Hard-hydrogenated tetrahedral amorphous carbon films by distributed electron cyclotron resonance plasma

International Journal of Refractory Metals & Hard Materials 24 (2006) 39–48 www.elsevier.com/locate/ijrmhm

Hard-hydrogenated tetrahedral amorphous carbon films by distributed electron cyclotron resonance plasma F. Piazza

*

Department of Physics, Faculty of Natural Sciences-Phase II, University of Puerto Rico, Office c-302, P.O. Box 23343, San Juan, PR 00931, USA Received 11 November 2004; accepted 12 July 2005

Abstract Hard-hydrogenated tetrahedral amorphous carbon films (ta-C:H) are deposited from acetylene-fed distributed electron cyclotron resonance (DECR) plasma over large area (300 mm diameter disk) at near room temperature (below 140
C). The effects of the ion flux and energy on the structure and physical properties are investigated. For a constant substrate bias V0 of
150 V, the mass–density, YoungÕs modulus and hardness reach a maximum value of 2.5 g/cm3, 280 GPa, and 45 GPa, respectively, and the hydrogen content reaches a minimum of 26 at.% at the maximum ion flux /+ of 6.3 · 1015 ions cm
2 s
1. For a constant ion flux and pressure, the mass–density and YoungÕs modulus reach a maximum at a substrate bias of
300 V, and the hydrogen content is minimised. Electron diffraction, and Raman spectra show that the films grown at the maximum ion flux and a negative substrate bias ranging between 150 and 500 V are ta-C:H. The films contain sp2-carbon clusters and chains. sp2-carbon clustering increases with the increase of the substrate bias and decreases with the increase of the ion flux. The disorder increases with the ion flux and decreases with the bias. The optical band-gap decreases with disorder and with sp2-carbon clustering. It depends primarily on disorder rather than on clustering.  2005 Elsevier Ltd. All rights reserved. Keywords: Tetrahedral amorphous carbon; Diamond-like carbon; DECR; Hardness; Raman spectroscopy; Band-gap

1. Introduction There is great interest in amorphous carbon films, hydrogenated (a-C:H) or not (a-C) containing a high fraction of sp3-bonds between carbon atoms [1]. The sp3-bonding between carbon atoms of a-C (and of aC:H) confers on it many of the beneficial properties of diamond, such as its high hardness and large YoungÕs modulus, low coefficient of friction, high wear resistance, smoothness, optical transparency in a wide range of wavelengths, chemical and biological inertness, low electron affinity, high electrical resistance, lack of magnetic response [1]. The films can be grown at near room temperature, and are much cheaper to produce than dia*

Tel.: +1 787 7640000x2602; fax: +1 787 7567717. E-mail address: [email protected]

0263-4368/$ - see front matter  2005 Elsevier Ltd. All rights reserved. doi:10.1016/j.ijrmhm.2005.07.004

mond. Therefore they have widespread applications, mainly as protective coatings in areas such as magnetic storage disks [2], optical windows, micro-electromechanical devices [1], optical storage discs [3,4], tools [5] and engines [6,7]. There are many forms of a-C (and a-C:H) films. The key parameters in such materials are: (1) the sp3-content; (2) the clustering of the sp2-phase; (3) the orientation of the sp2-phase; (4) the cross-sectional nano-structure; (5) eventually, the hydrogen content, and C–H bonding. The sp3-content alone mainly controls the elastic constants, but films with the same sp3- and hydrogen contents but different sp2-clustering, sp2-orientation or cross-sectional nano-structure can have different optical and electronic properties [1]. The bonding and properties of a-C:H fall into four regimes defined primarily by the energy of the ion

40

F. Piazza / International Journal of Refractory Metals & Hard Materials 24 (2006) 39–48

during the growth process [1]. At low energy, the films have large hydrogen and sp3-contents (40–50 at.% and up to 60 at.%, respectively). These films are called polymer-like carbon (PLC) as most of the sp3-bonds are C–H bonds. They are soft and have low mass–density (typically 1.2 to 1.6 g/cm3). At intermediate energy, the hydrogen content falls (30 at.%), the sp3-content is less, but the films are harder and have higher mass– density since the number of C–C sp3-bonds is higher. These films are called diamond-like carbon (DLC). When the sp3 C–C content is maximum (70– 85 at.%), but the hydrogen content is 20–30 at.%, the films are called hydrogenated tetrahedral amorphous carbon (ta-C:H). They are hard and their mass–density reaches 2.4 g/cm3. At high energy, the hydrogen content falls further and the bonding becomes mainly sp2-like. These films are called graphitic-like carbon (GLC). sp3-bonding is promoted by the deposition from a source of medium-energy ions [1]. At present, hydrogen-free ta-C with high sp3-content is deposited by mass selected ion beam, filtered cathodic vacuum arc or by pulsed laser deposition [8–14]. From an industrial point of view, it is desirable to achieve good mechanical properties in a-C:H films grown by plasma enhanced chemical vapour deposition (PECVD) from a hydrocarbon precursor. ta-C:H has been deposited by high plasma density techniques, such as the plasma beam source (PBS) [15] and electron cyclotron wave resonance source (ECWR) [16,17]. The maximum reported sp3-content and YoungÕs modulus are 75% [15,18] and 300 GPa [19] respectively, and a maximum mass–density of 2.4 g/cm3 [18,19]. Early claims of a mass–density of up to 2.9 g/cm3 for ta-C:H [15,16] were overestimated, due to an incorrect choice of the electron effective mass for mass–density evaluation via electron energy loss spectroscopy (EELS). Once the correct effective mass is taken into account the mass–density decreases to 2.4 g/cm3 [18]. Standard RF-PECVD a-C:H does not achieve such properties [1,20–23], due to the low degree of ionisation of the deposition flux, and dispersion of the ion energy. This paper reports on large area deposition and characterization of hard ta-C:H films from an alternative technique using distributed electron cyclotron resonance (DECR). The effects of the ion flux and energy on the structure and physical properties are studied.

