoneham, Ortuno and Somoza116].

2. 2. Work by another group provided evidence that WL correction may provide insight into the dominant mechanism for electrical conductivity through grain boundaries in N-UNCD films, at low temperatures (≪300 K), although the details of film microstructures in connection with mechanisms including 2D WL to 3D WL interplay and electron–electron (e–e) interaction effects in both dimensions have not been elucidated yet [Reference Bhattacharyya117]. This work provided evidence of 3D anisotropic and correlated electronic transport in N-UNCD films through conductivity and MR measurements and at magnetic fields as high as 12 Tesla. This work also provided data on Hall resistance measurements in N-UNCD films not only to establish delocalized transport in these films but also to explain the unusual magneto-transport and weak temperature variation of conductivity of the films in the 3D WL framework.

3. 3. N-UNCD films grown by another group, using the MPCVD process with 20% N₂ gas concentration, revealed relatively higher values than those grown with 10% N₂ films, in agreement with the data shown in Figure 1.18b. The estimated values of N atom concentration in the films of about 10¹⁹ cm^–3 can explain the high conductivity of these films and suggests that in these heavily doped UNCD films transport can reach the high end of the diffusive regime, enabling band-like conduction. Although the number of N atoms in the N-UNCD films decreases as the substrate temperature is reduced, the minimum value of conductivity remains >1 × 10¹³ cm^–2, which is a signature for delocalized conduction, which cannot be explained by the hopping conduction mechanism [Reference Shah, Churochkin, Chiguvare and Bhattacharyya118]. The research reported in [Reference Shah, Churochkin, Chiguvare and Bhattacharyya118] indicates that the dominant effect of anisotropic WL in three dimensions is associated with a propagative Fermi surface on the conductivity correction in N-UNCD films, and is supported by magnetoresistance studies at low temperatures. Also, low-temperature electrical conductivity shows WL transport in three dimensions, combined with the effect of e–e interactions in the N-UNCD films, which is remarkably different from the conductivity in 2D WL or strong localization regimes. The corresponding dephasing time of electronic wave functions in these systems, described as ~T^−p with p < 1, follows a relatively weak temperature dependence on conductivity compared to the generally expected conductivity for bulk metals having p ≥ 1. The temperature dependence of Hall (electron) mobility together with an enhanced electron density has been used to interpret the unusual magneto-transport features and show delocalized electronic transport in these N-UNCD films, which can be described as low-dimensional superlattice structures.

Electron Field Emission

As in the case of studies of electrical conductivity of N-UNCD films by several groups worldwide, studies on EFE of N-UNCD films were also performed by different groups worldwide, confirming all the initial studies already described. Studies included the following:

1. 1. N-UNCD films were grown on Si substrates using the MPCVD process, but using a liquid source of N atoms in the form of triethylamine (TEA) powder dissolved in methanol spun on the surface of the substrate [Reference Yuan, Fang and Feng119] to synthesize N-UNCD films on silicon substrates. The MPCVD process involved evaporating the N atoms from the surface of the Si substrate via heating to three different temperatures (760 °C, 830 °C, and 890 °C). The process replicated the previously demonstrated N-UNCD nanostructure of electrically conductive nanowires (see figure 3b,c in [Reference Yuan, Fang and Feng119]). The Raman analysis showed the characteristic two well-defined peaks of the D and G bands. The electrical conductivity of the N-UNCD films grown at a substrate temperature of 830 °C reached 1174 S cm^–1 and exhibited 3.4 V/µm, which is slightly higher than the one shown in Figure 1.18e, but a comparable EFE current density of 8.0 mA cm^–2 at 6.2 V/µm. Summarizing, the work reported in [Reference Yuan, Fang and Feng119] replicated the prior research already described.

2. 2. R&D by several other groups [Reference Jothiramalingan, Haenen, Haenen and Yang120–Reference Wang, Tong, Chen, Shih and Lin122] worldwide basically confirmed all the prior research discussed above on the synthesis and properties of MPCVD-grown N-UNCD films.

Growth of N-UNCD Films by MPCVD via the BEN-BEG Process

The first demonstration of the BEN-BEG process for growing N-UNCD films was done by Auciello et al. [Reference Ding, Krauss and Auciello79]. N-UNCD films were grown on high-conductivity Si substrates heated to 800 °C using microwave power of 600 W at a total pressure of 11 Torr. One group of films was grown with Si substrates biased at –100 V using a gas mixture of N₂ (1%) /CH₄ (1–20%) / H₂ (98–79%) [Reference Ding, Krauss and Auciello79], while a second group of films was grown with a substrate bias in the range –100 to –150 V in a gas mixture of N₂ (1%)/CH₄ (10%)/H₂ (89%) [Reference Ding, Krauss and Auciello79]. The field emission performance in terms of threshold field and emission current improved considerably as a function of increasing CH₄ concentration and negative bias voltage [Reference Ding, Krauss and Auciello79]. UV Raman analysis revealed that the field emission enhancement resulting from an increase in CH₄ concentration from 1% to 5% correlates with a decrease in the sp³-C atoms bonding in the N-UNCD films. The dependence of field emission on negative bias voltage appears to be correlated with ion bombardment-induced damage in the film during growth. SEM images of N-UNCD films grown with –150 V bias showed smaller surface topographic features compared to films grown under 0 and –100 V bias. J vs. E₀ measurements across a length of 40 mm over the N-UNCD film showed that films grown with a substrate bias of –150 V exhibited the lowest threshold field for electron emission demonstrated at that time (~2.0 V/µm) [Reference Ding, Krauss and Auciello79].

The initial R&D described above [Reference Ding, Krauss and Auciello79] opened a new field of R&D to optimize the growth of electrically conductive N-UNCD films for high-tech and biomedical applications, the latter motivated by the fact that N atoms inserted in biocompatible UNCD films provide probably the best biologically suitable atoms in relation to interactions with the human biological environment. The big advantage of using the BEN-BEG technique to grow N-UNCD films, as also discussed for UNCD films in Section 1.2.1.3, is that the time-consuming chemical seeding process is eliminated and the overall cost of growing N-UNCD films is also substantially reduced.

