Carbon diffusion in uncoated and titanium nitride coated ironsubstrates during microwave plasma assisted chemical vapordeposition of diamondPaul S. Weiser, Steven Prawer, Alon Hoffman, Rafael R. Manory, Peter J. K. Paterson et al. Citation: J. Appl. Phys. 72, 4643 (1992); doi: 10.1063/1.352119 View online: http://dx.doi.org/10.1063/1.352119 View Table of Contents: http://jap.aip.org/resource/1/JAPIAU/v72/i10 Published by the American Institute of Physics. Related ArticlesDiffusion and incorporation of Cd in solar-grade Cu(In,Ga)Se2 layers Appl. Phys. Lett. 99, 234101 (2011) Ion-beam mixing in crystalline and amorphous germanium isotope multilayers J. Appl. Phys. 110, 093502 (2011) Diffusion of Ag and Co in ultrafine-grained -Ti deformed by equal channel angular pressing J. Appl. Phys. 110, 083514 (2011) Diffusion of tin in germanium: A GGA+U approach Appl. Phys. Lett. 99, 162103 (2011) Ab initio calculations on Li-ion migration in Li2FeSiO4 cathode material with a P21 symmetry structure Appl. Phys. Lett. 99, 141909 (2011) Additional information on J. Appl. Phys.Journal Homepage: http://jap.aip.org/ Journal Information: http://jap.aip.org/about/about_the_journal Top downloads: http://jap.aip.org/features/most_downloaded Information for Authors: http://jap.aip.org/authors

Downloaded 31 Dec 2011 to 140.112.24.30. Redistribution subject to AIP license or copyright; see http://jap.aip.org/about/rights_and_permissions

Carbon diffusion in uncoated and titanium nitride coated iron substrates during microwave plasma assisted chemical vapor deposition of diamond

Paul S. Weiser and Steven Prawer School of Physics, University of Melbourne, Parkville, Victoria 3052, Australia

Alon Hoffman Chemistry Department, Technion, Hajia 32000, Israel

Rafael FL Manoty,a) Peter J. K. Patersorxb) and Sue-Anne Stuartb) Royal Melbourne Institute of Technology, GPO Box 2476K Melbourne, Victoria 3001, Australia

(Received 6 April 1992; accepted for publication 21 July 1992)

Auger electron spectroscopy has been employed to investigate the effectiveness of thin films of TiN as barriers to carbon diffusion during chemical vapor deposition (CVD) of diamond onto Fe substrates. Auger depth profiling was used to monitor the C concentration in the TiN layer, through the interface and into the substrate both before and after CVD diamond deposition. The

thick is sufficient to inhibit soot formation on the results show that a layer of TiN only 250 8, Fe surface and C diffusion into the Fe bulk.

INTRODUCTION

The growth of chemically vapor deposited (CVD) di- amond on many substrates, including tool materials (ce- mented carbide, SiAlON, Sic, and tungsten cutting tools) has been extensively studied in recent times.‘” Despite this, the deposition of diamond on Fe-based tools remains comparatively unreported and is poorly understood. Some authors4?’ have commented that it is difficult to grow dia- mond on Fe. The factors that have been suggested to mit- igate against good quality diamond growth on Fe (as com- pared to other metals such as tungsten), are: (i) Soot formation on the surface (presumably due to the catalytic effect of the Fe substrate), and (ii) the diffusion of C into the Fe bulk. The latter may also lead to embrittlement of the Fe substrate.

The presence of a suitable barrier layer on the Fe sub- strate may well overcome some or all of the above prob- lems. In this study, we report the effect of thin titanium nitride (TiN) layers on the nucleation and growth of dia- mond on Fe substrates. TiN was chosen because it is ex- tensively used as a hard coating on steel and cemented carbide tools and is also known to be very effective as a diffusion barrier.6 Herein, we show that the presence of even a very thin film of TiN on the Fe surface does indeed inhibit soot formation on the surface and the diffusion of C into the Fe bulk.