2. Experimental The films were grown on Sih1 0 0i and polycarbonate (PC) substrates using a DECR plasma reactor described previously in details [24]. A micro-wave power PMW of 800 W at a frequency of 2.45 GHz was used together with C2H2 as precursor. The plasma pressure P (pressure during plasma operation) was varied from 0.1 to

1.1 mTorr (from 13.3 to 146.6 mPa). Torr will be used as pressure unit (1 Torr = 133.32 Pa). The negative bias V0 applied to the 300 mm diameter stainless steel substrate holder was varied in the range 25–600 V using a power supply operating at 13.56 MHz. The bias was kept constant during deposition by an automatic modulation of the RF power. The substrate holder was watercooled enabling near room temperature deposition. In the experimental conditions considered, the temperature at the surface of the substrate holder during deposition remains lower than 140
C, the PC glass temperature. No damage of the PC substrates was observed after deposition. Prior to deposition the silicon substrates were exposed to an Ar plasma cleaning for 5 min (PMW = 800 W, P = 0.5 mTorr, V0 =
50 V). The expected low dispersion of the ion energy [25], especially when C2H2 is used as precursor [15] in the range of pressure considered, and the possibility of controlling the ion flux by varying the pressure, confer to the DECR technique interesting advantages over other techniques to control the film structure [26]. The hydrogen content NH and thickness were determined from nuclear reaction analysis (NRA) [27] and spectroscopic ellipsometry (SE) [24,27], respectively. The density q was deduced from the SE and NRA data [24,27]. The mean bond length r1 and angle a were deduced from electron diffraction [28,29]. SE was used to estimate the optical band-gap [24,27]. Unpolarized Raman spectra were acquired at 244 and 514.5 nm using two different spectrometers. The UV spectra at 244 nm were excited using an intracavity, frequency-doubled Ar ion laser. They were collected using a Renishaw micro-Raman 1000 spectrometer on a 40 · objective with 244 nm filter, and an UV-enhanced charge-coupled device camera. The spectral resolution was of 4–6 cm
1. UV Raman measurements are prone to damage hydrogenated samples. To avoid damage the power on the sample was kept below 1 mW. The spectra were collected for 90 s while the samples were rotating at a high speed (>3000 rpm). This ensured no visible damage and no change of the spectral shape during the measurements. A Renishaw Raman microscope system 2000 was used to get the spectra at 514.5 nm (argon ion laser). The power was kept below 1 mW [28,29]. Fourier transform infrared spectroscopy (FTIR) measurements were performed using a Bio-Rad FTS40 APC spectrometer in the transmission mode to assess the C–H bonding [27–29]. The YoungÕs modulus E was estimated by laser induced surface acoustic waves (LISAW) [30]. The thickness of the films analyzed is between 70 and 190 nm. Optical emission spectroscopy (OES) was performed using a Jobin Yvon HR 320 monochromator to record the evolution of the Ha (656.3 nm), CH (432.4 nm) and C2 (512.9 and 516.5 nm) spectral lines with the plasma pressure.

F. Piazza / International Journal of Refractory Metals & Hard Materials 24 (2006) 39–48

The ion flux /+ was deduced from the RF generator power delivered for a specific substrate bias. The secondary electron emission is assumed to be negligible within the energy range investigated [31]. This approach has already been used previously [32]. In the case of Ar plasma at 0.4 mTorr, good agreement was found between the substrate current density estimated in this way and the Bohm current density [26]. The later was assessed taking into account the electronic temperature and plasma density values determined using a Langmuir probe. The expression of the Bohm current density given in Ref. [26] is valid for a non-collisional sheath. This is the case in the present experiment conditions where the pressure is within the range of 0.1–1.1 mTorr. We assume that no power loss occurs in the RF device impedance regulator. Finally, the error involved in the ion flux estimation is believed to be of 20%. The /+ value is representative of the amount of the predominant ions into the low pressure C2H2 plasma which are known to be (C2Hx)+ ions [15]. However a variable production of H+ and (C4Hx)+ ions with the pressure is expected to contribute to /+ [15].

3. Results and discussion 3.1. Effect of the ion flux on C and H incorporation Fig. 1 shows the evolution of the ion flux /+ and flux of the film forming carbon /C as a function of the C2H2 flow, F, or plasma pressure, P, for a constant substrate bias V0 =
150 V. /C is derived from the deposition rate and mass–density. Fig. 1 shows that /C first increases with the pressure, reaches a maximum when 0.4 6 P 6 0.5 mTorr (30 6 F 6 40 sccm) and then decreases with a further increase of the pressure. The ion flux /+ follows a similar trend. The variation of /+ with the pressure is

-1

carbon φ (10

+

4

C

φC

6

2 40 4

Hydrogen content (at.%)

Flux (species 1015 cm-2 s -1)

-2

15

φ

6 cm s )

0.1

tentatively explained by the simultaneous evolution of the plasma density. The plasma density first increases since the probability of ionising collisions increases with the pressure. Then it decreases since the probability for a collision to lead to ionisation decreases when the electronic temperature decreases. OES measurements show that the intensity of the CH (432.4 nm) and C2 (512.9 and 516.5 nm) spectral lines first increases with the pressure, reaches a maximum when 0.4 6 P 6 0.5 mTorr, and then decreases with a further increase of the pressure. We suppose that the evolution of the intensity of the spectral lines traduces the evolution of the plasma density, although line intensity is not systematically correlated to species concentration. The maximum of the carbon incorporation corresponds to the maximum of the ion flux and to the maximum of the intensity of the optical emission lines of the CH and C2 radicals. The carbon incorporation increases when the ion flux is increased. This result is in good agreement with the film formation models based on the adsorbed layer model [33–37]. It confirms that the ions participate to the growth, either directly by subplantation process or indirectly by inducing chemisorptions of the radicals physisorbed to the growing film surface. The relationship between ion flux and carbon incorporation is further shown in Fig. 2a where the evolution of /C is plotted as a function of /+ for films grown with F > 30 sccm. Fig. 2a shows that carbon incorporation is maximized for the highest ion flux values. However Fig. 2a shows that /C drops from 5.7 · 1015 to 2.7 · 1015 cm
2 s
1 while the ion flux remains constant /+1.8 · 1015 cm
2 s
1. This confirms that the ions are not the only species that contribute to the film growth as presented in the film formation models based on the adsorbed layer model [33–37].

a

Plasma pressure (mTorr) 0.9 0.3 0.5

8

41

2

b

35 30 25

0 0

20

40 60 C2 H2 flow (sccm)

80

100

Fig. 1. Ion flux /+ and flux of carbon to the surface that is adsorbed /C as a function of the C2H2 flow and plasma pressure (V0 =
150 V). The dotted lines are guides to the eyes.