Diamond Film Growth.

HfO₂ thin films are grown via atomic layer deposition (ALD) using Tetrakis (dimethylamido) hafnium (Hf(NMe₂)₄) as the hafnium precursor and H₂O as the oxidation precursor. HfO₂ films of 5, 10, 30, and 100 nm thicknesses were grown on SiO₂/Si substrates heated up to 100 °C, using a nitrogen purge of 20 seconds between each precursor pulse. All substrates with the HfO₂ layers on the surface were subjected to the chemical-based diamond seeding process via immersion of the substrate in a commercial solution of methanol with functionalized diamond particles (to avoid seed particles agglomeration to induce high-density pinhole-free diamond films) in an ultrasound generator, which immersed UNCD (3–5 nm) particles from the seeding solution on the substrate surface via sound wave-induced vibration. Details of the seeding process can be found in [Reference Auciello and Sumant22]. Following the seeding process, the substrates were placed inside the HFCVD chamber on a rotating substrate holder to promote film uniformity. Details of the HFCVD geometry, gas flow, and filament and substrate heating were described in Section 1.3.2.2. Briefly, the substrates were heated to ~800 °C for growing MCD and NCD films and ~650 °C for growing UNCD films. The gas flows used to grow the diamond films were: (1) for MCD and NCD films, CH₄ (3 sccm)/H₂ (200 sccm) for 8 h and 4 h, respectively; and (2) for UNCD films, CH₄ (2 sccm)/H₂ (10 sccm)/Ar (90 sccm) for 1 h, keeping the total chamber pressure at 10 Torr. The details of the HFCVD process to grow MCD to UNCD films can be found in a recent paper [Reference Fuentes-Fernandez, Alcantar-Peña and Lee74].

The next step, after the diamond film growth, was to study their structural and chemical composition properties using complementary electron microscopy techniques and Raman and XPS analyses, respectively, as described below.

Characterization of Diamond Films Structure by Electron Microscopy.

Figure 1.24 shows SEM images of UNCD, NCD, and MCD films grown on HfO₂ layers on Si, demonstrating that the HfO₂ layer plays a critical role in inducing growth of dense PCD films with all structures.

Figure 1.24

SEM images of the top surface of UNCD films grown on SiO₂ (a), WO_x (b), TiAlO_x (c), and HfO₂ (d), showing that the densest UNCD film is grown on a HfO₂ layer. (e,f) Dense NCD and MCD films also grow with high-density structures on HfO₂ layers on Si.

(reprinted from Diam. Relat. Mater., vol. 69, p. 221, 2016 (Fig. 3) in [41] with permission from Elsevier Publisher)

Figure 1.25 shows cross-section SEM images correlated with the SEM images of the top surface shown in Figure 1.24, such that the cross-section confirms the information provided by the top surface images that UNCD, NCD, and MCD films are grown with extreme dense structures on HfO₂.

Figure 1.25

SEM images of the cross-section of UNCD films grown on SiO₂ (a), WO_x (b), TiAlO_x (c), HfO₂ (d), showing that the densest UNCD film is grown on a HfO₂ layer, with a nanoscale smooth UNCD/HfO₂ interface. (e,f) Dense NCD and MCD films also grow with high-density structures and nanoscale smooth UNCD/HfO₂ interfaces on HfO₂ layers on Si.

(reprinted from Diam. Relat. Mater., vol. 69, p. 221, 2016 (Fig. 4) in [41] with permission from Elsevier Publisher)

The fundamental mechanism for growth of UNCD films on HfO₂ layers was revealed by systematic detailed studies using HRTEM, Raman, and XPS analyses. Figure 1.26 shows the nanostructure of a 140 nm thick UNCD film grown uniformly on a 10 nm thick HfO₂ template layer on an Si surface, as shown in the LRTEM (low-resolution transmission electron microscopy) image of Figure 1.26a. Figure 1.26b shows a plain view HRTEM image of the UNCD film, revealing the characteristic 3–5 nm grain sizes of UNCD, with the diamond crystal structure confirmed in the diffraction pattern of the inset. Figure 1.26c shows an HRTEM cross-section image that reveals the Si substrate and HfO₂ layer, and the HfO₂/SiO₂ and UNCD/HfO₂ interfaces. The HfO₂ layer, amorphous as it is grown by ALD, crystallizes during the film growth, as indicated in the upper left inset diffraction pattern of Figure 1.26c, showing clear spots characteristic of the HfO₂ crystalline structure. The UNCD/HfO₂ interface exhibits a stripe-type structure, highlighted by the rectangle in Figure 1.26c, pointing to the upper right inset, which shows the diffraction patterns of the UNCD/HfO₂ interface. A comparison between the diffraction patterns of pure UNCD (inset in Figure 1.26b) and the UNCD/HfO₂ interface (upper right inset in Figure 1.26c) reveals the appearance of extra peaks for the diffraction at the interface, indicating the presence of a new phase at the interface. This new phase most probably appears during the first stages in the nucleation of the UNCD films, as explained later in the text. The HRTEM-EELS elemental analysis profile, shown in Figure 1.26d (scanning direction indicated with an arrow across the Si/HfO₂/UNCD layered structure shown in Figure 1.26a), confirms the elemental profile through the whole SiO₂/HfO₂/UNCD heterostructure).

Figure 1.26

LRTEM cross-section (a) and HRTEM plain (b) and cross-section (c) images of UNCD film on UNCD/HfO₂/SiO₂ heterostructure. (d) HRTEM-EELS elemental profile analysis across the UNCD/HfO₂/SiO₂ heterostructure showing the distribution of C, O, Hf, and Si atoms correlated with the mechanism of diamond film nucleation and growth described in the main text.

(reprinted from Diam. Relat. Mater., vol. 69, p. 221, 2016 (Fig. 1) in [41] with permission from Elsevier Publisher)

Chemical Analysis of UNCD Films.

