Superhard nc-TiN/a-BN and nc-TiN/a-TiBx/a-BN coatings prepared by plasma CVD and PVD: a comparative study of their properties

1

ICMCTF San Diego April 2002, Paper B1-2-2, Surf. Coat. Technol. (in press)

Superhard nc-TiN/a-BN and nc-TiN/a-TiBx/a-BN Coatings

Prepared by Plasma CVD and PVD: A comparative Study of their Properties

P. Karvankovaa, M. G. J. Veprek-Heijmana, O. Zindulkab, A. Bergmaierc and S. Vepreka *aInstitute for Chemistry of Inorganic Materials, Technical University Munich,

Lichtenbergstr. 4, D-85747 Garching b. Munich, Germanyb SHM Ltd., CZ-788 03 Novy Malin 266, Czech Republic

c Physics Department E12, Technical University Munich, James-Franck-Str. D-85747 Garching

Abstract

We present a comparative study of the preparation and properties of superhard "Ti-B-N" coat-

ings deposited by plasma CVD and by Vacuum Arc Evaporation (PVD) of Titanium combined

with Plasma CVD of TiB2 and BN. Using high frequency plasma CVD at a total pressure of sev-

eral mbar with TiCl4, BCl3, N2 and H2 as reactants, superhard (HV ≈ 40 � 50 GPa) nanocompo-

site coatings were successfully and reproducibly deposited and characterized in terms of me-

chanical properties (indentation & SEM), phase composition (XPS and ERD) and nanostructure

(XRD, SEM). Using reactive sputtering, several authors reported about the preparation of super-

hard TiN/TiB2 coatings with only a small fraction of BN. Efforts to increase the fraction of the

BN phase resulted in soft films. In contrast, plasma CVD yields superhard nc-TiN/a-BN and nc-

TiN/a-TiB2/a-BN coatings in a wide range of the fractions of BN and TiB2 phases. This is attrib-

uted to the high chemical activity of nitrogen under the conditions of plasma CVD. Industrial

scale vacuum arc evaporation PVD in combination with plasma CVD is used for preparation of

"Ti-B-N" coatings. In the system nc-TiN1-x/a-TiB2 with a minor fraction of the a-BN phase su-

perhard coatings with a very low fraction of microparticles and resultant low surface roughness

of Rm ≈ 0.15 µm were successfully prepared and those are intended for testing under a variety of

cutting conditions.

* Corresponding author: Tel.: ..49-89-2891 36 24; fax: ..49-89-2891 36 26 E-mail address: [email protected]

1. Introduction

Superhard (H ≥ 40 GPa) coatings are receiving increasing attention because of their po-

tential applications for wear protection (e. g. on tools for cutting, forming and stamping). Only

diamond and cubic boron nitride are intrinsically superhard. Superhardness can be achieved in a

variety of ordinary hard coatings either by energetic ion bombardment during their deposition

2

(which results in a variety of effect including a high compressive stress) or by the formation of

an appropriate nanostructure, such as superlattices and nanocomposites (see [1 - 4] and refer-

ences therein). The hardness enhancement due to the energetic ion bombardment is easy to

achieve in a variety of coatings including the so called "nanocomposites" M(1)nN/M(2). Here

M(1) is a hard and stable transition metal nitride, such as ZrN, Cr2N, TiN, ... and M(2) is another

metal which does not form any thermodynamically stable nitride, such as Ni, Cu,... (for a review

on these "nanocomposites" see [5] and also [1]). However, the superhardness is lost when the

high stress and other beneficial effects induced by the energetic ion bombardment relax upon an-

nealing to 400 � 600 °C [1] [4] or upon a long term storage in air.

The superhard nanocomposites prepared according to the generic design principle [6] [7]

show a very high thermal stability up to 1000 � 1100 °C [8] [9]. This design principle is based on

the formation of stable nanostructure due to a strong, spinodal decomposition in a binary or ter-

nary system [10]. The most studied systems so far were nc-TiN/a-Si3N4, nc-W2N/a-Si3N4, nc-

VN/a-Si3N4, nc-(TiAl)N/a-Si3N4, and nc-TiN/a-Si3N4/a- & nc-TiSix prepared by plasma induced

CVD at a relatively large partial pressure of nitrogen of several mbar which assured the desired

phase segregation (see [7 - 10]).

