Structural and micro-mechanical studies of CNx films deposited on silicon substrates in a remote nitrogen plasma

Surface and Coatings Technology 116–119 (1999) 59–64 www.elsevier.nl/locate/surfcoat

Structural and micro-mechanical studies of CN films deposited on x silicon substrates in a remote nitrogen plasma C. Jama a, *, O. Dessaux a, P. Goudmand a, J.-M. Soro b, D. Rats b, J. von Stebut b a Laboratoire de Ge´nie des Proce´de´s d’Interactions Fluides Re´actifs Mate´riaux-UPRES-EA no. 2698, Universite´ de Lille 1, Cite´ Scientifique, F-59655 Villeneuve d’Ascq, France b LSGS, UMR Cnrs-inpl-edf 7570, CIM-Ecole des Mines, F-54042 Nancy Cedex, France

Abstract This article reports on the synthesis of carbon nitride CN films on a non-heated substrate. The carbon nitride films are x deposited using pulsed CO laser ablation of carbon molecular fragments from a graphite target under nitrogen plasma afterglow. 2 Characterizations of carbon nitride films by Raman, XPS, FT-IR spectroscopy and micro-mechanical testing techniques are presented. XPS analyses clearly demonstrate that the film structure consists of carbon fragments in a graphitic form and in CN bonding in sp2 trigonal and sp3 tetrahedral forms. It also shows that higher N/C ratios with a higher CN sp3 concentration are obtained when the distance between the discharge and the deposition zone (d ) decreases. The aim of this work is to determine whether there is a relationship between the reactive nitrogen plasma afterglow species and film structure. © 1999 Elsevier Science S.A. All rights reserved. Keywords: Mechanical properties; Nitrogen plasma; Raman; XPS

1. Introduction The characterization of ultra-thin tribological hard protective coatings is an important concern for the Draconian demands of future requirements in several industrial applications and particularly in the field of electronics. For such applications, thin protective films (roughly 1 mm in thickness) that possess a high hardness, low friction coefficient, high wear resistance and low surface roughness are needed. Carbon nitride has recently attracted attention due to its high hardness, elastic recovery, and wear resistance [1,2]. Most of the techniques that have been used to synthesize carbon nitride, CN , are based on condensation of a thin film x onto a substrate. The type of CN films frequently x reported have a N/C ratio between 0.2 and 0.5 and are amorphous or partly crystalline. The present article reports on the micro-mechanical and chemical structure characterizations of carbon nitride films using TEA CO laser ablation of a graphite target under nitrogen 2 plasma afterglow conditions. Sputtered carbon frag* Corresponding author. Tel.: +33-3-20-33-62-99; fax: +33-3-20-43-41-58. E-mail address: [email protected] (C. Jama)

ments combined with reactive nitrogen plasma afterglow species lead to the formation of a film on the substrate. Nitrogen plasma discharges are known to generate two distinct spatial zones downstream from the power input region [3] ( Fig. 1(a)): The near afterglow or short-lived afterglow (SLA) zone, characterized by a pink luminescence. The reactive species are ions, electrons, short-lived particles, nitrogen atoms, as well as electronically and vibrationally excited nitrogen molecules. The far afterglow or cold remote nitrogen plasma (CRNP) zone is a non-ionized reactive zone. As in the SLA zone, electronically and vibrationally excited nitrogen molecules contribute to the reactive atmosphere. However, the main reactive species is made up of atomic nitrogen. Characterizations of carbon nitride films by Raman, X-ray photoelectron spectroscopy ( XPS), Fourier transform infra-red ( FTIR) spectroscopy and micromechanical testing techniques are presented. The micro-mechanical characterizations aimed to establish whether there is any relationship between the reactive nitrogen plasma afterglow species, film composition and structure and subsequently correlated to the film’s overall mechanical properties.