0

2

4 15

6 -2

8

-1

Ion flux (10 ions cm s )

Fig. 2. Flux of carbon effectively incorporated into the film /C (a) and hydrogen content (b) as a function of the ion flux /+ (D P 30 sccm, V0 =
150 V). The dotted lines are guides to the eyes.

42

F. Piazza / International Journal of Refractory Metals & Hard Materials 24 (2006) 39–48

The evolution of the carbon incorporation with the ion flux is correlated to the simultaneously evolution of the hydrogen content NH (Fig. 2b). The correlation between NH and /+ was previously discussed [27]. Two regimes are distinguished depending on the /+ value [27]. At ion flux higher than the threshold value T  1.9 · 1015 cm
2 s
1 the hydrogen desorption rate is almost in dynamic equilibrium with the adsorption and subplantation rates. When the ion flux /+ is reduced below the threshold value the hydrogen adsorption is significantly enhanced. The hydrogen incorporation mechanism is related to the hydrogen ions subplantation, to the passivation of dangling bonds by hydrogen radicals and to the exothermic hydrogenation of double bonds. The ion flux threshold value T, above which the desorption is reduced, is interpreted as corresponding to a complete coverage of the growing film surface by the individual ion impact zones. The range of the ions for V0 =
150 V is ˚ and the deposition rate is lower than 8 A ˚ /s. R8A This implies that the desorption is characterized by a time scale of the order of several hundred ms [27]. The amount of hydrogen remaining within the material depends on the competition between the chemisorption, subplantation and desorption. This is strongly dependent on the ion energy for plasma pressures higher than 0.1 mTorr in the experimental conditions considered (Section 3.2.2). At 0.1 mTorr the hydrogen chemisorption is predominant due to a high production of hydrogen radicals, as witnessed by a maximum in the intensity of the Ha spectral line [27]. Fig. 2a and b shows that the evolution of /C with /+ is related to the corresponding evolution of NH. The drop of /C for the lowest values of /+ is related to a simultaneous increase in the hydrogen content. The correlation between /C and NH is further demonstrated in Fig. 3 where /C is plotted as a function of NH. The car-

7

5

-2

-1

φC (10 carbon cm s )

6

4

15

3 2 1 0 20

25 30 35 Hydrogen content (at. %)

40

Fig. 3. Flux of carbon atoms effectively incorporated into the film /C as a function of the hydrogen content (D P 30 sccm, V0 =
150 V). The dotted line is guide to the eyes.

bon incorporation decreases linearly with the increase of the hydrogen incorporation. This is consistent with an increasing passivation of dangling bonds by hydrogen radicals. In conclusion, carbon and hydrogen incorporation depend on the plasma pressure and ion flux. In our case, the densest and less hydrogenated films are obtained when 0.4 6 P 6 0.5 mTorr. In this pressure range the ion flux reaches a maximum. 3.2. Density, YoungÕs modulus and diamond-like character 3.2.1. Effect of the ion flux As shown in Section 3.1, although constant substrate bias was used (i.e. constant ion energy), a significant increase in the carbon incorporation with the ion flux is observed at ion flux /+  >2 · 1015 ions cm
2 s
1 (Fig. 2a). In the same time the hydrogen content remains constant (Fig. 2b). The mass–density increases from 1.6 to 2.5 g/cm3 when the ion flux increases from 2 · 1015 ions cm
2 s
1 to 6.3 · 1015 ions cm
2 s
1. The carbon incorporation is maximized and the hydrogen incorporation is minimized for the highest ion flux values. The density and YoungÕs modulus reach a maximum value of 2.5 g/cm3 and 280 GPa, respectively, and the hydrogen content reaches a minimum of 26 at.% at the maximum ion flux of 6.3 · 1015 ions cm
2 s
1. The corresponding deposition rate is of 30 nm/min. Hardness, H, is related empirically to the YoungÕs modulus, E, by H ¼ 0:16 E

ð1Þ

within the approximation presented in [1]. Therefore the hardness is estimated to be of 45 GPa at the maximum ion flux of 6.3 · 1015 ions cm
2 s
1. The significant increase in the density and YoungÕs modulus observed at ion flux /+  >2 · 1015 ions cm
2 s
1 for a constant substrate bias while the hydrogen content remains constant traduces an increase in the C–C sp3-fraction and, thus, the diamond-like character. This is confirmed by the broadening of the infrared C–H stretching band and the disappearance of well-defined components for films deposited with /+ > 2 · 1015 ions cm
2 s
1 (Fig. 4). In conclusion for constant substrate bias (ion energy) the diamond-like character is increased by increasing the ion flux. This result is proposed to result from the propagation and/or interference of shock waves generated by ion impacts during the growth [38,39]. When /+ is decreased below 2 · 1015 ions cm
2 s
1, the carbon incorporation decreases and hydrogen content increases (Section 3.1). The mass–density decreases from 1.6 to 1.2 ions g/cm3. This implies an increase in the C–H sp3 fraction rather than the C–C sp3, and we

F. Piazza / International Journal of Refractory Metals & Hard Materials 24 (2006) 39–48

43

30

Hydrogen content (at.%)

600 0.4 mTorr; V0= - 300 V 15 -2 -1 φ+=6×10 cm s

-1

α (cm )

400

a

25

20 2.5

Density (g/cm3)

200 0.1 mTorr; V0= - 400 V

3100

3050

-2 -1

cm s

3000 2950 2900 2850 -1 Wavenumbers (cm )