Characterization of the chemical bonds in UNCD films was done using Raman and XPS analysis, as complementary studies, to confirm the results presented above from electron microscopy. Figure 1.27Aa shows visible Raman spectra from UNCD films grown on HfO₂ layers (with various thicknesses from 5 nm to 100 nm) grown by ALD on SiO₂/Si substrates. The Raman spectra were obtained with a visible 532 nm laser beam wavelength. The spectra show characteristic D band (1341.7 cm^–1) and G band (1588.5 cm^–1) peaks from the sp²-C atom bond hybridization, identical for all UNCD films grown on the HfO₂ layers with thickness in the 5–100 nm range (see Figure 1.27Aa). Deconvolution of the Raman peaks shows that there is a contribution of three extra peaks, green (at 1489.2 cm^–1), violet (at 1258.3 cm^–1), and yellow (at 1164.3 cm^–1), to the D and G peaks, reported as representative of C–C atoms bonds stretching in trans-polyacetylene (TPA) molecules, broadened vibrational density of states (VDOS) of small diamond clusters, tetrahedral amorphous carbon, and TPA C–H bending combined with C–C stretching, respectively [Reference Klauser, Steinmüller-Nethl, Kaindl, Bertel and Memmel140, Reference Ferrari and Robertson141]. The literature [Reference Klauser, Steinmüller-Nethl, Kaindl, Bertel and Memmel140, Reference Ferrari and Robertson141] shows that the diamond peak (1332 cm^–1) in UNCD films is always masked by the D mode peak of disordered carbon because visible Raman scattering from C atoms is mainly sensible to the dominating sp² bonds in the large network of UNCD grain boundaries, and much less to the sp³ C bonds of diamond inside the grains. Figure 1.27Ab shows Raman spectra produced by UV (244 nm wavelength) laser beam, which provides sufficient energy to excite the σ states of both sp² and sp³-C bonds of diamond, with a well-defined 1332 cm^–1 peak from a single crystal diamond, an NCD film (10–100s nm grain size), and even a UNCD film. The disappearance of the 1150 and 1458 cm^–1 peaks, shown in the visible (532 nm) Raman spectra, for the UV (244 nm) spectra (compare Figures 1.27Aa and Ab) indicate that those peaks are not the consequence of phonon confinement. This is due to the fact that if these peaks were zone boundary phonons, then they should be present in UV excitation as well [Reference Ferrari and Robertson141]. On the other hand, the visible Raman spectrum shows that the intensity of the broad peak from the graphite D band (1340–1350 cm^–1) is inversely proportional to the crystal size, as indicated in Figure 1.27Aa, although the appearance of this feature was attributed to the breakdown of the q = 0 selection rule. A Raman interpretation model [Reference Ferrari and Robertson141] indicates that for nanocrystalline diamond, the characteristic Raman diamond peak at 1332 cm^–1 is strongly reduced in intensity, and the D and G band peaks dominate the visible Raman spectra.

Analysis by XPS provided valuable information on the atom bonding in the initial UNCD film nucleation and growth stages. The appearance of the C1s, Hf4f, and O1s XPS peaks provided key information, complementing the HRTEM images, indicating that the UNCD films nucleate on HfO₂ due to the formation of a nanometer-thick HfC interface layer at the very initial nucleation stage. The first row of data in Figure 1.27Ba shows the XPS spectra of C1s, Hf4f, and O1s from the UNCD film grown for 5 min. Figure 1.27Bb and Bc show similar XPS data for UNCD films grown for 10 and 15 min, respectively. For UNCD films grown for 5 and 10 min, the Hf4f and O1s peaks dominate the XPS spectra, while a small but noticeable (C1s) peak appears at the binding energy of ~281.8 eV, which correlates, within experimental error, with the 281.5 eV C1s peak of HfC [Reference Špatenka, Shur, Erker and Rump142, Reference Ramqvist, Hamrin, Johansson, Fahlman and Nordling143]. The C1s peak of HfC does not appear in the spectrum of the UNCD film grown for 15 min, which is due to the thickness of the UNCD film covering the HfO₂ film. The C1s peak appearing at 184.5 correspond to the C atom bonds in the UNCD films.

The data from the Raman and XPS analyses shown in Figure 1.27 provide valuable information to confirm the mechanism of nucleation and growth of UNCD films on HfO₂ layers, which may play a critical role in the development of a new generation of micro/nanoelectronic devices.

Figure 1.27

A (a) Raman spectra of UNCD films grown on HfO₂ layers, obtained with a 532 nm wavelength visible laser beam, showing the characteristic D band (1341.7 cm⁻¹) and G band (1588.5 cm⁻¹) peaks; this figure includes a deconvolution of the Raman spectrum, showing the contribution of two peaks corresponding to the G band peak (1588.5 cm⁻¹ ): 1 (1589 [cm⁻¹] and 2 [1500 cm⁻¹] peaks), and two peaks corresponding to the D band peak ((1341.7 cm⁻¹): 3 (1341 cm⁻¹) and 4 [1298 cm⁻¹]), peaks; the deconvoluted peak 5 at ~ 1150 cm⁻¹) correspond to the TPA peak correlated with H-rich molecules in the grain boundaries; (b) Raman spectra of a UNCD film grown on a HfO₂ layer, obtained with a 244 nm wavelength UV laser beam, showing a very small but visible 1332 cm⁻¹ peak characteristic of diamond, which is exactly at the position where the strong 1332 cm⁻¹ diamond peak of crystalline diamond appears (Raman spectra for a NCD film and for a crystalline diamond are shown to confirm the position of the diamond pea); (c) Arrhenius plot ? growth rate (k) vs. 1/T – for UNCD films grown on HfO₂ layers at different temperatures (fitting of the experimental data points provides the activation energy for growth of UNCD films at ~18.4 kcal/mol, which is within the range of activation energies that yield high-quality diamond films).