EXPERIMENT

The substrates were cut from a piece of magnet iron as plates of dimension 10X 10X0.7 mm3. Auger spectroscopy revealed C (<2 at. % > and Si ( - 12 at. %) as the only impurities to within the sensitivity of the Auger technique (f at. % in these samples). The samples were mechani- cally polished to a 0.5 pm diamond paste finish and then

“Metallurgical Engineering. “Department of Applied Physics.

ultrasonically cleaned in acetone and ethanol. Using a TENCOR ALPHASTEP 250 profiler, it was found that the typical surface roughness was - 0.1 pm. Reactive mag- netron sputtering was used to deposit thin TiN films 250 and 500 A thick. A mask was used to shield half the spec- imen from the TiN deposition, leaving half of the sample uncoated.

The CVD diamond deposition system consists of an Evenson Cavity encasing a 1 in. vertical quartz tube con- taining the reactant gases (methane/hydrogen mix), which are excited using a frequency of 2.45 GHz.~ Prior to insertion in the CVD system, the specimens were again thoroughly cleaned in acetone and ethanol. The deposition parameters used were: pressure=30 Torr, flow rate= 100 seem; CHJH, = 1:99, temperature =900 “C; and deposi- tion times of 7; min and 2 h. Temperature was measured using an optical pyrometer.

Auger electron spectroscopy (AES) was used to inves- tigate the near surface region ( -20 A) and the composi- tion as a function of depth of the substrates before and after CVD diamond deposition. The measurements were per- formed with a Varian Cylindrical Mirror Analyzer (CMA) with a coaxial electron gun. A primary current of 5 PA and primary electron energy of 5 keV were used. The sputtering was via argon ion bombardment with an ion energy of 2 keV and a current density of 100 ,uA/cm’. The Auger line-shape provides clear signatures for different al- lotropes of carbon* and for carbides. These signatures have been used to gain qualitative information about the nature of the bonding in the films. The concentration of an ele- ment X was estimated by

IdSx 2 rx/sx ’

all elements

where I,Y is the peak to peak intensity of the Auger line and S, is the sensitivity factor.’ The sensitivity factors for Fe, Ti, and Si were previously measured (for our analyzer) to

4643 J. Appl. Phys. 72 (IO), 15 November 1992 0021-8979/92/224643-05$04.00 @ 1992 American Institute of Physics 4643

Downloaded 31 Dec 2011 to 140.112.24.30. Redistribution subject to AIP license or copyright; see http://jap.aip.org/about/rights_and_permissions

be 0.20, 0.34, and 0.29, which are very similar to those in the literature’ from which the sensitivity of N (0.22) was taken. The sensitivity factor for C in an Fe matrix was measured using a 1 wt. % carbon-steel standard and was found to be 0.42. This is to be compared with the literature value for graphite of only 0.14,’ the difference being par- tially due to the preferential sputtering in the Fe-C system resulting from the different sputtering rates of Fe and C, and the change in chemical bonding.

Scanning electron microscopy (SEM) was employed to investigate the morphology and nucleation density of the depositions. Raman microprobe spectra were taken using the 488 nm excitation line of an argon ion laser with a spot size of l-2 pm and resolution of 6 cm-‘.

RESULTS

SEM micrographs .of the TiN/Fe interface region after 7f min and 2 h of CVD diamond deposition are shown in Figs. 1 (a)-1 (c). After only 7; min, a continuous film formed on the Fe, while there was no observable growth on the TiN. The thickness of the film was measured by surface prolilometry to be 0.4 ,um. The material was soft and easily scratched by the profilometer stylus at loads >9 mg. After 2 h [Fig. l(b)], particles formed on both the coated and uncoated surfaces, but the particles on the Fe grew on top of the continuous film that formed in the first few minutes of growth. Raman microprobe spectra of these particles on the TiN and the Fe (Fig. 2) returned similar spectra clearly showing the presence of the 1332 cm-’ diamond line, together with broad peaks at 1525 cm-’ indicative of the presence of disordered carbons. It must be noted that the diamond particles present on the uncoated Fe have grown on top of a layer of soot.