2.0

2800

Fig. 4. Evolution of the infrared C–H stretching band with the plasma pressure (ion flux).

expect the films to become more polymer-like. The FTIR analysis shows that the shape of the infrared C–H stretching bands for films deposited with /+ < 2 · 1015 ions cm
2 s
1 is typical from polymer-like carbon film with well-defined components (Fig. 4). In conclusion the diamond-like character is tuned by changing the ion flux for constant ion energy. When the ion flux is decreased below 2 · 1015 ions cm
2 s
1 the increase in sp3-fraction (C–H bonds) is due to an increase in the hydrogen incorporation triggered by a high production of hydrogen radicals at low pressure (Section 3.1). The cross-linking between carbon is low and the films are polymer-like. When the ion flux is increased above 2 · 1015 ions cm
2 s
1 the increase in sp3-fraction (C–C bonds) is due to ion energy loss and transfer process. The cross-linking between carbon increases and the films are diamond-like. 3.2.2. Effect of the ion energy The ion energy is a key parameter in carbon films growth [1]. We investigated its influence on the structure and physical properties of DECR films. Fig. 5 shows the evolution of the hydrogen content (Fig. 5a), mass–density (Fig. 5b) and YoungÕs modulus (Fig. 5c) with the substrate bias for films deposited at 0.5 mTorr (with /+  5.2 · 1015 ions cm
2 s
1). The results displayed in Fig. 4a have been previously discussed [27]. The hydrogen content, NH, depends on the competition between the chemisorption, subplantation and desorption. It is thus expected to be strongly dependent on the ion energy. Fig. 5a shows that the hydrogen content first decreases with increasing substrate bias, indicating that ion bombardment related effects (sputtering, out-diffusion) dominate over hydrogen incorporation. Then it remains constant, indicating a dynamic equilibrium between hydrogen incorporation and loss. Fig. 5b and c shows that for films deposited at P = 0.5 mTorr the density and YoungÕs modulus first

1.5

Young's modulus (GPa)

15

φ+=10

0

b

c

250

200

150 0

200

400

600

Negative substrate bias (V) Fig. 5. Hydrogen content (a), mass–density (b), and YoungÕs modulus (c), as a function of the negative substrate bias (plasma pressure P = 0.5 mTorr).

increase with jV0j, reach a maximum when jV0j  300 V and then slightly decrease with a further increase of the substrate bias. The maximum of density and YoungÕs modulus is reached when the carbon atoms carried by (C2Hx)+ ions undergo a subplantation process with 150 eV energy. This is in good agreement with previous results, and was interpreted by the subplantation model (see Ref. [1]). For 0.5 mTorr the maximum of density and YoungÕs modulus are of 2.3 g cm
3 and 250 GPa, respectively. The corresponding hardness value deduced from YoungÕs modulus is of 40 GPa. The density, YoungÕs modulus and hardness values obtained here are similar to those corresponding to PBS and ECWR deposited ta-H films [15–19]. The DECR technique is suitable for coating large areas, at high rate. The deposition rate for the hardest film is >30 nm/min, which is at least three times faster than in the PBS system [15]. In the present experimental conditions, the samples are located 27 cm far away from the bottom of the magnetic racetracks [24,26]. The deposition rate is expected to be further increased by moving the substrate holder closer to the magnetic racetrack. It is possible to use the refractive index at 633 nm as a rapid means of estimating the diamond-like character and mechanical properties for a-C:H films with a mass–density above 1.6 g/cm3. Fig. 6 shows the evolution of YoungÕs modulus (Fig. 6a) and refractive index at 633 nm (Fig. 6b) as a function of mass–density q for films deposited at different plasma pressures and

44

F. Piazza / International Journal of Refractory Metals & Hard Materials 24 (2006) 39–48 300 120 graphite

Bond angle (˚)

Young's modulus (GPa)

a

200

100

0.1 mTorr 0.5 mTorr 0.9 mTorr

0 2.8

117 ta-C:H

114

111

b

108 1.41

diamond 0.5 mTorr PBS; Weiler et al.

1.44

1.47

1.50

1.53

1.56

Bond length (Å)

n (at 633 nm)

2.6

Fig. 7. Mean bond angle–bond length diagram (P = 0.5 mTorr, 200 V < jV0j 6 500 V) including data for graphite and diamond (stars), and data corresponding to PBS deposited ta-C:H [15] (triangles).

2.4 2.2 2.0 1.8 1.0

1.2

1.4 1.6

1.8

2.0

2.2

2.4

2.6

duced at 0.5 mTorr with 200 V < jV0j 6 500 V are tetrahedral. The other films produced are located in another part of the diagram. This shows that the ion flux enhances the tetrahedral character.

Mass-density (g/cm3) Fig. 6. YoungÕs modulus (a), and optical index at 633 nm (b) as a function of mass–density.

substrate bias. Fig. 6b shows that the refractive index increases in a linear manner with mass–density for films with q  >1.6 g/cm3. Fig. 6a shows that YoungÕs modulus increases in a linear manner with mass–density for films with q  >1.6 g/cm3, in a similar way to that found for H-free carbons [19]. Therefore, the optical index at 633 nm can be used to estimate the YoungÕs modulus, and diamond-like character. 3.3. Structure 3.3.1. Electron diffraction Electron diffraction was used to have a closer look on the structure. The technique was used to estimate the mean bond angle, a, and bond length, r1. Fig. 7 shows an a–r1 diagram where the data obtained for the films deposited at 0.5 mTorr with 200 V < jV0j 6 500 V are compared with the values for graphite and diamond, and data corresponding to PBS deposited ta-C:H [15]. Fig. 7 shows that the films produced here are character˚ ), very close ized by large bond length values (a P 1.50 A to the bond length in diamond, and bond angle values closer to the value in diamond than in graphite (r1 < 116.16
). The bond angle values obtained here are comparable to the values reported for PBS deposited ta-C:H films with 75% sp3-carbon fraction [15]. The bond length values obtained here are larger. This shows that the structure of the films produced at 0.5 mTorr with 200 V < jV0j 6 500 V is close to the structure of ta-C:H obtained by PBS. It confirms that the films pro-

