Surface and Coatings Technology 151–152(2002) 160–164

0257-8972/02/$ - see front matter� 2002 Elsevier Science B.V. All rights reserved.PII: S0257-8972Ž01.01639-5

Laser-assisted deposition of r-B C coatings using ethylene as carbon4

precursor

M.J. Santos , A.J. Silvestre , O. Conde *a b a,

Departamento de Fısica, Universidade de Lisboa, Ed. C8, Campo Grande, 1749-016 Lisboa, Portugala ´Instituto Superior de Transportes, R. Castilho, No. 3, 1269-074 Lisboa, Portugalb

Abstract

Rhombohedral B C coatings were synthesised on fused silica substrates by CO laser-assisted chemical vapour deposition4 2

(LCVD) using a dynamic reactive atmosphere of BCl , C H and H . Films with carbon content from 15 to 22 at.% were grown3 2 4 2

at deposition rates as high as 0.12mm s . The kinetics of the reactive system used to deposit the B C films and the influencey14

of growth conditions on the structure and morphology of the deposits were investigated.� 2002 Elsevier Science B.V. All rightsreserved.

Keywords: Rhombohedral boron carbide(r-B4C); Laser-CVD; Growth kinetics

1. Introduction

Rhombohedral boron carbide(r-B C) is a ceramic4

material of great interest for a wide variety of applica-tions because of its attractive mechanical, thermal andelectronic propertiesw1x. The r-B C has found applica-4

tion as a neutron absorbent material in the nuclearindustry due to its high neutron capture cross-sectionw2x. Also of particular importance are its low specificweight and high hardness, the latter even surpassingdiamond and boron nitride at temperatures over 11008Cw3x. Moreover, it presents a high melting point andmodulus of elasticity, and has great resistance to chem-ical agents. This combination of properties makes boroncarbide a prominent corrosion-resistant ceramic materialfor thin film applications. Furthermore, considering itshigh-temperature stability, large Seebeck coefficient andlow thermal conductivity, boron carbide could findpotential use as high-temperature thermoelectric materialfor energy convertersw1,4x.In previous studies conducted by our group, laser-

assisted chemical vapour deposition(LCVD) of boron-carbon films was carried out from a dynamic reactiveatmosphere of BCl , CH and H using a continuous3 4 2

wave CO laserw5–7x. Compared to methane, which is2

the conventional carbon precursor in CVD processes,

* Corresponding author. Tel.:q351-1-7500035; fax:q351-1-7573619.

E-mail address: oconde@fc.ul.pt(O. Conde).

ethylene presents several advantages due to its highabsorption coefficient at the CO laser wavelength and2

its higher sticking coefficient, enabling us to achievehigher deposition rates.It is the purpose of this paper to report on the kinetics

of CO laser-CVD of r-B C films using C H as carbon2 4 2 4

precursor and on the influence of growth conditions onthe structure and morphology of the deposited material.

2. Experimental procedure

The deposition system and the procedure used forlaser-assisted CVD of r-B C have been presented else-4

where w5x and only a brief description is given here.Boron carbide films were deposited on silica substratesusing a CO laser as heat source, operated in cw2

TEM mode at a wavelength of 10.6mm, and a00

dynamic reactive gas mixture of BCl , C H and H .3 2 4 2

Argon was used as buffer gas. The laser beam reachesthe substrate at perpendicular incidence with a 1ye2

spot diameter of 12 mm. No focus lens was used sincesilica absorbs 84% of the laser radiation. Prior to theirinsertion in the reactor, the substrates were cleaned inultrasonic baths of acetone and ethanol. The reactionchamber was always evacuated to a base pressure of-10 mbar before the introduction of the gaseousy6

reactants.In this study, the total pressure and argon flux were

kept constant at 133 mbar and 400 sccm, respectively.