B XPS spectra from UNCD films grown on HfO₂ layers for 5 min (a), 10 min (b), and 15 min (c). The spectra for the C1s and Hf 4f peaks reveal the presence of peaks at ~284.5 eV and ~14.7 eV, respectively, which confirm the formation of a HfC interface layer that induces nucleation of the UNCD films, the peaks for O 1s confirm the presence of O atoms contributing to form stoichiometric HfO₂ films.

(Reprinted from Diam. Relat. Mater., vol. 69, p. 221, 2016 (Fig. 2) in [41] with permission from Elsevier Publisher).

In concluding this section, it is relevant to note that in addition to the development of a process to produce high-quality UNCD coatings on HfO₂ layers on Si, Figure. 1.24f and 1.25f show that high-quality MCD films can be grown on HfO₂ layers on Si substrates. This is a key result that opens the way to develop the coating technology to cool down the next generation of Si-based micro/nanoelectronics, since it has been demonstrated that MCD exhibit a thermal transport of ~1900 W/K·m [Reference Auciello and Sumant22], very close to the best thermal transport of any material perpendicular and parallel to the film surface, which is for single crystal diamond (~2100 W/K·m) [Reference Auciello and Sumant22]. However, further R&D is needed to develop a process to grow MCD at ≤450 °C, as has been developed for UNCD coatings [Reference Auciello and Sumant22], in order to be within the thermal budget of Si microchips to avoid destroying the Si device, as occurs when heated above 450 °C.

Background on Research on Wettability of Diamond Surfaces

Before describing the new R&D process for tailoring the wettability of UNCD film surfaces, it is relevant to provide background information on the prior R&D done on the wettability of diamond surfaces. In this sense, from the information shown in Table 1.2 and the literature cited therein, it appears that there are no reports on systematic research on the influence of grain size of PCD films on the surface chemistry termination and their effects on the wettability of the treated surfaces. Therefore, the research reviewed in this section relates to work performed by Montano et al. [Reference Gabriela Montano-Figueroa, Alcantar-Peña and Tirado152], which was focused on performing systematic studies of O₂, H₂, and CF₄-based plasma treatment of MCD, NCD, and UNCD films, and the effect of combined surface chemistry and UNCD-coated 3D microstructured surfaces, all on surface wettability and related performance.

Growth of UNCD Films by Chemical Seeding Plus HFCVD Process

Montano et al. [Reference Gabriela Montano-Figueroa, Alcantar-Peña and Tirado152] investigated wettability of MCD, NCD, and UNCD film surfaces. However, only the studies on UNCD film surface wettability are reviewed here, and the reader is referred to [Reference Gabriela Montano-Figueroa, Alcantar-Peña and Tirado152] to read about the research on MCD and NCD films.

The UNCD films were grown using the HFCVD technique. The UNCD films were grown on electrically conductive р-Si (Reference Coffman, Cao and Pianetta111) wafers inserted in the HFCVD chamber onto a rotating substrate holder, located underneath an array of 10 parallel tungsten (W) filaments (see Section 1.3.2.2. and Figure 1.20 and [Reference Fuentes-Fernandez, Alcantar-Peña and Lee74]). The parameters used to grow the UNCD films are shown in Table 1.3, which correlates with the production of nanoscale dense UNCD films on Si surfaces (Figure 1.28). For the research focused on investigating the effect of 3D structures on wettability of UNCD films, these films were grown on a large array of micropatterned pillars on Si substrates (Figure 1.29) using well-established photolithography and reactive ion etching processes [Reference Gabriela Montano-Figueroa, Alcantar-Peña and Tirado152]. Because of the nanoscale size (3–5 nm) of UNCD, these films can be grown with excellent thickness uniformity (Figure 1.28) and density on high aspect ratio 3D structures like pillars (Figure 1.29c,d). The parameters used to grow UNCD films on the 3D Si pillars are shown in the first row of Table 1.3. The Si pillar dimensions were: diameter, 20 μm; height, 5 μm; and distance between pillars, 20 μm.

Treatment of PCD Films Surfaces with O₂, H₂, and CF₄ Plasmas to Tailor Wettability and Characterization

Characterization of Key Parameters of the Treated Surfaces of UNCD Films to Produce Super-hydrophilic to Super-Hydrophobic Surfaces

Thickness and Surface Morphology of Flat and 3D Structured UNCD films.

The thicknesses of the UNCD films were measured using cross-section SEM imaging (Figure 1.28), and the surface roughnesses were measured using AFM (top right insert in Figure 1.28). The AFM measurements revealed the characteristic nanoscale surface roughness of UNCD films in the range of 3–7 nm.

Figure 1.28

Cross-section SEM image of a UNCD film grown using HFCVD on the surface of an Si wafer; the top right insert shows an AFM image of the surface of the UNCD film, revealing an rms roughness of about 3–7 nm.

(reprinted from Carbon, vol. 139, p. 361, 2018 (Fig. 2) in [152] with permission from Elsevier Publisher)

UNCD films were grown on a large array of micropatterned pillars on Si substrates, such that the top surface exhibited the characteristic UNCD smooth structure (Figure 1.29a). Figure 1.29b shows a low-magnification SEM image of the large array of Si pillars coated with a very conformal UNCD film. Figure 1.29c,e shows high-magnification cross-section SEM images of an Si pillar coated with UNCD film, revealing the extreme uniformity of the UNCD film thickness achieved on the high aspect ratio 3D pillar structures [Reference Gabriela Montano-Figueroa, Alcantar-Peña and Tirado152]. Finally, Figure 1.29d shows a summary of the static WCA measurements on UNCD film surfaces with different chemical treatments on flat and 3D structures, revealing the orders of magnitude change in the surface nature of UNCD films from super-hydrophilic, when treated with an O₂ plasma, to super-hydrophobic when treated with a CF₄ plasma and grown on a microstructured Si large pillar array.