The Ti-B-N system represents another class of superhard coatings which were prepared

by a variety of PVD methods, such as vacuum arc evaporation (e.g. [11] [12]) and sputtering (e.

g. [13 - 16] and further references in the review [7]). Because of limited space available here we

can mention only some highlights of the many papers on the Ti-B-N system which are relevant

to our present work. In the PVD techniques used, the partial pressure of nitrogen is orders of

magnitude lower than in our plasma CVD and, therefore, the segregation into TiN and BN is not

completed. As result, a variety of Ti, B and N containing phases are formed depending on the

exact conditions (N2 pressure, substrate temperature, ion bombardment,...). Using conventional

sputtering at a low pressure results either in the formation of a homogeneous phase TiB2Nx [17 -

19], or a TiNxBy phase consisting of TiN lattice in which the nitrogen sites are occupied by 33 %

of boron, 49 % of nitrogen and 18 % of vacancies [20], or amorphous Ti-B-N coatings [21] [22].

Other authors have found phase segregation with the formation of binary phases TiN and TiB2

[13 - 15] [23] [24] corresponding to the "region 4" of the equilibrium phase diagram [15] 1. At a

lower boron content, nanocrystalline TiN and quasi-amorphous TiB2 phase are formed whereas a

higher boron content leads to a structure consisting of quasi-amorphous TiN and nanocrystalline

TiB2 [23] [24]. A large number of "Ti-B-N" coatings with different composition and structure

1 The authors refer to the original work of Novotny et al. from 1961 [25] not considering the more recent literaturein which a variety of other phases was reported [26].

3

prepared by a variety of methods have been reported, such as a mixture of phases TiN, TiB2 and

BN or nc-TiN phase imbedded in a BN matrix [27 - 29], but in most cases, the coatings were

relatively soft of ≤ 11 GPa [30] [31]. Héau and Terrat reported a hardness of 80 GPa in Ti-B-N

coatings deposited by magnetron co-sputtering from Ti and TiB2 targets and having the compo-

sition 13 at.% B, 37 at.% Ti and 50 at.% N, but the hardness was unstable and decreased to about

46 GPa after several days [32]. The BN matrix can be hexagonal, amorphous or cubic [30 - 33],

or even a mixture of fcc TiN, orthorhombic TiB, c-BN and hexagonal Ti-B-N [34 - 37]. Hard-

ness of 40 - 70 GPa was reported in many papers [13 - 20] [24] [27] [33 - 38] [40 - 42]. In gen-

eral, the hardness as a function of composition did not follow the rule of mixture but displayed a

maximum at a given composition where the microstructure of the films was very uniform (no

columnar growth) and the crystallite size was a few nm. With increasing nitrogen content, when

the h-BN phase was formed, the hardness of the film strongly decreased. This decrease was at-

tributed to the formation of soft h-BN [15] [19] [22] [28] [38] [40] [43].

The purpose of our present paper is to further prove the generality of our concept for the

design of superhard nanocomposites also for this system. The thermal stability of the fully segre-

gated, binary TiN + BN system depends on the pressure of nitrogen. The decomposition TiN +

2BN → TiB2 + (3/2)N2 commences about 1350 °C at nitrogen pressure of 1 mbar and the de-

composition temperature increases to 1600 °C at P(N2) = 500 mbar. At a lower nitrogen pressure

the decomposition temperature decreases and a complex multiphase diagrams applies including a

variety of TixBy, TinNm and further mixed phases TixByNz [26]. This explains the complexity of

phases found in coatings deposited by the PVD techniques (see above) under a low and variable

nitrogen pressure or even without nitrogen. Under conditions of plasma CVD, when the nitrogen

pressure of several mbar is sufficiently high, the binary system of stoichiometric TiN and BN

should be stable. Therefore, only under these conditions the spontaneous formation of the nc-

TiN/a-BN nanocomposite should occur. By analogy with our earlier studied systems nc-TiN/a-

Si3N4, nc-W2N/a-Si3N4, nc-VN/a-Si3N4, and nc-TiN/a-Si3N4/a- & nc-TiSi2 we predicted that the

maximum hardness should be obtained at the percolation threshold when there is about one

monolayer thin continuous tissue of a-Si3N4 or a-BN between the nanocrystals of a stable transi-

tion metal nitride [6] [44]. In this paper we provide an evidence for this prediction and compare

the properties of the coatings deposited by plasma CVD with those deposited by vacuum arc

PVD.