161M.J. Santos et al. / Surface and Coatings Technology 151 –152 (2002) 160–164

Table 1Process parameters for LCVD of r-B C films4

Experimental parameters Range of values

Laser power(W) 180–250Interaction time(s) 90BCl flow rate (sccm)3 41C H flow rate(sccm)2 4 1–3H flow rate (sccm)2 150–200

Fig. 1. GIXRD spectra of r-B C films with different carbon contents, as measured by EPMA, prepared with C H at the following experimental4 2 4

parameters:(a) ws0.13,F s200 sccm,Ps220 W; (b) ws0.13,F s150 sccm,Ps220 W; (c) ws0.09,F s200 sccm,Ps240 W; (d)H H H2 2 2

ws0.09,F s150 sccm,Ps200 W; (e) ws0.05,F s150 sccm,Ps240 W; (f) ws0.09,F s200 sccm,Ps200 W.H H H2 2 2

The other experimental parameters were varied in theranges shown in Table 1. The relative amount of carbonand boron in the reactive gas phase is given by theparameterws2F y(2F qF ), whereF is theC H C H BCl i2 4 2 4 3

flow rate of precursor i. In this studyw took valuesbetween 0.05 and 0.13.The structure of the as-deposited films was studied

by X-ray diffraction at glancing incidence(GIXRD) of18 with Cu–Ka radiation and the chemical compositionwas investigated by electron probe microanalysis(EPMA). The surface microstructure of the films wasexamined by scanning electron microscopy(SEM) andthickness profiles were measured by optical profilometry.

3. Results and discussion

3.1. Chemical and structural analysis

Chemical analysis by EPMA showed that films withuniform composition were produced in a broad range ofcarbon content, from 15 to approximately 22 at.% C,consistent with the B–C phase diagram.All X-ray spectra display the r-B C diffraction pattern,4

showing narrow diffraction lines. Depending on the

carbon content, the spectra match the JCPDS cards 33-0225 or 35-0798 corresponding to the B C and13 2

B C stoichiometries, respectively(Fig. 1). Moreover,12 3

the deposition of graphite was never observed and onlyboron carbide phase was detected by GIXRD.The two major diffraction peaks characteristic of r-

B C compounds correspond to the(104) and (021)4

reflections. Their 2u angular position varies with thecarbon content in the films, as can be seen from Fig. 1.Both peaks shift left approximately 0.28 when the carbonconcentration decreases from approximately 22 to 15at.% C. This observed shift is mainly related with thesubstitution of a carbon atom by a boron atom in thecentralC–B–Cintericosahedral chain(B C structure),12 3

leading to aC–B–B chain in the main diagonal of therhombohedral structure of B C boron carbidew7x.13 2

Similarly to the films synthesised with methanew6,7x,the coatings prepared from ethylene present an inversionof the relative intensities of the(104) and (021) linesfor carbon content lower than 17 at.% C, suggesting thedevelopment of a(104) texture(Fig. 1, diffractogramse and f). Although this trend to develop the(104)texture is not well understood, it is clearly independentof carbon precursor. Furthermore, preferential crystallo-graphic orientations in this range of carbon content havealso been observed in other CVD processes, namely inlow-pressure CVDw8x.

3.2. Growth kinetics

Gaussian or near flat thickness distribution profileswere observed, allowing us to calculate apparent depo-sition rates by taking the ratio of the maximum height,

162 M.J. Santos et al. / Surface and Coatings Technology 151 –152 (2002) 160–164

Fig. 2. Deposition rate as a function of H flow rate, for two different2

w values and for the following laser power values:j, Ps200 W;h,Ps220 W;d, Ps240 W.