0257-8972/99/$ – see front matter © 1999 Elsevier Science S.A. All rights reserved. PII: S0 2 5 7- 8 9 7 2 ( 9 9 ) 0 0 06 9 - 9

60

C. Jama et al. / Surface and Coatings Technology 116–119 (1999) 59–64

(a)

(b)

Fig. 1. (a) Nitrogen post-discharge afterglow. (b) Experimental set-up for CN deposition. x

2. Experimental

lived afterglow (SLA), whereas for d=0.9 m, they were subjected to the far afterglow (CRNP) condition.

2.1. Film preparation 2.2. Structural characterization A schematic diagram of the experimental set-up is presented in Fig. 1(a) and (b). The pulsed TEA CO 2 laser (203 LUMONICS ) parameters used were the wavelength, l=10.56 mm, a pulse duration of 50 ns, and a repetition rate of 1 Hz. The energy per laser irradiation pulse was 10 J, and the laser beam was properly focused on the target at an incidence angle of 45°. The substrates were placed at a distance of 3 cm from the graphite target. CN films were deposited over a period of 60 min x on a silicon substrate by laser ablation from a graphite target under nitrogen plasma afterglow conditions. Nitrogen flow was excited in an electrode-less discharge by means of a SAIREM 2450-MHz microwave generator. Unless otherwise specified, the transmitted power was fixed at 1000 W. The discharge was produced in a quartz tube, and, by continuous pumping, the plasma was fed into the reaction chamber located at two different distances from the discharge d=0.5 and 0.9 m. For d=0.5 m, the substrates were subjected to the short-

The films were characterized by Raman (Dilor spectrometer using a 160-mW laser at a wavelength of 541.5 nm) and FTIR (Perkin Elmer 1600 spectrometer) spectroscopies. The surface composition of the deposited films was characterized by XPS (LHS 10 spectrometer). An AlKa X-ray source was operated at 13 kV and 20 mA emission current. The pressure during analysis was below 1.3×10−6 Pa. The peaks were deconvoluted, assuming Gaussian profiles. The relative atomic stoichiometries N/C and O/C, denoted X and X , respectively, N O were determined according to Ref. [4]. 2.3. Mechanical properties Surface mechanical properties were assessed by depth-sensing indentation (DSI ) (NHT by CSEM Switzerland ) and by an original micro-scratching rig, as described, respectively, in Refs. [5–7]. In the latter case,

C. Jama et al. / Surface and Coatings Technology 116–119 (1999) 59–64

61

both conventional single pass under progressive loading and multipass operation under constant load were practised. Surface damage was assessed by means of standard reflected light microscopy (RLM ) and scanning electron microscopy (SEM ), when necessary for reasons of depth resolution. 3. Results and discussion 3.1. Infra-red spectroscopy and Raman The infra-red spectra of carbon nitride films deposited by plasma decomposition of hydrocarbons and N were 2 characterized by Han et al. [8]. Bands at 1370 and 1570 cm−1 detected for nitrogen-doped amorphous carbon were attributed to the sp2 graphitic bonds by Kaufman et al. [9] and Kumar et al. [11]. For CN x films deposited by pulsed laser, the bands at 1500–1750 cm−1 and 1200–1450 cm−1 were respectively identified as C앞N and C–N by Zhao et al. [10]. FTIR spectra (not shown) for CN films deposited under x CRNP or SLA conditions show a broad band in the region of 1000–1600 cm−1, which may be related to the diamond-like carbon film structure [12]. The 1250–1400 cm−1 band could correspond to the sp3 carbon single bonded to nitrogen C–N in tetrahedral environment. The 1400–1600 cm−1 band could be due to the C앞N vibrational mode [11]. Raman spectra for carbon nitride films exhibit the characteristic D and G bands of graphite, as shown in Fig. 2. Fig. 2(a) presents films obtained at 500 W under CRNP. These films displayed a lower-intensity D band centred at 1350 cm−1 and higher G band centred at 1560 cm−1. A slight increase in the intensity of the D band is observed for films obtained under CRNP at 1000 W [Fig. 2(b)], which is qualitatively evident from the higher ratio of intensities of the bands (I /I ). This D G ratio is a measure of the degree of disorder in amorphous carbon [9]. Hence, an increase in structural disorder is due to the increase in microwave power. Furthermore, a slight shift towards low wavenumber values is observed in comparison to films obtained under the same conditions with a He–N plasma [13]. The D band was 2 observed at 1363.3 cm−1, and the G band was detected at 1583.7 cm−1. This could be due to a higher contribution of carbon atoms in a sp3 tetrahedral bonding configuration in our films [14,15]. 3.2. XPS spectroscopy Table 1 shows the evolution of film composition as a function of pressure for CN films obtained under CRNP x or SLA conditions. Film composition values provide evidence that a higher nitrogen composition ratio relative to carbon N/C (X ) is obtained under SLA conditions. N