2. Experimental

4

The nc-TiN/a-BN coatings were deposited by means of plasma CVD in a high frequency

(HF) discharge operating at 13.56 MHz and power 100 Watts. The apparatus was similar to that

described in our earlier papers [6] [44] with the following modifications: The reactor, made of

silica glass, was inserted into an electrical oven and kept at 565 °C. The substrate holder was

used as one electrode, the other one was a grounded nickel sheet around the silica tube. At a con-

stant total pressure of 3 mbar the gas flows were: N2 � 5 sccm, H2 � 50 sccm, TiCl4 � 1.7 sccm

and BCl3 was varied between 0 and 1.7 sccm. Under these conditions the deposition rate was

around 0.9 nm/s and the thickness of the coatings between 6 and 10 µm. Because the energy of

ions bombarding the growing film is under these conditions small, the biaxial compressive stress

was below about 1 GPa and, therefore, there was no hardness enhancement due to the stress in

the coatings.

The "Ti-B-N" (essentially nc-TiN/a-TiB2) coatings were deposited by vacuum arc evapo-

ration of Ti as described in [33] and introducing borazine (B3N3H6) but no nitrogen. For the two

series of samples studied here the deposition temperature was 600 and 700 °C, respectively and

the bias of the substrates � 100 V.

The crystallite size and crystalline phases were determined by means of X-Ray diffrac-

tion (XRD) using both Bragg-Brentano and glancing incidence geometry. The crystallite size

was determined by means of Warren-Averbach analysis [45] because for the largest crystallite

sizes the contribution of random strain to the Bragg peaks broadening is significant and, there-

fore, the simple Scherrer formula cannot be used. The morphology was investigated by means of

scanning electron microscopy (SEM) equipped with energy dispersive analysis of X-rays (EDX).

Because of the low sensitivity of EDX to low-Z elements and because the cut-off due to the be-

ryllium window at the energy corresponding to the B line, the elemental analysis was done by

means of elastic recoil detection (ERD) as described in [46] [55]. The annealing experiments

were done in 1 atm. of pure nitrogen (99.999) keeping the sample at the given temperature for

0.5 hour.

The hardness measurements were done by means of the automated load-depth sensing

technique as a function of the applied load and only the load independent hardness (typically

between a load of 30 to 150 mN and indentation depth ≥ 0.3 µm) was taken as reliable. The in-

dentometer Fischerscope 100 equipped with a microscope and possibility to program a series of

indentations at different chosen lateral positions on the coatings was used. The hardness values

obtained from the Fischerscope was verified by measuring the size of the remaining indentation

in SEM and calculating the hardness from the equation H = 0.927⋅L/AP where L is the applied

5

load and AP is the projected area of the indentation [47]. Further details of our hardness meas-

urements are given in [7] [48].

Leybold LH 10 surface analytic system was used for the measurement of the phase com-

position by means of X-ray photoelectron spectroscopy (XPS). The spectra were excited with the

AlKα source and recorded in ∆E = const. mode. Repetitive scans of selected spectral regions and

signal averaging were used in order to obtain a sufficient signal-to-noise ratio.

3. Results and discussion

3.1. Coatings deposited by plasma CVD

The only crystalline phase which was detected by means of XRD in all coatings dis-

cussed here was the fcc TiN showing a preferential (200) orientation. The lattice parameter was

by about ≤ 0.3 % slightly enhanced but no systematic variation with the boron content could be

found. Although the coatings had gold-like color the ERD showed a slight substoichiometry of

about 4 to 8 at. %.

In Fig. 1 we report results of two series of samples. Initially, the substrate was fixed to

the sample holder at one point only (open symbols) later by two (full symbols) which clearly

helped to improve the lateral homogeneity of the deposition and to lower the chlorine content at

high total flow rates of the chlorides. In the first series the chlorine content of the coatings in-

creased above the flow rate ratio of 0.6 to 2.5 � 4.5 at. % which deteriorated the properties. In the

second series the chlorine content was less than 1.5 at. % even at the highest chlorine flow rates

and below about 0.6 at. % for flow ratio ≤ 0.6. Although all recent coatings are deposited with

the two point fixing we include here also the older ones in order to underline the general depend-

ence of the hardness on the coverage of the surface of TiN nanocrystals by BN (we removed the

error bars for the sake of lucidity). From Fig. 1a we can see that with increasing BCl3/TiCl4 flow

ratio the hardness increases up to 50 GPa and, at even higher flow rates, it slightly decreases still

being in the range of about 40 GPa. The vertical line in Fig.1a separates the region where the

fraction of TiB2 phase measured by XPS is small and can be neglected as compared to BN (left

hand side) and that where the content of this phase increases and becomes comparable (right

hand side). The crystallite size (Fig. 1b) of 10 � 40 nm for flow ratio ≤ 0.6 is much larger than in

our other coatings (nc-TiN/a-Si3N4, nc-W2N/a-Si3N4, nc-VN/a-Si3N4, and nc-TiN/a-Si3N4/a- &

nc-TiSi2) and it decreases to 3 � 7 nm at higher BCl3 flow.

