Diamond and Related Materials 12(2003) 1213–1219

0925-9635/03/$ - see front matter� 2003 Elsevier Science B.V. All rights reserved.PII: S0925-9635Ž02.00393-X

Doping and electrical properties of amorphous silicon carbon nitride films

Y.C. Chou , S. Chattopadhyay , L.C. Chen *, Y.F. Chen , K.H. Chena a a, b c

Centre for Condensed Matter Sciences, National Taiwan University, Taipei 106, Taiwan, ROCa

Department of Physics, National Taiwan University, Taipei 106, Taiwan, ROCb

Institute of Atomic and Molecular Sciences, Academia Sinica, Taipei 106, Taiwan, ROCc

Abstract

Electrical properties and annealing behaviour of undoped and doped amorphous silicon carbon nitride(a-SiC N) thin films,x y

deposited by ion beam sputtering techniques, have been studied. Doping of the a-SiC N thin films with magnesium(Mg), andx y

phosphorous(P) was carried out by ion implantation techniques, and subsequent annealing effect on the electrical conductivity(s) and activation energies for electrical conduction have been reported. It was found that the undoped films were insulatingwith electrical conductivities in the range of 10 –10 Sycm. Annealing of these films at high temperatures aided in somey6 y8

structural relaxation and hence an increase ins by several orders of magnitude without showing any indications for crystallization.Suitable doping(Mg) of the films resulted in increaseds, and in some cases of low phosphorous doping a decrease ins wasalso found which indicated that the films may be intrinsically p-type.� 2003 Elsevier Science B.V. All rights reserved.

Keywords: Physical vapour deposition; Ion implantation; Electrical conductivity; Photoelectron spectroscopy

1. Introduction

The family of carbon based covalent materials contin-ues to provide newer directions to material scienceresearch. Diamond and silicon carbide group of materialsfor example, along with the crystalline carbon nitrides,have created avenues for both experimental and theor-etical studies. With the advent of nanoscale science theystill remain a formidable family that promises to deliverin several fronts of material science. The reports ofnanotubes, nanorods and nanotips of carbon or siliconcarbide or silicon carbon nitride(SiC N ) support suchx y

claims. SiC N material, although relatively new to thisx y

family, have shown interesting opticalw1x, mechanicalw2x and thermalw3x properties, owing to the small sizeand covalent nature of their constituent atoms. Crystal-line SiC N (c-SiC N ) films exhibited a direct and widex y x y

band gap of 3.8 eV and an optical emission approxi-mately 2.8 eVw1,4x. Nanoindentation measurements onthe crystalline SiC N suggested its hardness and bulkx y

modulus to be in the vicinity of 30 and 322 GPa,respectively w2x, whereas for amorphous SiC N(a-x y

SiC N ) films, the hardness value was approximately 22x y

*Corresponding author. Tel.:q886-2-2366-8228; fax:q886-2-2362-0200.

E-mail address: chenlc@ccms.ntu.edu.tw(L.C. Chen).

GPa w2x and a Young’s modulus of approximately 245GPaw5x.

Although optoelectronic, mechanical and thermalproperties of a-SiC N have been reported, informationx y

on their electrical properties and doping efficiencies hasbeen rather limited. Organosilicon polymer derived non-stoichiometric SiC N has been reported to have highx y

electrical conductivityw6x where nitrogen was believedto be a dopant in the ceramic matrix. Electrical proper-ties of crystalline compacts derived out of hot-pressedSi–C–N powders were also reportedw7x. Haluschka etal. w8x demonstrated the effects of chemical composition,solid state structure and microstructure of the polysila-zane derived silicon carbonitride ceramics on their elec-trical properties. The amorphous phase we areinvestigating is derived from a physical vapour deposi-tion of the constituent Si, C and N atoms in a thin filmform. This paper presents a systematic study on theelectrical conductivity in a-SiC N , with specific treat-x y

ments such as annealing and doping through ionimplantation.