Figure 1.29

(a) SEM top view of a flat UNCD surface on top of an Si pillar. (b) Low-magnification image of UNCD-coated Si pillars array. (c,e) Part of a high-magnification image of a UNCD-coated Si pillar, showing the excellent dense/conformal coating achieved on the 3D structured Si surface. (d) WCA vs. chemical surface treatment of UNCD coatings grown on flat Si substrates and on a 3D structured Si pillar array, showing the controllable wettability properties by different means, tailoring from super-hydrophilic surface with WCA ~ 0 to super-hydrophobic surface (UNCD-coated Si pillar treated with CF₄ plasma) with WCA 178° (including the measurement error).

(reprinted from Carbon, vol. 139, p. 361, 2018 (Fig. 3) in [152] with permission from Elsevier Publisher)

The importance of controlling the wettability properties of a large range of PCD films through a relatively easy and short-duration process is that it has a direct impact on a wide range of applications to biomedical devices and protheses coated with UNCD films.

Water Contact Angle Measurements on UNCD Film Surfaces.

Figure 1.30 shows the systematic measurement of the static WCA from all UNCD film flat surfaces and their XPS survey analysis, as well as the microtextured UNCD surfaces grown on the Si pillar arrays. The pictures in Figures 1.30a–c show the results from WCA measurement of flat UNCD surfaces and their change in surface energy [Reference Gabriela Montano-Figueroa, Alcantar-Peña and Tirado152]:

1. 1. Super-hydrophilic (total wetting of the surface) with high surface energy, after 30 s O₂ plasma treatment (Figure 1.30a), inducing O atoms and/or O–H molecules attaching to the surface, and reacting chemically with water molecules.

2. 2. Medium hydrophobic (with UNCD-H terminated surface by MPCVD H-plasma processing (Figure 1.30b).

3. 3. Hydrophobic, after 60 s CF₄ plasma treatment (Figure 1.30c) with low surface energy.

4. 4. Highly hydrophobic, pillar textured Si surface coated with a UNCD film with the surface terminated by H atoms via H-plasma treatment with an MPCVD H-plasma process, such that the H atoms, chemically bonding to C open bonds on the UNCD surface, inhibit chemical reactions of the C atoms with water (H₂O) molecules (Figure 1.30d) (WCA = 159°) [Reference Gabriela Montano-Figueroa, Alcantar-Peña and Tirado152].

5. 5. Pillar array textured Si surface coated with UNCD films, plus exposure to a CF₄ plasma inducing a super-hydrophobic surface, induced by the combination of texture plus UNCD surface termination with F atoms, inducing the highest water drop contact angle demonstrated today (~178° water drop contact angle to the UNCD surface) [Reference Gabriela Montano-Figueroa, Alcantar-Peña and Tirado152].

XPS analysis of the surface of the treated UNCD films, shown in Figure 1.30f, showed the following results:

1. 1. The oxygen plasma treatment resulted in replacement of H atoms on the surface of the as-grown UNCD films by O and OH groups. This agrees with results reported by other groups [Reference Popov, Vasilchina and Kulisch154, Reference Park, Son and Kim162].

2. 2. In the case of CF₄ plasma treatment, a single XPS narrow peak appeared for the fluorinated UNCD surface, which is consistent with the formation of a carbon-monofluoride species on the surface, as also confirmed by other authors [Reference Denisenko, Romanyuk, Pietzka, Scharpf and Kohn158]. So, the presence of the prominent F1s peak confirms the surface fluorination of the UNCD film, which was not shown in the survey XPS spectra for the films treated with O₂ or not treated at all. The bonding energy of the C–F bond is very high, thus enabling high stability of the fluorinated surface.

Figure 1.30

(a) Flat UNCD film surface treated with an O₂ plasma for 30 s. (b) Flat UNCD surface terminated with H atoms. (c) Flat UNCD surface treated with a CF₄ plasma for 60 s. (d) UNCD-coated Si pillar array with the UNCD surface terminated with H atoms. (e) Fluorinated UNCD-coated Si pillar array treated with a CF₄ plasma for 50 s. (f) Survey XPS spectra of as-grown UNCD film (bottom curve), H-surface terminated UNCD film (second curve from bottom), O-surface terminated UNCD film (third curve from bottom), and finally F-surface terminated UNCD film (top curve) [Reference Gabriela Montano-Figueroa, Alcantar-Peña and Tirado152].

(reprinted from Carbon, vol. 139, p. 361, 2018 (Fig. 4) in [152] with permission from Elsevier Publisher)

The results reported in [Reference Gabriela Montano-Figueroa, Alcantar-Peña and Tirado152] indicate that when a water drop interacts with a textured surface, such as the Si micropillars, two distinguished wetting states can be formed: the Wenzel state (θ_W), where the liquid penetrates the structures, wetting the whole surface; or the Cassie–Baxter state (θ_CB), where the water droplet is suspended on top of the textures and the droplets placed on them remain almost spherical [Reference Nosonovsky and Bhushan163, Reference Kim, Choi and Williams164]. Hence, θ_W and θ_CB were calculated by the researchers of [Reference Gabriela Montano-Figueroa, Alcantar-Peña and Tirado152] for the two cases: (1) Si micropillars coated with UNCD film; and (2) Si micropillars coated with UNCD film exposed to CF₄ plasma. θ_W and θ_CB calculated values are shown in Table 1.4. The values of f (the fraction of the liquid droplet touching solid and not air) represents the fraction of the solid–liquid interface in the entire composite surface beneath the liquid, obtained by using the Cassie–Baxter formula, and r (the ratio of the actual area of the solid–liquid interface to the normally projected area), obtained by using the Wenzel formula, was used to calculate θ_W and θ_CB, enabling correlation with experimental measurements. The calculations indicate that the Wenzel model correlates better with the wettability state of the UNCD-coated Si pillar array without fluorine treatment, while the Cassie–Baxter model correlates better with the state of the fluorinated UNCD-coated Si pillar array (Table 1.4).

Table 1.4Values of θ_w and θ_CB calculated using the Cassie–Baxter and Wenzel formulas, respectively (reprinted from Carbon, vol. 139, p. 361, 2018, in [152] with permission from Elsevier Publisher).