Fig. 2. Raman spectra of CN films deposited under CRNP conditions. x

At a pressure of 1000 Pa, X =0.55 is obtained with the N CRNP, whereas X increases to 0.74 under SLA condiN tions. Furthermore, X increases with increasing nitrogen N pressure in the CRNP case, whereas there is no significant change in the case of SLA conditions. Fig. 3 shows the deconvoluted C 1s peaks for films deposited under CRNP and SLA conditions, at 1000 Pa. The C 1s peak consists of four major contributions. The peak at a binding energy (B.E.) of 284.6 eV is attributed to C bonded to C in a graphitic form. The contribution at 286.2 eV is assigned to sp2 trigonal CN bonding, and the peak at 287.8 eV is attributed to sp3 tetrahedral CN bonding. The contribution at 288.9 eV corresponds to CO bonds formed upon air exposure. These attributions are similar to those presented in the literature [16 ]. The N 1s peak (Fig. 4) shows two main peaks: the first at 398.7 eV corresponds to sp3 CN contribution, and the second at 400.2 eV is due to sp2 CN bonding. A small contribution at 401.6 eV is probably due to some N–O contamination upon air exposure. This deconvolution confirms the two different CN bonding configurations detected on the C 1s peaks. Data from Table 1 clearly show that for the SLA condition, sp3 bonding configurations for nitrogen incorporation in the film are favoured. In the CRNP case, it is difficult to say which CN bonding configuration is favoured for nitrogen incorporation in the film. N 1s

62

C. Jama et al. / Surface and Coatings Technology 116–119 (1999) 59–64

Table 1 Nitrogen and oxygen N/C (X ) and O/C (X ) atomic ratios and C 1s and N 1s deconvolutions for CN films deposited under CRNP and N O x SLA conditions Sample

SLA CRNP

C 1s

C3 C4 C2 C1

N 1s

X

C–C

sp2CN

sp3CN

C–O

sp3CN

sp2CN

N–O

B.E.a (eV )

284.6

286.2

287.8

288.9

398.7

400.2

401.6

500 Pa 1000 Pa 500 Pa 1000 Pa

20.5 29.0 29.1 17.2

28.0 22.9 44.5 52.8

43.5 39.7 20.7 21.6

8.0 8.4 5.7 8.4

58 59.4 47.0 53.0

37 33.8 48.5 42.5

5.0 6.8 4.5 4.5

N

0.77 0.74 0.48 0.55

X

O

0.20 0.13 0.08 0.09

a B.E.: binding energy.

Fig. 3. Decomposition of C 1s spectra ofCN films deposited at x 1000 Pa: (a) CRNP; (b) SLA.