Interestingly, Fig. 1 (c) shows that the maximum nu- cleation density of diamond occurs on the TiN in the prox- imity of the interface between the coated and uncoated portions of the Fe substrate. This enhanced nucleation may be due to the sputtering of the soot formed on the uncoated Fe substrate. Sputtered Fe and C may be deposited on the TiN where they may act as centers for the nucleation of diamond. This is consistent with recent observations by Salvador-i et al. lo They show that using a solid graphite ring as the source of carbon within the hydrogen plasma, the greatest growth rate occurs in close proximity to the ring.

An Auger depth sputter profile of the uncoated Fe substrate after 7i min of CVD deposition time is shown in Fig. 3. It shows that a sputter time of more than 150 min is required to obtain the pre-CVD carbon concentration within the iron (C<2 at. %). The depth of the sputter crater was estimated by profilometry to be approximately 1 pm (10 times greater than the initial surface roughness). From these measurements, it is clear that there has been considerable diffusion of the carbon into the Fe bulk.

The inset (a) to Fig. 3 shows the Auger line shape of the carbon close to the surface, that is typical of graphitic carbon, being slightly asymmetric, with the satellite peak at - 30 eV to the left of the main peak at 272 eV. As the Fe concentration increases deeper into the sample the

FIG. 1. SEM micrographs of the TiN (250 a) and Fe interface after (a) 7; min and (b) 2 h of CVD diamond deposition at high magnification. (c) shows a low magnificktion micrograph after 2 h of CVD deposition.

C(KLL) line shape, inset (b), shows evidence of the pres- ence of carbidic carbon (presumably Fe&!) with the emer- gence of two more satellite peaks displaced 11 and 19 eV to the left of the 272 eV peak. The observed line shape, how- ever, does not appear to be one of pure carbide since it is asymmetric. This suggests that the layer is a mixture of graphitic and carbidic carbons at this depth. The line shape becomes increasingly carbidic as one sputters deeper into the film. However, once the C level reaches that observed in the virgin Fe samples, the line shape once again appears to be graphitic. This fact, together with the observation of a graphitic Auger line shape after considerable sputtering

4644 J. Appl. Phys., Vol. 72, No. 10, 15 November 1992 Weiser et al. 4644

Downloaded 31 Dec 2011 to 140.112.24.30. Redistribution subject to AIP license or copyright; see http://jap.aip.org/about/rights_and_permissions

5-T .e 5 9 ‘Z 0 a, A x .e i! -F -

I 1000 1200 1400 1600 1800

Raman Shift (cm-‘) Sputtering Time (minutes)

FIG. 2. Raman microprobe spectra of the growth on (a) Fe and (b) TiN (500 A) after 2 h of CVD diamond deposition. FIG. 4. Auger depth profile of an Fe sample coated with 250 A of TiN

after 7; min of CVD diamond deposition. The inset (a) shows the car- bidic nature of the C(KLL) line shape within the TiN.

of the surface [inset (a), Fig. 31 shows, that the observed carbidic line shape [inset (b), Fig. 31 is not an artifact produced by the Argon ion beam sputtering process.

An AES depth profile of the portion of the sample coated with 250 A of TiN on Fe after 7f min of CVD diamond deposition is shown in Fig. 4. For comparison, the AES depth profile before CVD deposition is shown in Fig. 5. Estimation of the N content of TiN films is com- plicated by the overlap of the N(KLL) and Ti(LMM) transitions. Based on the method described in Ref. 11, we estimate that the Ti/N ratio is 0.90*0.05.