3.3.2. Resonant Raman spectroscopy Resonant Raman spectroscopy is a useful technique to monitor the carbon bonding. Raman scattering is a resonant process in which the configurations whose band-gap matches the excitation energy are preferentially excited. For visible energies, the Raman scattering of sp2-carbon is a resonant process between 50 and 230 times more effective than that of sp3-carbon [40]. UV Raman gives a more evenly weighted probe of sp3 and sp2-sites, and of sp2-clusters and chains. All carbon films show common features in their Raman spectra in the 800–2000 cm
1 region, the so called G and D bands, which lie at 1560 and 1360 cm
1 respectively for visible excitation, and the T band at 1060 cm
1, which is visible only with UV excitation [41]. The G band is due to the bond stretching of all pairs of sp2-carbon atoms in both rings and chains. The D band is due to the breathing modes of sp2-carbon atoms in rings [42]. The T band is due to the sp3 C–C vibrations [43]. Two main factors determine the band shapes: the sp3-content and the arrangement of sp2-carbon (clusters and chains). The arrangement of sp2-carbon can vary independently from the sp3-content, so that for particular sp3-content and excitation energy, we can have a number of different Raman spectra or, equivalently, similar Raman spectra for different sp3contents. A multi-wavelength Raman analysis is thus needed to fully characterize the samples. Two useful parameters, derived by resonant Raman studies, are the G band full width at half maximum (FWHM), and G band dispersion. Both are related to the degree of disorder. The G band FWHM always increases as the disorder increases, at every excitation wavelength and for every type of carbon [41]. At fixed

F. Piazza / International Journal of Refractory Metals & Hard Materials 24 (2006) 39–48

Intensity (a.u.)

G band

a

D band 514.5 nm T band

244 nm

G band

b

Intensity (a.u.)

D band

514.5 nm

244 nm 500

750

1000

1250

1500

1750

Raman shift (cm-1) Fig. 8. Typical visible (at 514.5 nm) and UV Raman spectra of films deposited at 0.5 mTorr (V0 =
300 V) (a), and at 0.1 mTorr (V0 =
400 V) (b).

G band dispersion (cm-1/nm)

0.4

a

0.3

0.2

0.1

0.1 mTorr 0.5 mTorr

0.0

b 1620

G band position (cm-1)

excitation energy, ta-C has the largest G band FWHM, as it has the largest disorder. On the other hand polymeric a-C:H has a very narrow G band, consistent with the low disorder in this materials deposited at low ion energies [41]. The G band only disperses in disordered carbon because of the presence of a range of configurations with different local band-gap and different phonon modes [41]. The dispersion arises from a resonant selection of sp2-configurations or clusters with wider p band gaps, and correspondingly higher vibration frequencies [41]. The dispersion is proportional to the degree of disorder [41]. The G band of graphite and nano-crystalline (nc) graphite does not disperse, while ta-C presents the higher dispersion (0.45 cm
1/nm). In materials with only sp2-clusters, the UV G band position saturates at a maximum of 1600 cm
1, the position in nc-graphite. In contrast, in materials also containing sp2-chains, the UV G band rises past 1600 cm
1 [41]. The high G band position is due to short, strained double bonds chains. Thus UV Raman gives a weighted probe of rings and chains. It is not biased towards sp2-configurations of lower band gap as visible Raman. In UV excitation, increasing clustering lowers the G band position. Fig. 8a shows characteristic visible (at 514.5 nm) and UV (at 244 nm) Raman spectra of films deposited at 0.5 mTorr (V0 =
300 V). As a comparison the corresponding spectra of films deposited at 0.1 mTorr (V0 =
400 V) are displayed in Fig. 8b. Fig. 9 shows the evolution of the G band dispersion (a), and UV position (b) as a function of the negative substrate bias for films deposited at 0.1 and 0.5 mTorr. The spectra were fitted using a combination of a Breit–Wigner–Fano (BWF) and a Lorentzian for respectively the G and D band. A detailed analysis of the Raman spectra will be published elsewhere [44]. The spectra displayed in Fig. 8a are characteristic of ta-C:H [41,45]. In particular, in the UV, there is no clear dip at 1500 cm
1 between the D and G band, and a weak T band can be observed. In addition the UV G band position is higher than 1600 cm
1 (Fig. 9b) while at the same time its width and dispersion are large (Fig. 9a). The UV G band position, FWHM and dispersion vary from 1617 to 1600 cm
1 (Fig. 9b), 180 to 156 cm
1, and 0.29 to 0.20 cm
1/nm (Fig. 9a), respectively, in the range of bias considered. Those features distinguish ta-C:H from other form of a-C:H. As a comparison, the spectra displayed in Fig. 8b are characteristic of PLC [41,45]. In the UV spectra there is a clear dip at 1500 cm
1 between the D and G band. The T band is very weak. The UV G band position is still higher than 1600 cm
1 as for ta-C:H (Fig. 9b) but at the same time its width (91 cm
1) and dispersion are low (0.1 cm
1/nm) (Fig. 9a). Beside the information on the tetrahedral character of the films, multi-wavelength Raman spectroscopy

45

1610

1600

1590

0

200 400 600 Negative substrate bias (V)

Fig. 9. G band dispersion (a) and position in the UV (b) for films deposited at 0.5 mTorr (circles) and at 0.1 mTorr (squares).