Fig. 4. Decomposition of N 1s spectra of CN films deposited at x 1000 Pa: (a) CRNP; (b) SLA.

and C 1s decomposition does not seem to be in good agreement. Hence, C 1s shows a major contribution of sp2 CN bonding, whereas N 1s, shows a high sp3 bonding contribution. This disagreement is probably due to the presence of sp CN bonding, which appears in the C 1s peak at 286.2 eV (also sp2 region) and at 398.7 eV (also sp3 region) in the N 1s peak. Raman and FTIR spectroscopies do not give a clear diagnosis of the presence for sp CN bonding in the CRNP condition. The corresponding low-intensity peak observed in FTIR spectroscopy was not detected using the Raman technique.

3.3. Mechanical properties 3.3.1. Depth sensing indentation (DSI) Numerical results are compiled in Table 2. All experiments have been run with a standard Vickers indenter, a loading rate of 10 mN min−1, and a holding time with a maximum load of 10 s. Inspection of these data calls for various comments. As evidenced in Table 2, the major experimental variation is measured for the penetration depth at maximum load. This is indicative of experimental deficiencies corresponding to surface roughness and

63

C. Jama et al. / Surface and Coatings Technology 116–119 (1999) 59–64 Table 2 Experimental results in depth sensing indentation (NHT-CSEM ) Sample

Coating process conditions

Maximum contact loads and experimental results in-depth sensing indentation

d (m)

N 2 (slpm)

P (Pa)

P a m (mN )

D b m (nm)

Standard deviation

H V (GPa)

Standard deviation

E (GPa)

Standard deviation

C1 C2

0.9 0.9

4 2

1000 500

C3 C4

0.5 0.5

2 4

500 1000

5 1 3 5 5 5

400 200 324 399 466 329

10 9 17 5 5 23

1.17 0.77 1.03 1.17 0.87 1.78

– 0.06 0.10 0.03 0.03 0.33

38 22 38 51 39 74

– 1 2 1 2 2

a P : maximum indentation load. m b D : maximum indentation depth. m

dispersion in mechanical strength of these pilot coatings. DSI, in the present case, cannot be expected to yield reliable absolute experimental values, but rather gives an indication of general coating quality. Indeed, in all of these low load indentations, radial cracking and lateral cracking are observed on the optical micrographs. It may explain the increase in hardness for increasing maximum contact load (Sample C2, Table 2) which is contradictory with respect to common knowledge on load dependence of hardness for thin coatings. With these data, it is certainly risky to conclude on any cross-correlation between microstructure of the coatings as resulting from the process parameters and the corresponding mechanical properties. 3.3.2. Microscratch testing For the microscratch experiments run with an original instrument based on the deflection of a macroscopic cantilever, the following experimental conditions were adopted: $ $ $

indenter tip radius: 10 mm maximum load in progressive loading : 80 mN sliding speed: 0.05 mm s−1.

The numerical results of these experiments are compiled in Table 3. A comparison of the significance of the two mechanical testing techniques is definitely in favour of micro-scratch testing. However, the two techniques are altogether coherent, considering the hardness values H in static indentaV tion and the scratch hardness (H ). H is commonly s s defined as the ratio between the normal force (F ) and N the load bearing area (A ) in a constant load test. Using LB a spherical tip, this leads to [17,18]: F F H = N =8 N . S A pd2 LB The most significant mechanical parameter in Table 3 is the number of cycles to failure, N . C The engineering conclusions for future pursuit of our coatings is as follows: $

$

The distance of 0.9 m from the discharge appears to be the optimum situation. A high working pressure and a high nitrogen flow appear to be equally in favour of good mechanical properties. Indeed, specimen C2, as for C1 in CRNP

Table 3 Experimental results in microscratch testing Sample

C1 C2 C3 C4

Coating conditions

Progressive loading (3-mm sliding distance)

Constant load multipass operation (1-mm sliding distance)

da (m)

N 2 (slpm)

P (Pa)

L b (mN ) C

Surface damage

F c n (mN )

N d C

d (mm)

H e S (GPa)