One can see from Fig. 1 that although the crystallite size and the boron content signifi-

cantly change between the BCl3/TiCl4 flow ratio of 0.2 and 1.1 the hardness does not show any

significant change in that range. The only clear change is the increase of the hardness from the

6

usual value of 23 � 25 GPa for pure TiN to 40 � 50 GPa with the addition of boron for the

BCl3/TiCl4 gas flow ratio between 0 and 0.15. This is further emphasized by Fig. 2a which

shows the hardness vs. crystallite size for coatings which were so far completely analyzed (i. e.

also by ERD and XPS, which is very much time consuming). On the other hand, if the specific

surface area is calculated from the measured crystallite size and the coverage of the surfaces of

the TiN nanocrystals is calculated from the boron content determined by ERD and XPS (see be-

low) it is seen in Fig. 2b that maximum hardness is achieved at about one monolayer coverage.

The specific surface area of the TiN nanocrystals was calculated assuming a regular

shape. This is based on our earlier investigations of the nc-TiN/a-Si3N4 [49] and nc-TiN/a-

Si3N4/a-TiSi2 [50] by means of high resolution transmission electron microscopy which reveals

fairly regular and uniform shape of the TiN nanocrystals. However, because the TiN nanocrys-

tals are randomly oriented, it is impossible to obtain image of the grain boundaries and of the ex-

act shape of the nanocrystals even when high resolution, at which one can image the TiN lattice

plains, is used (see Fig. 3 in [49]). In spite of these problems, the results shown in Fig. 2b agree

surprisingly well with those obtained earlier for the binary nc-TiN/a-Si3N4 (see Fig. 5 in [6] and

Fig. 4 in [10]), nc-W2N/a-Si3N4 (see Fig. 3 in [44]) and ternary nc-TiN/a-Si3N4/a-TiSi2 (see Fig.

3 in [10]). In spite of the limited accuracy of such calculations due to unavoidable errors of the

analyses and lack of knowledge of the exact shape of the TiN nanocrystals, this agreement is

very encouraging because it lends a strong support to our generic concept [6].

3.2. Coatings deposited by vacuum arc PVD

Also in these coatings the only crystalline phase detected by XRD was the fcc of TiN

with a preferential (220) orientation and crystallite size of 6 � 7 nm. The boron content was 22

and 24 at. % for samples from the two series studied here. In spite of such a relatively high boron

content no Bragg reflection corresponding to a TiBx phases was found. Because no nitrogen was

introduced during the deposition (i. e. all nitrogen originated from the addition of borazine) the

samples were substoichiometric nc-TiN1-x/a-TiB2 with a minor contribution of the BN phase (see

below) with about 23 - 26 at. % of N and they had metallic gray color which changed to gold

upon post-annealing in N2. The load independent hardness varied between about 45 and 55 GPa

(for an indentation depth of ≥ 0.3 µm).

Because the coatings were deposited at negative bias they had a biaxial compressive

stress of ≥ 5 GPa which relaxed upon annealing to a temperature of ≥ 800 °C. This was accom-

panied by a decrease of the measured hardness to about 30 GPa. The exact value of this decrease

is difficult to quantify because even in the pure nitrogen the residual oxygen impurity due to the

7

desorption from walls caused appreciable oxidation of the TiB2 phase, evaporation of BOx and

resultant loss of boron from the coating (see below). The nc-TiN/a-BN superhard nanocompo-

sites are more stable against oxidation than the nc-TiN/TiB2 as already found also for the nc-

TiN/a-Si3N4 [6].

3.3. XPS investigation

The XPS investigation of Ti containing coatings is subjected to an unavoidable problem

of the contamination with oxygen during the transport of the coatings from the deposition gear to

the XPS which cannot be fully removed by sputtering due to a final surface roughness. Moreo-

ver, the Ar+ sputter-cleaning results in a preferential removal of nitrogen. For example, when a

pure stoichiometric TiN, whose stoichiometry was verified by means of Rutherford backscatter-

ing spectroscopy (RBS), was sputter-cleaned by 3 keV Ar+ and, afterwards the stoichiometry

determined by means of XPS a composition of about TiN0.85 was found in our system [51]. Ac-

counting for these changes the overall composition determined by ERD and XPS agrees rea-

sonably well. However, because of these artefact we used ERD to calibrate our XPS spectra.