F. Piazza / International Journal of Refractory Metals & Hard Materials 24 (2006) 39–48

3.4. Optical band-gap For films deposited at pressure P > 0.1 mTorr the optical absorption has very broad tail at low photon energies (inset of Fig. 10a). The optical band-gap E04 was deduced from the optical absorption spectra (film thickness of 150 nm). E04 is defined as the energy at which the optical-absorption coefficient reaches 104 cm
1. It is generally regarded as a reliable value of the band-gap for a situation of very broad absorption edges, such as here. For the most absorbent film E04 was estimated from a parabolic fit. In carbon films the optical band-gap is determined by the gap between the p and p* states of the sp2-sites as these lie closest to the Fermi level [46]. It was first predicted that the gap depends on the sp2-clusters sizes distribution [46]. The local band-gap associated to a cluster varies inversely with its size. Thus the greater clusters determine the gap. Later, simulation results [47–51], and neutron diffraction experiments [52,53] showed that the clusters size is in fact smaller that the value expected

a

E04 (eV)

2.0

5.5

1.5

0.5 mTorr

5.0

0.9 mTorr

4.5 4.0 1

2

1.0

3 4 E(eV)

0.5 0.5 mTorr 0.9 mTorr

0.0

2.5

0

200

-V0 (V)

400

600

2.0

2.5

b

2.0

E04 (eV)

1. The films produced here contain sp2-clusters and chains because the UV G peak position is always higher than 1600 cm
1 (Fig. 9b). 2. The increase of the ion flux induces a decrease in clustering because the G position increases with the ion flux. This is consistent with the electron diffraction results that show that the tetrahedral character increases with the ion flux (Section 3.3.1). 3. In an opposite way, the increase of the negative substrate bias induces an increase in clustering because the G position decreases with the negative substrate bias (Fig. 9b). 4. An increase in the hydrogen incorporation reduces clustering because the G band position increases with hydrogen content. 5. The films deposited at 0.5 mTorr (with /+  5.2 · 1015 ions cm
2 s
1) and 0.9 mTorr (with /+  2 · 1015 ions cm
2 s
1) contain more disorder than the low-density (1.2 g/cm3) films deposited at 0.1 mTorr (with /+  2 · 1015 ions cm
2 s
1) because the FWHM and dispersion of the G band is higher at 0.5 and 0.9 mTorr. We believe that this is due to higher ion bombardment at 0.5 mTorr, and to lower hydrogen radicals flux at 0.9 mTorr compared to 0.1 mTorr (Section 3.1). 6. The disorder decreases with the increase of the negative substrate bias because the G band dispersion decreases in this condition (Fig. 9a).

2.5

-1

can be used to get information concerning the arrangements of sp2-carbon atoms. From a detailed analysis of the spectra [44] the following conclusions can be drawn:

log(α (cm ))

46

1.5 1.0 0.5

0.5 mTorr 0.9 mTorr Kim et al.

0.0 1.0

1.5

Density (g/cm3) Fig. 10. Optical band-gap E04 as a function of the negative substrate bias (a), and as a function of mass–density (b), including data from Kim and Grotjohn [56]. The dotted lines are guides to the eyes. The inset is the optical absorption spectra of films deposited with V0 =
300 V.

from the relationship between gap and cluster size. This was explained by the competition between sp2-carbon clustering on one hand and disorder induced by the intense ionic bombardment characteristic of most of the deposition process, on the other hand [54,55]. The local gap corresponding to a given sp2-chain or cluster decreases with distortions [54,55]. Therefore the gap is first determined by the distortions of the clusters and/or chains introduced by the high bombardment and stress rather than by the cluster sizes distribution. Fig. 10 shows the evolution of E04 with the substrate bias (a), and with mass–density (b), including data from Ref. [56]. Fig. 10a shows that for a given pressure E04 remains constant for jV0j 6 200 V and decreases when the negative substrate bias is increased from 200 to 600 V. Note that the hydrogen remains constant when the bias is increased from 200 to 600 V (Fig. 5a). This is explained by an increase of the clustering as confirmed by Raman spectroscopy analysis (Section 3.3). Whatever the substrate bias considered, films deposited at 0.5 mTorr with the highest ion flux present the lowest band-gap. This is explained by higher disorder and distortions because of higher ion flux, as confirmed by Raman spectroscopy analysis (Section 3.3). Fig. 10b shows that for a given pressure, E04 starts to shrink when

F. Piazza / International Journal of Refractory Metals & Hard Materials 24 (2006) 39–48

the mass–density increases above 2 g/cm3. This trend was previously observed [56]. For a given mass–density the gap is lower for films deposited at 0.5 mTorr than at 0.9 mTorr. This shows that there is no clear relationship between gap and mass–density. It shows that films of the same density can have different band-gap because they have been obtained using different ion bombardment, leading to variation in clustering and distortions. Since the films deposited at 0.5 mTorr contain more disorder and less clustering than the films deposited at 0.9 mTorr, the fact that the gap is lower for films deposited at 0.5 than at 0.9 mTorr shows that the gap is primarily determined by distortions rather than by clustering.

4. Conclusions ta-C-:H films with up to 2.5 g/cm3 mass–density, 280 GPa YoungÕs modulus, and 45 GPa hardness have been deposited from acetylene-fed distributed electron cyclotron resonance (DECR) plasma at a deposition rate of 30 nm/min over large area (300 mm diameter disks) at substrate temperature below 140
C. The tetrahedral character of the films was confirmed by electron diffraction and multi-wavelength Raman spectroscopy. The films structure and physical properties depend strongly on the ion flux and energy. For a given ion flux the films properties (diamond-like character) are optimum at a substrate bias of
300 V or 150 eV per carbon atom coming from a C2 Hþ x ion. The result is consistent with the subplantation model. For a given substrate bias, the diamond-like character improves for increasing ion flux above of 2 · 1015 ions cm
2 s
1. The results are interpreted in terms of the propagation and/or interference of shock waves generated by ion impacts during the growth. The films contain sp2-carbon clusters and chains. sp2-carbon clustering increases with the increase of the substrate bias and decreases with the increase of the ion flux. The disorder increases with the ion flux and decreases with the bias. The optical bandgap decreases with disorder and with sp2-carbon clustering. It depends primarily on disorder rather than on clustering.