2. Experimental

The a-SiC N samples were deposited via ion beamx y

sputtering(IBS) technique. A 3-cm Kaufmann type ion

1214 Y.C. Chou et al. / Diamond and Related Materials 12 (2003) 1213–1219

Table 1Composition and electrical conductivity of a-SiC N films having dif-x y

ferent carbon content

Silicon Carbon Nitrogen Electricalcontent content content conductivity(at.%) (at.%) (at.%) (Sycm)

21 47 32 2.33=10y8

20 50 30 6.3=10y7

20 55 25 5.6=10y6

Fig. 1. Representative AES depth profile of an a-SiC N film showing the uniform composition along the thickness of the sample.x y

source was used for sputtering targets of laminatedsilicon carbide or silicon and graphite in nitrogen ambi-ent. A detail of the IBS reactor is given elsewherew9x.The films were grown on quartz substrates for theelectrical conductivity measurements and on silicon forother characterisations. Ion implantation technique wasadopted for doping the films. An ULVAC IMX-3500ion implanter was used at 75 KeV to dope the sampleswith zinc (Zn), magnesium(Mg) and phosphorus(P)at different incident ion fluxes varying from 10 to12

10 ionsycm . The annealing of the films were carried16 2

out by rapid thermal processing(RTP) in controlledatmosphere. RTP uses rapid heating of the samples tothe desired temperature and a shorter hold time(2–10min) unlike other conventional annealing techniques.The a-SiC N films on quartz were systematically stud-x y

ied for the electrical conductivity(s) by the commontwo-probe technique. Gold electrodes were evaporatedon to the samples for proper contact whose ohmiccharacteristics were then verified by the linear current–voltage(I–V) characteristics. The chemical compositionand bondings in the films were determined by Auger(AES) depth profile and high resolution scans of Si(2p),B(1s), C(1s) and N(1s) peaks by X-ray photoelectronspectroscopy(XPS). A PHI 1600 system was used forthe XPS measurements. A Perkin-Elmer scanning Augernanoprobe system(SAN 760) was used for Augerelectron spectroscopy(AES). Depth profiling of thedoped samples were also done using secondary ion massspectroscopy(SIMS) after implantation.

3. Results

The a-SiC N samples were first studied for theirx y

composition and depth profiles using AES. A represen-tative composition depth profile is shown in Fig. 1,

which reveals the homogeneity of composition alongdifferent depth of the sample. The sample studied had50 at.% carbon(C), 25 at.% nitrogen(N), 20 at.%silicon (Si) and less than 10 at.% oxygen(O) in it.

Films with different compositions ands are tabulatedin Table 1. Films with higher carbon content was foundto have a highers. Fig. 2 shows the deconvoluted C 1sXPS spectra for the films with carbon contents of 47and 55 at.%, respectively. The main contribution to thepeak comes from the C(sp )–C and C(sp)–N with2

binding energies of 284.6 and 286.5 eV, respectivelyw3x. The film with higher carbon content had a 37%higher ratio of the C(sp)–N bonds to C(sp )–C bonds2

relative to the lower carbon content film.The sensitivity of s towards the carbon content

indicated a rather strong structural dependence, and asimple way to alter the structure of a network isannealing whereby some stress in the network and somedefects could be efficiently removed. All electricalmeasurements that will be discussed were carried outon the a-SiC N sample with 50 at.% carbon, if notx y

specifically mentioned. Fig. 3 shows the changes ins

for an undoped a-SiC N sample with different annealingx y

temperatures. The electrical conductivity increased byfour orders of magnitude after annealing at temperatures

1215Y.C. Chou et al. / Diamond and Related Materials 12 (2003) 1213–1219

Fig. 2. Deconvoluted XPS C(1s) spectra for a-SiC N films with(a) 47 and(b) 55 at.% of carbon.x y

Fig. 3. Variation of the room temperature electrical conductivity(s) in undoped a-SiC N film with 50 at.% of carbon, as a function of thex y

annealing temperature.

1216 Y.C. Chou et al. / Diamond and Related Materials 12 (2003) 1213–1219

Fig. 4. Variation of thesT product as a function of the inverse3y2

temperature for undoped a-SiC N films with 50 at.% carbon,x y

annealed at(d) 800, (m) 900 and(�) 1000 8C. The activationenergies are shown on the graph for each case.

Fig. 5. A depth profile by SIMS of a-SiC N film annealed at 1000x y

8C and doped with P at(a) 1=10 , (b) 1=10 and(c) 1=1012 14 16

ionsycm .2

ranging from 600 to 10008C. Fig. 4 shows the temper-ature dependence ofsT for samples annealed at 800,3y2

900 and 10008C. From the slope of thesT vs.3y2

T plots, activation energies for conduction werey1

calculatedw8x. It was found that the activation energieswere decreased with higher annealing temperatures.Activation energy was 153 meV when annealed at 8008C, which decreased to 87 meV when annealed at 10008C.