above the substrate surface, to the irradiation time.Deposition rates between 0.03 and 0.12mm s werey1

measured, which are one order of magnitude larger thanthose obtained in CVD processesw9,10x. Fig. 2 showsthe deposition rate of r-B C films as a function of H4 2

flow rate, for ws0.09 and 0.13 and for three distinctlaser power values. As can be seen, the deposition rateincreases slightly with thew parameter while laser powerand hydrogen concentration in the gaseous mixture arethe process parameters that most strongly determine therate at which films are grown. For a given reactiveatmosphereysubstrate system and interaction time, themaximum temperature attained during the laser–materialinteraction increases as laser power increases. As aconsequence, the deposition rate is greatly influencedby this process parameter.As also depicted in both graphs in Fig. 2, the

deposition rate varies withF going through a maxi-H2

mum at 175 sccm. The increase of the deposition rate

with the hydrogen flux, in the lowF range, is expectedH2

because hydrogen favours the deposition reaction ofboron carbide by reducing the BCl and preventing3

graphite formation. On the contrary, the decreasingbehaviour for the higherF values could be explainedH2

by considering that the deposited thin films are etchedby the hydrogen. Nevertheless, the usual temperaturefor this process in bulk B C is;1200 8C w11x, which4

is much higher than the deposition temperature valuescalculated in this work. Another hypothesis to under-stand the decrease of the deposition rate at highF isH2

based on Soret’s effect, by which the heavier BCl and3

C H molecules are relegated to the colder zones, a2 4

substantial fraction of the adsorption sites in the centralirradiated region being occupied by the lighter H2

moleculesw12x. This leads to a limited access of the Band C precursors to the central reaction zone, andtherefore to an inhibition of film growth. The balancebetween these two processes yields aF value at whichH2

the deposition rate of r-B C presents a maximum. For4

this optimised H flux, the number of nucleation centres2

is higher leading not only to higher deposition rates butalso to films presenting a much more compact anduniform microstructure.Arrhenius diagrams are a common convenient way to

infer on the reaction mechanism that controls depositionw13x. Mass transport in the vapour phase and surfacekinetics are the rate-limiting mechanisms in conventionalCVD and also in LCVD, although the latter process hasa more complex structure. The Arrhenius plot presentedin Fig. 3 shows the ln of the r-B C deposition rate as a4

function of the reciprocal central deposition temperature.The surface temperature achieved at the centre of thefilms during deposition was estimated between 780 and1430 K, following the calculation technique describedby Oliveira and Condew6x. It can be seen from theArrhenius diagram that two straight lines with differentslopes are needed to fit the data. Thus, within thedeposition temperature range induced by the laser radi-ation and for the reactive atmosphere utilised, two rate-limiting mechanisms were identified for the LCVD ofr-B C films:4

1. At high deposition temperatures(T)1050 K), therate-limiting step for film growth is the mass transportof the reactive gaseous species. In this temperatureregion the slope of the straight line has a small valueof 5.3 kJ mol . Due to the reduced number ofy1

experimental data used in the fitting procedure, ahigh standard error of"2.7 kJ mol was calculated.y1

2. At low deposition temperatures(T-1050 K), theapparent activation energy deduced from the Arrhe-nius plot is E s30.7"2.1 kJ mol . We can thusy1

a

conclude that surface chemical reaction kinetics isthe rate-limiting step for film growth at temperaturesbelow 1050 K.

163M.J. Santos et al. / Surface and Coatings Technology 151 –152 (2002) 160–164

Fig. 3. Logarithm of the apparent deposition rate as a function of reciprocal temperature.j, measured values; —, least square fitting.

Fig. 4. Evolution of surface microstructure of films processed at different experimental conditions.

164 M.J. Santos et al. / Surface and Coatings Technology 151 –152 (2002) 160–164

It should be noted that the magnitude of the apparentactivation energy deduced from the kinetic regime ismuch lower than theE values that usually characterisea

this regime in purely thermal CVD processesw13x.Because ethylene, as well as boron trichloride, absorbsthe infrared CO laser radiation through vibrational2

molecular excitationw14–16x, different reaction path-ways can be opened between the excited reactantsleading to a lower reaction activation energy.