From Fig. 3 one can clearly see that in the nanocomposites prepared by means of plasma

CVD at a relatively high partial pressure of nitrogen of about 0.3 mbar as compared with the

PVD techniques BN is the dominant boron containing phase when the total boron content does

not exceed about 8 at.%. The TiB2 fraction can be neglected in these coatings and we can calcu-

late the coverage of the TiN nanocrystals by the amorphous BN. In the case of coatings with

higher boron content we have to take the fraction of a-BN and a-TiB2 into the calculation of the

coverage.

The assignment of the peaks to BN and TiB2 is in agreement with the data of other

authors [14] [52] [53]. The peak at a binding energy of about 186 was assigned to TiB or to bo-

ron dissolved in substoichiometric TiN1-x [54]. However, the following argument supports the

assignment of this peak to boron atoms at the TiN surface being bonded simultaneously to Ti and

N. This is supported by results shown in Fig. 4 which shows the fraction of the three phases vs.

the crystallite size. It is well known that less than about 10 � 20 nm small nanocrystals are free of

any defects because of a relatively high, destabilizing contribution of such defect to the total free

energy of such nanocrystal combined with a short diffusion length to the grain boundaries. In

Fig. 4 we cannot see any dependence of the fraction of the phases on the crystallite size although

the latter changes between about 7 and 35 nm. Therefore we attribute, at least preliminary, the

peak at Eb ≈ 186 eV to boron atoms located at the surfaces of the TiN nanocrystals.

8

The dominant boron phase in the PVD "Ti-B-N" coatings is TiB2 with a much smaller

fraction of BN and even much smaller TiB component (see Fig. 3b). This is in agreement with

the majority of results reported by other authors for Ti-B-N coatings prepared by PVD tech-

niques.

3.4. Thermal stability

The thermal stability of nc-TiN/a-BN nanocomposites deposited by plasma CVD on

stainless steel substrates is comparable to that of the nc-TiN/a-Si3N4 and nc-TiN/a-Si3N4/a-TiSi2

reported earlier [8] [9]. One example is shown in Fig. 5.

Because the crystallite size remains fairly constant upon annealing up to 900oC the de-

crease of the hardness upon annealing to ≥ 900 °C is due most probably to the diffusion of chro-

mium from the stainless steel substrate as observed by EDX. Relaxation of a compressive stress

is unlikely because the lattice parameter measured by XRD remained almost unchanged up to the

highest annealing temperature (not shown here for lack of space). In our earlier paper we have

found that the thermal stability of the previous superhard nanocomposites increased to ≥ 1100 °C

with crystallite size decreasing to about 3 nm (see Fig. 6 in [8]). The stability of the nc-TiN/a-BN

nanocomposites reported here seems to be similar for the range of crystallite sizes obtained so far

but more work is needed in order to prepare such nanocomposites with crystallite size below 5

nm.

The "Ti-B-N" coatings prepared by vacuum arc PVD and CVD from borazine show a

smaller oxidation resistance and also somewhat smaller thermal stability upon annealing in pure

nitrogen. As already mentioned, there is an indication that the somewhat larger decrease of the

hardness upon annealing to 800 °C may be due partially to the relaxation of compressive stress.

However, because the XPS showed clearly a lost of boron after the annealing of these coatings,

evidently due to the above mentioned oxidation by residual O-impurities in the oven and subli-

mation of BOx (not shown here for lack of space), we attribute the observed softening to the lost

of the TiB2 tissue. Thus, it is quite clear that the high hardness of 45 � 55 GPa in our Ti-B-N

coatings is due predominantly to the formation of nanostructure and not to enhancement caused

by energetic ion bombardment during the deposition as in the case of the ZrN/Ni, Cr2N/Ni [4]

and other so called "nanocomposites" [5] prepared by magnetron sputtering at a low pressure.

Moreover, Hammer et al. [13] have shown that in the Ti-B-N coatings deposited at room tem-

perature and having a relatively low hardness of about 20 GPa the superhardness of 40 GPa can

be achieved upon thermal post-treatment which results in a spontaneous formation of a nanos-

tructure. In many other papers there is also a contribution of the high energetic ion bombardment

9

during the deposition and the formation of compressive stress to the higher values of the reported

hardness and it is difficult to distinguish whether the nanostructure or the ion bombardment play

the decisive role (see discussion in [1]).