Undoped a-SiC N samples were then doped withx y

different dopants and different dopant densities to studytheir effect on the electrical conductivity. A typicalSIMS depth profile analysis of C, N, O and P ions inthe a-SiC N samples with different incident ion fluxesx y

(of phosphorous) and subsequent annealing at 10008C,are shown in Fig. 5. At lower flux densities of 1012

ionsycm the phosphorous signal was low, but with the2

flux increasing we observed a non-uniform depth profileof phosphorous. Most of the dopants were detected at adepth of 0.1mm from the sample surface and a lowdensity of dopants at the surface was observed. Evenannealing at 10008C could not initiate an efficientphosphorous diffusion throughout the thickness of thesample to reduce such inhomogeneity. One set of undo-ped film was separately implanted with P and Mg ionswith a flux density of 1=10 ionsycm and annealed15 2

at 900 8C. The electrical conductivity increased uponMg ion implantation, but s decreased with smallamounts of P ion implantation. The activation energies,as found from thesT vs. T plots of these doped3y2 y1

samples(Fig. 6), also show a decrease to 87 meV withMg doping in one sample whereas an increase to 182meV upon P doping in the other, with respect to the

undoped activation energy of 123 meV. Mg dopingusing various flux densities and subsequent annealing at900 8C, produced different electrical conductivities andactivation energies(Fig. 7). Initially there was a dropin s, but thereafters increased.

4. Discussion

Electrical properties of amorphous materials are ofparticular interest since they are generally more defective

1217Y.C. Chou et al. / Diamond and Related Materials 12 (2003) 1213–1219

Fig. 6. Variation of thesT product as a function of the inverse temperature for(d) undoped,(j) phosphorus doped, and(m) magnesium3y2

doped, a-SiC N films with 50 at.% carbon, annealed at 9008C. The activation energies are shown on the graph for each case.x y

Fig. 7. Variation of thesT product as a function of the inverse temperature for Mg doped a-SiC N films with 50 at.% carbon, annealed at3y2x y

900 8C under incident ion flux of(d) 1=10 , (m) 1=10 and(�) 1=10 ionsycm . The activation energies are shown on the graph for14 15 16 2

each case.

and have high density of mid-gap defect states, such asin amorphous silicon(a-Si). a-SiC N is expected tox y

possess similar electrical conduction mechanism as thatof a-Si. The deposition routes used, IBS in this case,for the preparation of a-SiC N films produced variousx y

defects such as dangling bonds, voids as well as lowercoordinated bondsw3,5x. The presence of these defectsin the films would reflect in their electrical conductivi-ties. From the XPS studies we have found that in a-SiC N , carbon–carbon, carbon–nitrogen, andx y

silicon–nitrogen bonds are most easily detected withinsignificant or no presence of silicon–carbon bonds.Kinetics for the bond formation mechanism favouredthe C–C and C–N growth, which may be responsiblefor the insignificant Si–C bonds present in the system.Again from the Auger depth profile(Fig. 1) a uniformcomposition across the depths of the sample was foundwhich is important to understand the electrical conduc-tivity results of the coming sections. From Table 1 astrong increase ofs was observed with small variations

1218 Y.C. Chou et al. / Diamond and Related Materials 12 (2003) 1213–1219

in carbon content. Since the incorporated carbon linksmainly with either carbon or nitrogen atoms, the increasein s would most probably be due to some particularbonds involving these atoms. From Fig. 2 we found thatwith increase in carbon content there is an increase inthe sp bonded carbon atoms. The ratio of2

C(sp )–C:C(sp)–N increases from 0.8 to 1.1, when the2

carbon content increase from 47 to 55 at.%. These sp -2

carbon imparts the high electrical conductivity in graph-ite. Hence the increase ins by approximately 2 ordersof magnitude is probably due to the contribution ofthese sp bonds.2

The increase in the electrical conductivity of the a-SiC N materials with annealing is expected because ofx y

the structural relaxation that results in different levelsof defect removal. When annealed at different tempera-tures between 600 and 10008C, s showed steadyincrease from 10 to 10 Sycm (Fig. 3). The increasey6 y2

in s is not due to any crystallization of the material asfound by X-ray diffraction and transmission electronmicroscopy studies. The increase ins is hence attributedto purely structural relaxation of the amorphous matrix.

In the range of temperatures(R.T toy50 8C) probed,sT vs. T showed a linear variation according to3y2 y1

B EEay3y2 C FsAT exp yD GKT

whereE denotes the donoryacceptor levels with respecta

to the conductionyvalence band. The rate of coolingwill possibly affect thes values, but a quasi thermalequilibrium was maintained during all such measure-ments with sufficient relaxation times for a reliable andverifiable s. The activation energies, calculated fromthe slope of thesT vs. T plots in Fig. 4, for3y2 y1

undoped a-SiC N films annealed at different tempera-x y

tures also showed steady decrease with increasingannealing temperatures. This activation for electricalconduction is mainly due to hopping of carriers amongthe high density of states in the mobility gap. A higherslope of thesT vs. T plot at higher temperatures3y2 y1

were not observed. The reduction in the activationenergies indicates defect removal from the band tailstates as a result of thermal annealing.