Sample	θmeas	θW	θCB
Si pillars + UNCD	130.43 + 8.01	141.10	158.39
Si pillars + UNCD + CF4	178. ± 1.0	NA	176.40

The results described above revealed the key role that CF₄ plasma treatment of UNCD film surface has on enhancing the hydrophobic intrinsic behavior of a material such as UNCD, by controlling the surface chemistry, combined with the microstructure of the substrate, resulting in a synergistic effect that leads to the formation of Cassie–Baxter droplets, where the liquid sits partially on the air trapped underneath. The experimental observation is that water does not adsorb at all on the fluorinated UNCD micro-/nanostructured surface, which provides a new material paradigm for application to a broad range of blood-interfacing implantable medical devices.

XRD Analysis.

XRD analysis on all films (Figure 1.33) confirms the information from the Raman analysis, revealing that the UNCD films tend to grow at and near the edge faster than at the center and middle of the substrate. Figure 1.33a shows that for BEG UNCD films grown for 0.5 and 1.0 h, the main peaks are located at positions 1 (31.5°), 2 (35.5°), 3 (39.7°), and 5 (48.5°), which correspond to WC (001), WC (100), W₂C (Reference Collins, Lightowlers and Field101), and WC (Reference Collins, Lightowlers and Field101), respectively [Reference Dash, Nayak and Abhangi178]. Diamond (Reference Coffman, Cao and Pianetta111) appears at 2θ = 44°, represented by peak 4 in Figure 1.33a, which starts showing strong intensity after 2.0 h BEN-BEG of UNCD films. Diamond (220) at 75° is convoluted with the WC (Reference Coffman, Cao and Pianetta111) and W₂C (221) at 73° and 76° (peaks 7 and 8 in Figure 1.33a). All diamond directions were confirmed by HRTEM diffraction (Figure 1.35). For 2.0 and 2.5 h BEN-BEG of UNCD films, the diamond (311) can be seen in HRTEM, but it cannot be corroborated with XRD since it is a non-allowed diffraction due to the FCC nature of the diamond structure. Figure 1.33 (b) shows the XRD spectra from analysis on the center, middle, and edge areasof a UNCD film grown for 1 h. by BEN-BEG. The XRD peaks correspond mainly to the WC nucleation layer, correlating with the XRD spectra shown in Figure 1.33 (a) for films grown up to 1.5 hrs, with the small diamond (111) peak (5), at 2θ = 44˚, just starting to appear on the edge area, where the film nucleates faster. However, there is no intense diamond (111) peak because the film needs to be grown for ~ 2 hrs to exhibit the dense structure. These results support the need for using complementary analytical techniques to obtain reliable information.

Figure 1.33

(a) XRD spectra from analysis of BEN-BEG UNCD films grown for 0.5–2.5 h plus additional growth time of 2 h without bias. : b) XRD spectra from analysis on center, middle, and edge areas of a UNCD film grown for 1 h. by BEN-BEG (the XRD peaks correspond mainly to the WC nucleation layer, correlating with the XRD spectra shown in (a) for films grown for up to 1.5 hrs., with the small diamond (111) peak (5) at 2θ = 44˚, just starting to appear on the edge area, where the film nucleate faster (see discussion of the peaks identification in the text).

(reprinted from Diam. Relat. Mater., vol. 78, p. 1, 2017 (Fig. 3) in [135] with permission from Elsevier Publisher)

SEM Analysis.

SEM imaging of the top UNCD film surface on the center, middle, and edge areas (Figure 1.34) confirm the data from Raman and XRD analysis, showing that the UNCD film growth from edge to center, combined with the longer BEN-BEG time (2.5 h), produces the denser UNCD films. The cross-section SEM images show good film thickness uniformity across the 100 mm diameter substrates.

Figure 1.34

SEM top view images of HFCVD BEN-BEG UNCD films grown for times in the range 0.5–2.5 h, with additional growth time of 2 h without bias. The inset figures (top right corner) for the center, middle, and edge areas of the films grown for 2 and 2.5 h show cross-section SEM images of the full dense films grown across the 100 mm diameter substrates.

(reprinted from Diam. Relat. Mater., vol. 78, p. 1, 2017 (Fig. 4) in [135] with permission from Elsevier Publisher)

HRTEM Studies.

HRTEM images (top view) of BEN-BEG UNCD films grown for times in the range 0.5–2.5 h, plus 2 h growth without bias, are shown in Figure 1.35. All films showed grain sizes of 2–5 nm, characteristic of the UNCD nanostructured films for many applications [Reference Naguib, Birrell, Elam, Carlisle and Auciello21–Reference Shenderova and Gruen23, Reference Carlisle, Gruen, Auciello and Xiao40, Reference Alcantar-Peña, Lee and Fuentes-Fernandez41, Reference Auciello, Shenderova and Gruen46, Reference Tzeng, Auciello, Liu, Lin and Cheng48–Reference Krauss, Ding and Auciello50, Reference Shi, Jin, Chen and Auciello58, Reference Auciello59, Reference Fuentes-Fernandez, Alcantar-Peña and Lee74].

EELS Analysis.

EELS analysis was performed also for all BEN-BEG UNCD films grown in the range 0.5–2.5 h, plus 2 h growth without bias (see a typical curve, similar for all films, in Figure 1.35f). The curve shown in this figure reveals the characteristic spectrum intensity vs. energy loss (step up at about 290 eV and peaks at about 303 and 310 eV) of UNCD obtained in EELS analysis of numerous UNCD films grown with the conventional chemical seeding process using both the MPCVD and the HFCVD processes by many groups worldwide.

Figure 1.35

Top view HRTEM images of BEN-BEG UNCD films grown for (a) 0.5 h, (b) 1 h, (c) 1.5 h, (d) 2 h, and (e) 2.5 h, plus 2 h growth without bias. (f) Spectrum from EELS analysis corresponding to all BEN-BEG UNCD films, for which HRTEM images are shown in (a)–(e). The inset top right figures show electron diffraction patterns from the UNCD films (the arrows in the diffraction patterns in (a)–(e), pointing to 2.08, 206, and 2.05 Å, correspond to diamond (Reference Coffman, Cao and Pianetta111) orientation; the arrows pointing to 1.27, 1.29, and 1.22 Å correspond to diamond (220) orientation, and the arrows pointing at 1.02 Å correspond to diamond (311) orientation.