0.9 0.9 0.5 0.5

4 2 2 4

1000 500 500 1000

17 9 10 Spontaneous, large-area spalling on cleaning in ethanol

Spalling Blistering Blistering

6 6 6

5–10 5–10 4

5 3 4

0.61 1.70 0.95

a d: track width. b L : critical contact load for spalling. C c F : contact load on slider. n d N : number of cycles to failure. C e H : scratch hardness. S

64

C. Jama et al. / Surface and Coatings Technology 116–119 (1999) 59–64

conditions at a distance of 0.9 m from the discharge rate, clearly has a shorter ‘service life’ in low cycle friction fatigue operations, as practised in multipass unidirectional scratching under a contact load of 6 mN.

4. Conclusions Carbon nitride CN films have been deposited on x silicon wafers on a non-heated substrate using pulsed CO laser ablation of carbon molecular fragments from 2 a graphite target under nitrogen plasma afterglow. Fairly high values of N/C ratios (up to 0.77) were measured in the films. Raman, XPS and FT-IR spectroscopies demonstrate the existence of two major environments for nitrogen, one in which N is single bonded to the sp3 tetrahedral carbon (CN sp3), and another in which N is double-bonded to carbon. XPS results show that higher N/C ratios with higher CN sp3 concentrations are obtained for films when the distance between the discharge and the deposition zone (d ) decreases.

Acknowledgement All of the fastidious depth-sensing indentation experiments as well as the corresponding micrographs for failure diagnosis have been done by B. Boreux.

References [1] C.J. Torng, J.M. Sivertsen, J.H. Judy, C. Chang, J. Mater. Res. 5 (1990) 2490. [2 ] H. Sjo¨stro¨m, L. Hultman, J.E. Sundgren, S.V. Hainsworth, T.F. Page, G.S.A.M. Theunissen, J. Vac. Sci. Technol. A 14 ( 1996 ) 56. [3] P. Supiot, O. Dessaux, P. Goudmand, J. Phys. D 28 (1995) 1826. [4] C.J. Powell, P.E. Larson, Appl. Surf. Sci. 1 (1978) 186. [5] N.X. Randall, C. Julia-Schmutz, J.M. Soro, J. von Stebut, G. Zacharie, Thin Solid Films 308/309 (1997) 297. [6 ] R. Consiglio, N. Durand, K.F. Badawi, P. Macquart, F. Lerbet, M. Assoul, J. von Stebut, Surf. Coat. Technol. 97 (1997) 192. [7] M. Dupeux, A. Bossebœuf, R. Kouitat, J. von Stebut, Surf. Coat. Technol. (1999) in press. [8] H.X. Han, B.J. Feldman, Solid State Commun. 65 (1988) 921. [9] J.H. Kaufman, S. Metin, D.D. Saperstein, Phys. Rev. B 39 (1989) 13053. [10] X.A. Zhao, C.W. Ong, Y.C. Tsang, Y.W. Wong, P.W. Chang, C.L. Choy, Appl. Phys. Lett. 66 (1995) 2652. [11] S. Kumar, T.L. Tansley, Thin Solid Films 256 (1995) 44. [12] P.V. Nagarkar, K.E. Sichel, J. Electrochem. 136 (1982) 2979. [13] C. Jama, V. Rousseau, O. Dessaux, P. Goudmand, Thin Solid Films 302 (1997) 58. [14] D. Beeman, J. Silverman, R. Lynds, M.R. Anderson, Phys. Rev. B 30 (1984) 970. [15] R.O. Dillon, J.A. Woollam, V. Katkanant, Phys. Rev. B 29 (1984) 3482. [16 ] D. Marton, K.J. Boyd, A.H. Al-Bayati, S.S. Todorov, J. Rabalais, Phys. Rev. Lett. 73 (1994) 118. [17] J.A. Williams, Tribol. Int. 29 (8) (1999) 675. [18] P.J. Blau, in: ASM Handbook Vol. 18 (1992) 414.