The constituent atoms in a-SiC N films are smallx y

and covalent in nature, but a-SiC N films have a highx y

electrical resistivity. Efficient doping of this kind ofsystems are always possible although there will be acompetition between active doping and defect creationduring ion implantation. While explaining the electricalconductivity data of the doped samples the effect ofnon-uniform dopant distribution along the thickness ofthe sample has to be borne in mind also. The energyloss mechanism of the impinging ions during the implan-tation process is responsible for such non-homogeneity,which even annealing at 10008C could not remove

(Fig. 5). The electrical measurements were carried outin a planar electrode configuration, where an electricallyresistive channel near the surface would affect ourresults since the dopant density peaked at a depth of 0.1mm from the surface of the sample(Fig. 5). Mg dopingin the undoped films efficiently increases the electricalconductivity whereas P doping reduced the electricalconductivity to a certain extent(Fig. 6). Mg is a p-typedopant in our system while P is an n-type dopant. FromFig. 6, the indication is that the a-SiC N films arex y

intrinsically p-type, so that Mg will further lower theactivation energies hence increasing the electrical con-ductivity. This is only indicative since Hall measure-ments were not carried out on these samples. P dopingon the other hand, will compensate some of theseacceptor states and move the Fermi level toward mid-gap increasing the activation energy. This will reducethe electrical conductivity. Another reason for thedecrease ins while doping with P may be the defectsintroduced in the network during ion implantation. It isworth mentioning here that at sufficiently high fluxdensities(10 ionsycm ) of P ions we did find an16 2

effective increase ins. Increasing amounts of Mg inthe film obtained by implanting at higher flux densitieswill increase the electrical conductivity by more than anorder of magnitude with concomitant decrease in theactivation energies from 110 to 98 meV(Fig. 7).

5. Conclusions

Electrical properties of undoped(as-deposited) anddoped (ion-implanted) a-SiC N thin films have beenx y

discussed. It is found that the electrical conductivityincreases with the carbon content in the films forundoped SiC N samples. Annealing of the films byx y

RTP is useful in structural relaxation of the material aswell as annihilation of certain defects and a resultantincrease in the electrical conductivity of these films.Annealing in this case did not produce the generallyobserved crystallization of the amorphous matrix. Dop-ing of the matrix by magnesium and subsequent anneal-ing will efficiently increase the electrical conductivityof the films whereas phosphorus doping initially reducedthe electrical conductivity, indicating that the films maybe intrinsically p-type. Decreasing activation energyvalues with annealing at high temperatures reveals low-ering of defect levels aiding in increased electricalconduction. The low diffusion of dopants even at highannealing temperatures speaks of the superior thermalstability of the a-SiC N matrix.x y

Acknowledgments

The work was carried out under a project funded bythe National Science Council, Taiwan, under ContractNos. NSC90-2112-M002-048, and Ministry of Educa-

1219Y.C. Chou et al. / Diamond and Related Materials 12 (2003) 1213–1219

tion under Contract Nos. 90-W-FA01-2-4-5. One of theauthors(S.C) acknowledges the award of a post-doctoralfellowship by the National Science Council, Taiwan.

References

w1x D.Y. Lin, C.F. Li, Y.S. Huang, et al., Phys. Rev. B 56(1997)6498.

w2x L.C. Chen, K.H. Chen, S.L. Wei, et al., Thin Solid Films 355–356 (1999) 112.

w3x S. Chattopadhyay, L.C. Chen, C.T. Wu, et al., Appl. Phys.Lett. 79 (2001) 332.

w4x L.C. Chen, C.K. Chen, S.L. Wei, et al., Appl. Phys. Lett. 72(1998) 2463.

w5x G. Lehmann, P. Hess, J.J. Wu, et al., Phys. Rev. B 64(2001)165305.

w6x M. Scarlete, J. He, J.F. Harrod, I.S. Butler, NATO ASI SeriesE, Applied Sciences 297(1994) 125.

w7x A. Sawaguchi, K. Toda, K. Nihara, J. Am. Ceram. Soc. 74(1991) 1142.

w8x C. Haluschka, C. Engel, R. Riedel, J. Eur. Ceram. Soc. 20(2000) 1365.

w9x J.J. Wu, T.R. Lu, C.T. Wu, et al., Diamond Relat. Mater. 8(1999) 605.