Acknowledgments The authors would like to thank D. Grambole and F. Herrmann (Forschungszentrum Rossendorf, Germany) for assistance in obtaining the NRA data; D.A.M Smith and D. Batchelder (Leeds University, UK) for access to UV Raman facilities; G. Relihan (NMRC, Ireland) for performing spectroscopic ellipsometry analysis; D. Schneider (Fraunhofer Institute for Material and Beam Technology IWS, Germany) and S. Schulze (Chemnitz University of Technology, Germany) for LISAW

47

and electron diffraction measurements, respectively; Y. Arnal and J. Pelletier for assistance in obtaining the OES data and helpful discussions (CNRS, France); A. Golanski (CNRS, France) for interesting and helpful discussions; J. Robertson and A.C. Ferrari (Cambridge University, UK) for logistic assistance in materials characterization. This work was supported by the European Community (DIAMCO; Brite-EuRam Contract No. BRPR-CT98-0749).

References [1] Robertson J. Diamond-like amorphous carbon. Mater Sci Eng R 2002;37:129–281. [2] Ferrari AC. Diamond-like carbon for magnetic storage disks. Surf Coat Technol 2004;180–181:190–206. [3] Piazza F, Grambole D, Zhou L, Talke F, Casiraghi C, Ferrari AC, et al. Large area deposition of hydrogenated amorphous carbon films for optical storage disks. Diamond Relat Mater 2004;13:1505–10. [4] Piazza F, Grambole D, Schneider D, Casiraghi C, Ferrari AC, Robertson J. Protective diamond-like carbon coatings for future optical storage disks. Diamond Relat Mater 2005;14:994–9. [5] Hakovirta M. Hardmetal woodcutting tool tips coated with tetrahedral amorphous carbon. Diamond Relat Mater 1999;8: 1225–8. [6] Malaczynski GW, Hamdi AH, Elmoursi AA, Qiu X. Diamondlike carbon coating for aluminum 390 alloy—automotive applications. Surf Coat Technol 1997;93:280–6. [7] Wu L, Holloway BC, Beesabathina DP, Kalil C, Manos DM. Analysis of diamond-like carbon and Ti/MoS2 coatings on Ti– 6Al–4V substrates for applicability to turbine engine applications. Surf Coat Technol 2000;130:207–17. [8] Ishikawa J, Takeiri Y, Ogawa K, Takagi T. Preparation and structure of carbon film deposited by a mass-separated C + ion beam. J Appl Phys 1987;55:188–93. [9] Lifshitz Y, Kasi SR, Rabalais JW. Subplantation model for film growth from hyperthermal species: application to diamond. Phys Rev Lett 1989;62:1290–3. [10] McKenzie DR, Muller D, Pailthorpe BA. Compressive-stressinduced formation of thin-film tetrahedral amorphous carbon. Phys Rev Lett 1991;67:773–6. [11] Lossy R, Pappas DL, Roy PA, Cuomo JJ, Sura VM. Filtered arc deposition of amorphous diamond. Appl Phys Lett 1992;61: 171–3. [12] Fallon PJ, Veerasamy VS, Davis CA, Robertson J, Amaratunga G, Milne WI, et al. Properties of filtered-ion-beam-deposited diamondlike carbon as a function of ion energy. Phys Rev B 1993;48:4777–82. [13] Pappas DL, Saenger KL, Bruley J, Krakow W, Cuomo JJ, Gu T, et al. Pulsed laser deposition of diamond-like carbon films. J Appl Phys 1992;71:5675–84. [14] Xiong F, Chang YY, Chang RPH. Complex dielectric function of amorphous diamond films deposited by pulsed-excimer-laser ablation of graphite. Phys Rev B 1993;48:8016–23. [15] Weiler M, Sattel S, Giessen T, Jung K, Ehrhardt H, Veerasamy VS, et al. Preparation and properties of highly tetrahedral hydrogenated amorphous carbon. Phys Rev B 1996;53:1594–608. [16] Weiler M, Lang K, Li E, Robertson J. Deposition of tetrahedral hydrogenated amorphous carbon using a novel electron cyclotron wave resonance reactor. Appl Phys Lett 1998;72:1314–6. [17] Morrison NA, Muhl S, Rodil SE, Ferrari AC, Nesladek M, Milne WI, et al. The preparation, characterization and tribological

48

[18]

[19]

[20]

[21]

[22] [23]

[24]

[25]

[26] [27]

[28]

[29]

[30]

[31] [32] [33] [34]

[35]

[36]

[37]