(reprinted from Diam. Relat. Mater., vol. 78, p. 1, 2017 (Fig. 5) in [135] with permission from Elsevier Publisher)

EDAX Analysis.

Energy-dispersive X-ray spectroscopy (EDAX) analysis was performed during the HRTEM studies, through the cross-sections of all the UNCD films, from the surface into the Si substrate. Figure 1.36 shows the EDAX cross-section analysis profiles from the surface of the UNCD films to the diamond/W/Si interfaces for all BEN-BEG UNCD films grown for 0.5–2.5 h, plus 2 h growth without bias. For all UNCD films, it can be seen that Si diffuses into the W layer and vice versa, induced by the high-temperature (600–700 °C) growth process, as observed by other groups [Reference de Luca, Poravoce and Texier179]. In addition, as the time of exposure of the substrate to the BEN-BEG process increases, the electric field plus ion bombardment may induce further interdiffusion of W into the UNCD films, which can explain the observation of W particles into the diamond network for 2.0 and 2.5 h BEN-BEG process. A key feature shown by the EDAX analysis is that Si atoms diffused into the W layer do not reach the surface of the W layer, correlating with the fact that WC regions formed at the W layer surface appear to be the dominant regions contributing to nucleation of the UNCD grains.

Figure 1.36

EDAX analyses through the cross-sections of BEN-BEG UNCD films grown for (a) 0.5 h, (b) 1 h, (c) 1.5 h, (d) 2 h, and (e) 2.5 h, plus 2 h growth without bias. The EDAX spectra correspond to the HRTEM images shown in

(reprinted from Diam. Relat. Mater., vol. 78, p. 1, 2017 (Fig. 6) in [135] with permission from Elsevier Publisher)

Studies of UNCD/W Layer Interfaces via Cross-Section HRTEM.

A key study enabled understanding of the nucleation process for HFCVD BEN-BEG of UNCD films on a W layer (~40 nm thick) on an Si substrate. Figure 1.37 shows cross-section HRTEM images of the UNCD–W layer interface at the UNCD film nucleation sites for the BEN-BEG UNCD films grown (a) 0.5 h, (b) 1 h, (c) 1.5 h, (d) 2 h, and (e) 2.5 h, plus 2 h growth without bias. All cross-section images were taken at diamond nucleation sites identified via electron diffraction. Determination of the diamond lattice spacing in the UNCD films was done via combination of experimental measurements and modeling using the GMS3 software, applied to different crystal structures containing W and C atoms, such as W, WC, W₂C, graphite, and diamond, detected at the UNCD–W interface. The cross-section HRTEM images, shown in Figure 1.37, revealed that crystalline W nuclei are incorporated in the diamond structure during the growth process. Although EDAX cross-section analysis shown in Figure 1.36 indicates that Si atoms can diffuse close to the WC interface, SiW nuclei or Si were not detected via XPS analysis at the WC–C interface. The prevalent diamond orientation in the UNCD grains, as indicated by the XRD analysis shown in Figure 1.33a is (111), because of minimization of the surface energy [Reference Liu and Dandy180]. The interfaces investigated (Figure 1.37) showed the appearance of WC with different orientations (Figure 1.37a) in all the interfaces studied, suggesting that WC plays a key role in the nucleation of diamond crystals. Furthermore, the diamond (111) and the WC (101) have very similar structural spacing, allowing for heteroepitaxial growth. Other crystal structures like graphite and pure W could also be present in the interface, which subsequently starts inducing the nucleation of WC regions on the W layer grown on the Si surface (confirmed by the HRTEM [Reference Alcantar-Peña, de Obaldia and Montes-Gutierrez135]) that incubates the growth of the UNCD grains.

Figure 1.37

Cross-section HRTEM images of BEN-BEG UNCD?W (40 nm thick) interfaces on Si substrates, grown for (a) 0.5 h, (b) 1 h, (c) 1.5 h, (d) 2 h, and (e) 2.5 h, plus 2 h growth without bias. (f) cross-section LRTEM image of UNCD film grown on W (40 nm thick) layer on Si, showing the nucleation of UNCD film on WC and the extended UNCD grown layer for which HRTEM is shown in (c). (g) table showing identification of W, WC, and diamond phases via electron diffraction from the areas in the squares, where the d-spacing values are correlated with the numbers indicated by the arrows in each inset showing the e-diffraction patterns.

(reprinted from Diam. Relat. Mater., vol. 78, p. 1, 2017 (Fig. 8) in [135] with permission from Elsevier Publisher)

XPS Analysis of UNCD Films.