Conclusions

Superhard nanocomposites consisting of dominant phases nc-TiN/a-BN were prepared by

plasma CVD. Their hardness reached 45 � 50 GPa at a composition when the surface of the TiN

nanocrystals was covered by approximately one monolayer of a-BN. This lends further support

to our generic concept for the design of novel superhard nanocomposites in a variety of different

systems. "Ti-B-N" coatings deposited by PVD in combination with CVD of boron phases from

borazine contain as main boron phase TiB2. In agreement with many earlier papers they also

reach hardness of about 50 GPa but show, at least so far known, a lower oxidation stability. The

latter could be the reason for somewhat worse performance of these coatings in cutting tests un-

der conditions of hard, fast and dry machining. This is in agreement with the conclusions of [13].

Several open questions require further study. In particular it is necessary to elucidate the possi-

bilities of preparing perfectly stoichiometric superhard nc-TiN/a-BN with crystallite size less

than 5 nm and to study their oxidation resistance and thermal stability.

Acknowledgment

This work was supported by the European Commission under the project G1RD-CT99-

0222 "NACODRY" and by NATO SfP Project 972379 Protection Coatings.

10

References

[1] S. Veprek, P. Karvankova and H. Männling, MRS Symp. Proc. 697 (2002) in press.

[2] S. Veprek and A. S. Argon, Surf. Coat. Technol. 146-147 (2001) 175.

[3] S. Veprek and A. S. Argon, J. Vac. Sci. Technol. B 20 March/April (2002).

[4] P. Karvankova, H. Männling, S. Veprek, Surf. Coatings Technol. 146-147 (2001) 280.

[5] J. Musil, Surf. Coat. Technol. 125 (2000) 322.

[6] S. Veprek, S. Reiprich, Thin Solid Films 268 (1995) 64.

[7] S. Veprek, J. Vac. Sci. Technol. A 17 (1999) 2401.

[8] A. Niederhofer, P. Nesladek, H.-D. Männling, K. Moto, S. Veprek and M. Jilek, Surf. Coat-

ings Technol. 120-121 (1999) 173.

[9] H.-D. Männling, D. S. Patil, K. Moto, M. Jilek and S. Veprek, Surf. Coating Technol. 146-

147 (2001) 263.

[10] A. Niederhofer, T. Bolom, P. Nesladek, K.Moto, C. Eggs, D. S. Patil and S. Veprek, Surf.

Coat. Technol. 146-147 (2001) 183.

[11] R. A. Andrievski, I. A. Anisimova and V. P. Anisimov, Thin Solid Films 205 (1991) 171.

[12] R. A. Andrievski, J. Mater. Sci. 32 (1997) 4463.

[13] P. Hammer, A. Steiner, R. Villa, M. Baker, P. N. Gibson, J. Haupt and W. Gissler, Surf.

Coat. Technol. 68/69 (1994) 194.

[14] T. P. Mollart, M. Baker, J. Haupt, A. Steiner, P. Hammer and W. Gissler, Surf. Coat.

Technol 74-75 (1995) 491.

[15] W. Gissler, M. A. Baker, J. Haupt, P. N. Gibson,R. Gilmore and T. P. Mollart, Diamond

Film and Technol. Vol. 7, No.3 (1997) 165.

[16] P. Losbichler, C. Mitterer, P. N. Gibson, W. Gissler, F. Hofer and P. Warbichler, Surf.

Coat. Technol. 94-95 (1997) 297.

[17] M. S. Wong and Y. C. Lee, Surf. Coat. Technol. 120-121 (1999) 194.

[18] H. Deng, J. Chen, R. B. Inturi and J. A. Barnard, Surf. Coat. Technol. 76-77 (1995) 609.

[19] T. P. Mollart, J. Haupt, R. Gilmore and W. Gissler, Surf. Coat. Technol. 86-87 (1996) 231.

[20] C. Héau, R. Y. Fillit, F. Vaux and F. Pascaretti, Surf. Coat. Technol. 120-121 (1999) 200.

[21] R. Wiedemann, V. Weihnacht and H. Oettel, Surf. Coat. Technol. 116-119 (1999) 302.

[22] R. D. Ott, C. Ruby, F. Huang, M. L. Weaver and J. A. Barnard, Thin Solid Films 377-378

(2000) 602.

[23] P. H. Mayrhofer and C. Mitterer, Surf. Coat. Technol. 133-134 (2000) 131.

[24] C. Mitterer, P. H. Mayrhofer, M. Beschliesser, P. Losbichler, P. Warbichler, F. Hofer, P. N.

Gibson, W. Gissler, H. Hruby, J. Musil and J. Vlcek, Surf. Coat. Technol. 120-121 (1999) 405.