F. Piazza / International Journal of Refractory Metals & Hard Materials 24 (2006) 39–48 properties of TA-C:H deposited using an electron cyclotron wave resonance plasma beam source. Phys Status Solidi (a) 1999;172: 79–90. Ferrari AC, Libassi A, Tanner BK, Stolojan V, Yuan J, Brown LM, et al. Density, sp3 fraction, and cross-sectional structure of amorphous carbon films determined by X-ray reflectivity and electron energy-loss spectroscopy. Phys Rev B 2000;62: 11089–103. Ferrari AC, Robertson J, Beghi MG, Bottani CE, Ferulano R, Pastorelli R. Elastic constants of tetrahedral amorphous carbon films by surface Brillouin scattering. Appl Phys Lett 1999;75: 1893–5. Koidl P, Wild C, Dischler B, Wagner J, Ramsteiner M. Properties and characterization of amorphous carbon films. Mater Sci Forum 1990;52:41. Tamor MA, Vassell WC, Carduner KR. Atomic constraint in hydrogenated ‘‘diamond-like’’ carbon. Appl Phys Lett 1991;58: 592–5. Zhou JW, Reichelt K, Schmidt K, Dischler B. The deposition and study of hard carbon films. J Appl Phys 1989;65:3914–8. Mo¨ller W. Plasma and surface modelling of the deposition of hydrogenated carbon-films from low pressure methane plasmas. Appl Phys A-Mater 1993;56:527–46. Golanski A, Piazza F, Werckmann J, Relihan G, Schulze S. Stress field effects on the microstructure and properties of a-C:H thin films. J Appl Phys 2002;92:3662–70. Lagarde T, Pelletier J, Arnal Y. study of the etching of SiO2 in SF6 plasmas: modeling of the etching kinetics and validation. Plasma Sources Sci Technol B 1999;17:118–26. Piazza F. The`se de Doctorat de lÕUniversite´ Louis Pasteur, Strasbourg, France, 2001. Piazza F, Grambole D, Herrmann F, Relihan G, Barthe MF, Desgardin P, et al. Incorporation of hydrogen and oxygen into (t)a-C:H thin films deposited using DECR plasma. Mater Res Soc Symp Proc 2001;675:W10.3.1. Piazza F, Golanski A, Schulze S, Relihan G. Transpolyacetylene chains in hydrogenated amorphous carbon films free of nanocrystalline diamond. Appl Phys Lett 2003;82:358–60. Piazza F, Schulze S, Relihan G, Golanski A. Transpolyacetylene chains in DECR plasma deposited a-C:H films. Diamond Relat Mater 2003;12:942–5. Schneider D, Schwartz T, Scheibe HJ, Panzner M. Non-destructive evaluation of diamond and diamond-like carbon films by laser induced surface acoustic waves. Thin Solid Films 1997;295: 107–16. Benninghoven A, Ru¨denauer FG, Werner HW. Secondary ion mass spectroscopy. New York: Wiley-Interscience, 1994. Keller JH, Pennebaker WB. Electrical properties of RF sputtering systems. IBM J Res Dev 1979;23:3–15. Jacob W. Surface reactions during growth and erosion of hydrocarbon films. Thin Solid Films 1998;326:1–42. Reinke P, Jacob W, Mo¨ller W. Influence of the ion energy on the growth and structure of thin hydrocarbon films. J Appl Phys Lett 1993;74:1354–61. Von Keudell A, Jacob W. Growth and erosion of hydrocarbon films investigated by in situ ellipsometry. J Appl Phys 1996;79: 1092–8. Mo¨ller W, Fukarek W, Lange K, von Keudell A, Jacob W. Mechanisms of the deposition of hydrogenated carbon films. Jpn J Appl Phys 1995;34:2163. Godet C, Heitz T, Boure´e JE, Drevillon B. Growth and composition of dual-plasma polymer-like amorphous carbon films. J Appl Phys Lett 1998;84:3919–32.

[38] Golanski A, Stoquert JP, Piazza F, Kern P, Laemmer E, Deruyver A, et al. Hypersonic shock waves and hybridization of a-C:H thin films. J Appl Phys 2003;93:5911–9. [39] Golanski A, Stoquert JP, Ken P, Piazza F, Schulze S. Impact of energy density and stress fields on the nucleation dynamics of plasma deposited a-C:H films. Nucl Instrum Meth Phys Res B 2003;206:731–5. [40] Wagner J, Wild C, Koidl P. Resonance effects in Raman scattering from polycrystalline diamond films. Appl Phys Lett 1991;59:779–81. [41] Ferrari AC, Robertson J. Interpretation of Raman spectra of disordered and amorphous carbon. Phys Rev 2000;61:14095–107; Origin of the 1150 cm
1 Raman mode in nanocrystalline diamond. Phys Rev 2001;63:121405–9; Resonant Raman spectroscopy of disordered, amorphous, and diamondlike carbon. Phys Rev 2001;64:075414–27. [42] Tuinstra F, Koening JL. Raman spectrum of graphite. J Chem Phys 1970;53:1126–30. [43] Gilkes KWR, Prawer S, Nugent KW, Robertson J, Sands HS, Lifshitz Y, et al. Direct quantitative detection of the sp3 bonding in diamond-like carbon films using ultraviolet and visible Raman spectroscopy. J Appl Phys 2000;87:7283–9. [44] Piazza F, Schneider S, Casiraghi C, Ferrari AC, Robertson J. Hydrogenated tetrahedral amorphous carbon films by distributed electron cyclotron resonance plasma. J Appl Phys, submitted for publication. [45] Casiraghi C, Piazza F, Ferrari AC, Grambole D, Robertson J. Bonding in hydrogenated diamond-like carbon by Raman spectroscopy. Diamond Relat Mater 2005;14:1098–102. [46] Robertson J, OÕReilly EP. Electronic and atomic structure of amorphous carbon. Phys Rev B 1987;35:2946–57. [47] Frauenheim T, Blaudeck P, Stephan U, Jungnickel G. Atomic structure and physical properties of amorphous carbon and its hydrogenated analogs. Phys Rev B 1993;48:4823–34; Molecular dynamic investigations of amorphous carbon: p bonding vs. electronic defect generation. Diamond Relat Mater 1994;3:462–9. [48] Galli G, Martin RM, Car R, Parinello M. Structural and electronic properties of amorphous carbon. Phys Rev Lett 1989; 62:555–8. [49] Wang CZ, Ho KM, Chan CT. Tight-binding molecular-dynamics study of amorphous carbon. Phys Rev Lett 1993;70:611–4. [50] Wang CZ, Ho KM. Structure, dynamics, and electronic properties of diamondlike amorphous carbon. Phys Rev Lett 1993;71: 1184–7; The electronic structure of diamond-like amorphous carbon. J Phys Condens Matter 1994;6:L239–44. [51] Drabold DA, Fedders PA, Strumm P. Theory of diamondlike amorphous carbon. Phys Rev B 1994;49:16415–22. [52] Li F, Lannin JS. Radial distribution function of amorphous carbon. Phys Rev Lett 1991;65:1905–8. [53] Burke TM, Newport RJ, Howells WS, Gilkes KWR, Gaskell PH. The structure of a-C:H(D) by neutron diffraction and isotropic enrichment. J Non-Cryst Solids 1993;164:1139–42. [54] Lee CH, Lambrecht WRL, Segall B, Kelires PC, Frauenheim T, Stephan U. Electronic structure of dense amorphous carbon. Phys Rev B 1994;49:11448–51. [55] Robertson J. Structural models of a-C and a-C:H. Diamond Relat Mater 1995;4:297–301. [56] Kim BK, Grotjohn TA. Comparison of a-C:H films deposited from methane–argon and acetylene–argon mixtures by electron cyclotron resonance–chemical vapor deposition discharges. Diamond Relat Mater 2000;9:37–47.