To complete the systematic analysis of the HFCVD BEN-BEG UNCD films via complementary characterization techniques, XPS analysis was performed (Figure 1.38) on all the BEN-BEG UNCD films described above. The only elements found at the surface of the UNCD films were carbon and tungsten (Figure 1.38a), in addition to O, which is due to exposure of the UNCD surface to the atmospheric environment when transferring the sample from the HFCVD system to the XPS system. No Si was found, and the oxygen in the surface disappeared after the Ar⁺ sputtering cleaning with 1 KeV Ar⁺ ions for 30 s. Figure 1.38a shows that the C atoms in the 0.5 h BEN-BEG UNCD films are bound to W atoms, which changes the binding energy of C atoms from 284.5 eV to 282.7 eV [Reference Liu, Li and Zhai181]. As the BEN-BEG film growth process develops, the amount of C atoms and CH_x radicals increases, such that C atoms that do not bind to W are free to start binding between each other, enabling the formation of diamond nuclei sites. The XPS W peaks are present as long as the surface is still being converted from W to W₂C and WC. For BEN-BEG UNCD films grown for 2.0 h and 2.5 h, the layer of crystalline diamond is dense enough that no C atoms appear bound to W and no W atoms are visible (green flat curve in figure S8B in [Reference Alcantar-Peña, de Obaldia and Montes-Gutierrez135]). The XPS C peak bond energy (285 eV) for UNCD films grown for 1.5–2.5 h corresponds to the C peak bond energy characteristic in the diamond phase. To further understand the Si–W and WC–C interfaces, the 0.5 h HFCVD BEN-BEG-grown UNCD films were analyzed as a function of depth from the film surface, using XPS profiling, involving Ar⁺ ion bombardment to produce physical etching of the film. The depth profiling showed no evidence of Si at the WC–C interface (Figure 1.38b). The W–WC interface could not be determined, but it was clear that there was a very thin layer of SiO₂ between the Si–W/WC interface, as shown by the presence of a sharp O peak at the Si–W interface (Figure 1.38b). The data shows that, initially, the CH_x⁺ (x = 1, 2, 3) radical ions and C⁺ ions produced in the plasma are accelerated toward the negatively biased substrate, producing sub-plantation at the W layer surface, inducing preferential formation of WC instead of sp³-C bond formation characteristic of diamond. Once the W layer surface is transformed to WC, then formation of sp³-C bonds occurs, corresponding to the diamond phase [Reference Liu, Xie, Sun, Ning and Jiang182].

Figure 1.38

XPS analysis of HFCVD BEN-BEG-grown UNCD films, showing the presence of C, O, and W atom binding energy (a) at 0.5 h BEN-BEG time, and the XPS analysis profile spectra (b) when inducing physical etching of the film, via in situ Ar⁺ ion beam bombardment-induced sputtering in the XPS system.

(reprinted from Diam. Relat. Mater., vol. 78, p. 1, 2017 (Fig. 7) in [135] with permission from Elsevier Publisher)

Complementary Experiment and Modeling of HFCVD BEN-BEG of UNCD Films.

The mechanism proposed for explaining the growth of UNCD film by HFCVD BEN/BEG on W-covered Si wafers can be seen in Figure 1.39. The initial step, according to this figure, is the formation of a plasma between the filaments and the substrate and substrate holder at a bias voltage of ~200 V, applied between the positively biased filaments and the negatively biased substrate. Figure 1.39a shows that the plasma covers the entire wafer surface. However, Figure 1.39b shows that there is an electric field–enhanced concentration at the relatively sharp substrate holder edges, inducing higher amounts of CH_x⁺ and C⁺ ions striking the edge of the wafer, initially inducing sub-plantation of the ions on the W layer to induce formation of the W₂C/WC nucleation sites as a precursor for the formation of diamond nanocrystals. Figure 1.39c represents the actual nucleation and growth of the diamond nanocrystals, such that as the W surface layer gets transformed to W₂C/WC, the nucleation and growth of the diamond cluster propagates toward the center. The WC layer then spreads across the substrate surface from edge to center with different crystal orientations until enough nuclei of W (001) and (Reference Collins, Lightowlers and Field101) are grown to induce formation of diamond (111), based on the fact that the WC and diamond crystals have similar lattice parameters and low activation energy of 1.39 eV [Reference Koeck and Nemanich183]. The optimum time determined for the HFCVD BEN-BEG of UNCD film on a W surface is ~2.5 h, based on the observation that UNCD films produced for this growth time are very uniform, as shown in Figure 1.39d, revealing a fully coated 100 mm diameter W-coated Si substrate with a ~200 nm thick UNCD film.

Figure 1.39

(a) Overall schematic of electric field lines across filaments over the W-coated 100 mm diameter Si substrate. (b) Schematic showing the BEN-BEG process inducing WC formation on the surface of the W layer grown on the Si substrate and Si–W reaction areas at the Si–W layer interface. (c) Model representation of plasma–electric field lines inducing diamond nuclei formation from the substrate edge toward the center, due mainly to the initial electric field concentration at the W-coated Si substrate edges during the BEN-BEG growth process. (d) Picture of uniform BEN-BEG UNCD film across 100 mm W-coated Si substrate after 2.5 h of BEN-BEG followed by 2 h of growth without bias.

(reprinted from Diam. Relat. Mater., vol. 78, p. 1, 2017 (Fig. 10) in [135] with permission from Elsevier Publisher)

The HFCVD BEN-BEG growth process to grow UNCD films on Si and other substrates opened the path to new technological applications of UNCD films grown by this unique process, as described in the next section.

Selective Growth of UNCD Films on W-Patterned Layers on a Si Substrate.

The electric fields generated between the filaments and the W-coated Si substrate during growth of UNCD films by the HFCVD BEN-BEG process can be used to induce selective growth of diamond films on patterned metallic layers grown on Si or other semiconducting or insulating substrates to produce UNCD film-based patterned devices. This process was demonstrated by producing BEN-BEG of UNCD films on W film–patterned lines on Si substrates. Figure 1.40a shows a BEN-BEG UNCD film grown on patterned W lines resembling a diamond grown on an SiO₂/Si substrate. Figure 1.40b shows the plasma concentrated on the W lines during the HFCVD BEN-BEG of UNCD film selectively grown on the electrically conductive W lines grown on an insulating SiO₂/Si substrate. This experiment demonstrates the feasibility of using the HFCVD BEN-BEG technique for growing UNCD and other diamond films selectively on W-patterned layers on Si-based substrates.

Figure 1.40

(a) Optical picture of UNCD film grown selectively on electrically conductive W-patterned lines with a diamond shape, grown on an SiO₂/Si substrate exposed to 2.0 h of HFCVD BEN-BEG with additional growth time of 2 h without bias. (b) Picture taken during the HFCVD BEN-BEG UNCD film growth process, showing the plasma concentrated on the electrically conductive W lines, connected to the electrically conductive substrate holder with thin wire (compare this picture with the one shown in Figure 1.31a, which shows the bright plasma uniformly distributed across the whole W-coated Si substrate).

(reprinted from Diam. Relat. Mater., vol. 78, p. 1, 2017 (Fig. 12) in [135] with permission from Elsevier Publisher)