11

[25] H. Novotny, F. Benesovsky, C. Brukl and O. Schob, Mh. Chem. 92 (1961) 403.

[26] P. Rogl and J. C. Suster, eds., Phase Diagrams of Ternary Boron Nitride and Silicon Nitride

Systems, ASM International, Materials Park, Ohio 1992.

[27] L. Chaleix and J. Machet, Surf. Coat. Technol. 91 (1997) 74.

[28] J. L. He, S. Miyake, Y. Setsuhara, I. Shimizu, M. Suzuki, K. Numata and H. Saito, Wear

249 (2001) 498.

[29] M. A. Baker, T. P. Mollart, P. N. Gibson and W. Gissler, J. Vac. Sci. Technol. A 15(2)

(1997) 284.

[30] J. F. Pierson, F. Bertran, J. P. Bauer and J. Jolly, Surf. Coat. Technol. 142-144 (2001) 906.

[31] L. Maya, C. E. Vallet and J. N. Fiedor, J. Vac. Sci. Technol. A 15(4) (1997) 2007.

[32] C. Héau and J. P. Terrat, Surf. Coat. Technol. 108-109 (1998) 332.

[33] P. Holubář, M. Jílek and M. �íma, Surf. Coat. Technol. 120-121 (1999) 184.

[34] Q. Q. Yang, L. S. Wen, X. Z. Chen, Y. Z. Zheng and Y. Z. Zhuang, Vacuum Vol. 46

(1995) 181.

[35]Q. Q. Yang, L. H. Zhao and L. S. Wen, Surf. Coat. Technol. 78 (1996) 27.

[36] Y. Qiaoqin, Z. Lihua, X. Hanning, Z. Nanfang and H. Qizhong, Trans. Nonferrous Met.

Soc. China 7 (1997) 97.

[37] Q. Q. Yang, L. H. Zhao, H. Q. Du and L. S. Wen, Surf. Coat. Technol. 81 (1996) 287.

[38] C. Mitterer, M. Rauter and P. Rödhammer, Surf. Coat. Technol. 41 (1990) 351.

[39] C. Mitterer, P. Losbichler, F. Hofer, P. Warbichler, P. N. Gibson and W. Gissler, Vacuum

50 (1998) 313.

[40] C. Rebholz, H. Ziegele, A. Leyland and A. Matthews, J. Vac. Sci. Technol. A 16(5) (1998)

2851.

[41] C. Héau, N. Guillon, R. Y. Fillit and J. Machet, Surf. Coat. Technol. 97 (1997) 60.

[42] C. Pfohl and K.-T. Rie, Surf. Coat. Technol. 116-119 (1999) 911.

[43] C. Mitterer, P. Rödhammer, H. Störi and F. Jeglitsch, J. Vac. Sci. Technol. A 7(4) (1989)

2646.

[44] S. Veprek, M. Haussmann and S. Reiprich, J. Vac. Sci. Technol. A 14(1996)46.

[45] H. P. Klug and L. E. Alexander, X-Ray Diffraction Procedures, 2nd ed., J. Wiley, New York

1974.

[46] S. Veprek, A. Niederhofer, K. Moto, T. Bolom, H.-D. Männling, P. Nesladek, G. Dollinger

and A. Bergmaier, Surf. Coating Technol. 133 (2000) 152.

[47] G. Tabor, The Hardness of Metals (Clarendon Press, Oxford, 1951).

12

[48] S. Veprek, S. Mukherjee, P. Karvankova, H.-D. Männling, J. L. He, K. Moto and A. S. Ar-

gon, ICMCTF 2002, Session F1/E4, Invited Paper, Abstract No. 163.

[49] S. Christiansen, M. Albrecht, H.P. Strunk and S. Veprek, J. Vac. Sci. Technol. B 16 (1998)

19.

[50] S. Christiansen et al., unpublished results.

[51] G. Ruhl, Ph D Thesis, Technical University Munich (1996).

[52] T. A. Carlson, Photoelectron Spectroscopy, 2nd ed., Plenum Press, New York 1978.

[53] D. Briggs and M. P. Seah, Practical Surface Analysis, Vol. 1, 2nd ed., John Wiley Chichester

1990.

[54] M. A. Baker, C. Rebbholz, A. Leyland and A. Matthews, European Vacuum Conf. and 3rd

European Topical Conf. on Hard Coatings, Madrid, September 17 � 20 (2001) in press.

[55] A. Bergmaier, G. Dollinger and C. M. Frey, Nucl. Instr. and Methods B 99 (1995) 488; B

136-138 (1998) 638.