3.3. Microstructure

SEM analysis of the coatings surface shows a crys-talline morphology, the size and shape of the crystalsdepending on the deposition conditions. Fig. 4 illustratesthe microstructure evolution with both laser power andgas phase composition. Also, a cross-section of a coatingis presented exhibiting good adherence and columnargrowth. As can be seen in Fig. 4 sequence(a), hydrogenconcentration in the gaseous mixture plays an importantrole in the microstructure. Following the discussiongiven in the previous section, the more compact anduniform microstructure is achieved for the optimisedH flux, i.e. 175 sccm. The microstructure is also2

strongly determined by the relative amount of carbonand boron in the reactive atmosphere. Fig. 4b shows theevolution asw increases from 0.05 to 0.13. At lowwvalues, the growth of approximately spherical nodulesis observed. For higher carbon content in the gas phase,a second nucleation and growth mechanism takes placeleading to a fine and uniform grain structure that coversthe underlying nodular structure. At maximumw values,nodules disappear and a continuous and uniform mor-phology develops showing well-defined crystal facetswith a pyramidal geometry. Moreover, SEM micrographsreveal a substantial growth of the grain size(Fig. 4c)as laser power increases by only 20 W(from 220 to240 W) emphasising the remarkable influence of thelaser powerydeposition temperature as referred topreviously.

4. Conclusions

Laser-CVD from BCl and C H as boron and carbon3 2 4

precursors, respectively, yield r-B C films with good4

adherence, well developed grain structure and carboncontent in the range 15–22 at.%. The deposition rateand the surface microstructure strongly depend on laserpower and hydrogen content in the gas phase. For theexperimental conditions used in this work, it was shownthat growth kinetics is dominated by a surface reactionmechanism atT-1050 K while mass transport in thevapour phase is the controlling mechanism of thin filmgrowth atT)1050 K.

Acknowledgements

This work was partially funded by Fundacao para a˜¸Ciencia e Tecnologia under POCTI.ˆ

References

w1x D. Emin, Phys. Rev. B 38(1988) 6041.w2x X. Deschanels, D. Simeone, J.P. Bonal, J. Nucl. Mater. 265

(1999) 321.w3x R. Telle, in: M.V. Swain(Ed.), Structure and Properties of

Ceramics, Mater. Sci. Technol, 11, VCH Publishers, Weinheim,1994, p. 173.

w4x M. Olsson, S. Soderberg, B. Stridh, U. Jansson, J.-O. Carlsson,Thin Solid Films 172(1989) 95.

w5x J.C. Oliveira, M.N. Oliveira, O. Conde, Surf. Coat. Technol.80 (1996) 100.

w6x J.C. Oliveira, O. Conde, Thin Solid Films 307(1997) 29.w7x O. Conde, A.J. Silvestre, J.C. Oliveira, Surf. Coat. Technol.

125 (2000) 141.w8x J. Rey, G. Male, Ph. Kapsa, J.L. Loubet, J. Phys. Coll. C5 50

(1989) 311.w9x U. Jansson, J.-O. Carlsson, B. Stridh, J. Vac. Sci. Technol. A5

(1987) 2823.w10x O. Postel, J. Heberlein, Surf. Coat. Technol. 108-109(1998)

247.w11x K.A. Schwetz, A. Lipp, Ullmann’s Encyclopaedia of Industrial

Chemistry, A4, VCH Verlag, Weinheim, 1985, pp. 295–307.w12x J.O. Hirschfelder, C.F. Curtiss, R.B. Bird, in: M.G. Mayer

(Ed.), Molecular Theory of Gases and Liquids, John Wiley &Sons, New York, 1964.

w13x J.-O. Carlsson, in: D.S. Rickerby, A. MatthewsŽEds..,Advanced Surface Coatings — A Handbook of Surface Engi-neering, Blackie & Son, 1991, pp. 162–193.

w14x N. Karlov, Appl. Optics 13(1974) 301.w15x T. Shimanouchi, Tables of Molec. Vibrational Frequencies

Consolidated, vol. I, NSRDS-NBS 39, 1991, p. 74.w16x J.I. Steinfeld, Laser and Coherence Spectroscopy, Plenum

Press, New York, 1978, p. 62.