13

Figure Captions:

Fig. 1: a: Hardness and b: crystallite size determined from the Warren-Averbach analysis vs. the

chlorine gas flow ratio. The corresponding boron content is shown on the top of Fig. 1a (see

text).

Fig 2: There is only a small influence of the crystallite size on the hardness (Fig. 2a) but a clear

dependence on the surface coverage of TiN nanocrystals by BN (Fig. 2b).

Fig. 3: B 1s region of the XPS spectra of a nc-TiN/a-BN nanocomposites prepared by plasma

CVD and b of the "Ti-B-N" coatings prepared by combined vacuum arc PVD of Ti and CVD of

boron from borazine.

Fig. 4: Fraction of the BN, TiB and TiB2 phases vs. crystallite size (see text).

Fig. 5: Example of thermal stability of two nc-TiN/a-BN nanocomposite coatings deposited by

plasma CVD upon annealing in pure nitrogen for 0.5 hour for each T-step. "WA" is the crystal-

lite size determined by the Warren-Averbach analysis which, unlike the simple "Scherrer" for-

mula (with integral half width) accounts for the contribution of random strain to the broadening

of the Bragg reflections.

14

a:

Fig. 1: Hardness (Fig. 1a) and crystallite size (Fig. 1b) determined from the Warren-Averbach

analysis vs. the chlorine gas flow ratio. The corresponding boron content is shown on the top of

Fig. 1a (see text).

0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 1.105

1015202530354045505560

0 1 2 3 4 5 6 7 8 9 10 11 12 13 14

nc-TiN/a-BN

Boron Content [at.%]

Flow Ratio - BCl3/TiCl4 [-]

2 points connection 1 point connection

Har

dnes

s [G

Pa]

nc-TiN/a-BN/a-TiB2

0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 1.10

5

10

15

20

25

30

35

40

450 1 2 3 4 5 6 7 8 9 10 11 12 13 14

2 points connection1 point connection

Boron Content [at.%]

Cry

stal

lite

Size

WA

[nm

]

Flow Ratio - BCl3/TiCl4 [-]

b

15

a

b

Fig 2: There

dependence o

0 5 10 15 20 25 30 35 4005

1015202530354045505560

Hard

ness

[

GPa

]

Crystallite Size WA [nm]

is only a small influence of the crystallite size on the hardness (Fig. 2a) but a clear

n the surface coverage of TiN nanocrystals by BN (Fig. 2b).

0.0 0.5 1.0 1.5 2.0 2.5 3.005

1015202530354045505560

Hard

ness

[G

Pa]

Coverage of nc-TiN by a-BN [monolayer]

16

a

b

Fig. 3: B 1s region

CVD and b of the "T

boron from borazine

205 200 195 190 185 180

195000

200000

205000

210000

215000 TiB

TiB2

BN

Inte

nsity

(

coun

ts)

Binding Energy (eV)

Total BN TiB TiB2

of the XPS spectra of a nc-TiN/a-BN nanocomposites prepared by plasma

i-B-N" coatings prepared by combined vacuum arc PVD of Ti and CVD of

.

205 200 195 190 185 180

36000

38000

40000

42000

44000

46000

48000

50000

52000

54000

TiB

BN

TiB2

Inte

nsity

(Cou

nts)

Binding Energy (eV)

Total TiB

2 BN TiB

17

Fig. 4: Fraction of the BN, TiB and TiB2 phases vs. crystallite size (see text).

0 5 10 15 20 25 30 35 400.0

0.2

0.4

0.6

0.8

1.0

Frac

tion

of th

e Ph

ases

Crystallite Size [nm]

BN TiB TiB2

18

a

b

Fig. 5: Example of thermal stability of two nc-TiN/a-BN nanocomposite coatings deposited by

plasma CVD upon annealing in pure nitrogen for 0.5 hour for each T-step. "WA" is the crystal-

lite size determined by the Warren-Averbach analysis which, unlike the simple "Scherrer" for-

mula (with integral half width) accounts for the contribution of random strain to the broadening

of the Bragg reflections.

0 500 600 700 800 900 1000 110005

10152025303540455055

0

5

10

15

20

25

30

35

40

WA

Cry

stal

lite

Size

[nm

]

Annealing Temperature [oC]

Hard

ness

[GPa

]

Scherrer

0 500 600 700 800 900 1000 110005

10152025303540455055

0

5

10

15

20

25

30

35

40

Crys

talli

te S

ize

[nm

]

Annealing Temperature [oC]

Har

dnes

s [G

Pa]

Scherrer

WA