The sputter profile shown in Fig. 4 is very similar to that obtained from depth profiling of the TiN layer prior to CVD deposition (Fig. 5). In particular, there is no increase in the C concentration underneath the TiN layer following exposure to the CVD plasma. The inset to Fig. 4 shows the

-2 8 E \ \UI r “\ 9 I

, 200 300

-k 20 \ L-?- eV C: *-%A-* .% ' '-.I u

- (a) 31 ---b Si 1

0 ‘t”‘~““‘lll’ 0 50 100 150

Sputtering Time (minutes)

FIG. 3. Auger depth profile of the uncoated Fe substrate after 7; min CVD deposition time. The inset shows the two C(KLL) line shapes ob- served within the sputter profile [(a) graphitic and (b) graphitic/carbidic carbon].

line shape of the carbon in the TiN layer, which is char- acteristic of a carbide (presumably Tic) ; which is present both before and after CVD deposition and is presumably attributable to carbon contamination in the TiN magne- tron sputter system. After 2 h of CVD deposition (Fig. 6), similar results were obtained with 500 A of.TiN; the AES depth profile showing no increase in the C concentration below the TiN layer following CVD diamond deposition. The inset to Fig. 6 shows the C(KLL) line shape on the surface prior to depth profiling. Similar to inset (b) of Fig. 3, the line shape appears to be a mixture of graphitic and carbidic carbons .(presumably Tic). By contrast, the Ra- man measurements of these 2 h depositions displayed a diamond and disordered carbons line shape. However, it must be recalled that the Raman microprobe measure- ments are spatially selective ( -2 pm resolution), whereas

1

2 .- .e :: e 3 & .E G 2 l?

oo,,..,,,,,,,.,.,,,,,,,,,,,,,,,,,,,,,,

80

60

20

0 0 1 2 3 4 5 6 7

Sputtering Time (minutes)

FIG. 5. Auger depth profile of an Fe sample coated with 250 A of TiN prior to CVD diamond deposition.

4645 J. Appl. Phys., Vol. 72, No. 10, 15 November 1992 Weiser et a/. 4645

Downloaded 31 Dec 2011 to 140.112.24.30. Redistribution subject to AIP license or copyright; see http://jap.aip.org/about/rights_and_permissions

.: 80

.e 0-J z E 60 g

0 .- -- 0 2 -a 6 8 IO Sputtering Time (minutes)

FIG. 6. Auger depth profile of an Fe sample coated with 500 .& of Tii after 2 h of CVD diamond deposition. The insef.shows the C(KLL) line shape of the surface prior to Auger depth profiling.

the Auger measurements involve a large electron beam ap- proximately 100 ym in diameter. Hence, the Auger spectra sample isolated particles and the surrounding substrate, while the Raman measurements show *only the spectra ob- tained from individual particles. After 6 h of growth (not shown), the diamond particles coalesce to form a contin- uous film on both the TIN coated and-uncoated Fe sub- strates. The Auger spectra of these 6 h samples display the well-known diamond line shape.

DISCUSSION

The 7: min CVD deposition onto uncoated Fe shows that in the very early stages of exposure a thick film forms on the Fe substrate. This film is graphitic in nature as has been confirmed by Auger and Raman measurements. Sur- prisingly, even though the film is of the order of- 0.4 pm thick after only 7; min of diamond deposition, a consider- able amount of Fe (up to 50 at. %) is present on the surface. It is not clear whether this is due to Fe diffusing up through the deposited layer or due to the formation of a complicated mixture of FesC and graphite. The equilib- rium phase diagram for Fe/C (Ref. 12) (which is not strictly applicable for the nonequilibrium conditions present in the microwave plasma) shows the possibility of the formation of a mixture of Fe& and graphite at 900 “C! at high carbon concentrations (025 at. %). However, since the analysis is performed at room temperature, we cannot distinguish between carbide formed at 900-C and carbide formed upon decomposition of austenite into fer- rite and Fe& during normal cooling. Formation of such a mixture, possibly concomitant with swelling, is consistent with our observations of both carbidic and graphitic car- bons in this layer; although the possibility of diffusion of Fe to the surface cannot be discounted. ,Under the plasma conditions in our CVD apparatus, significant diffusion of C into Fe was observed ( 1 pm penetration in 7; min). This large diffusion rate, despite the low concentration of C in

the source gas, is attributable to-plasma enhanced diffusion that has also been observed in the case of dc plasma im- mersion of steels in methane.13

The graphitic layer formed on the Fe is observed for depositions up to an hour in duration.14 After 2 h of CVD deposition, diamond particles nucleate and grow upon this layer, culminating in the formation of continuous films for deposition times in excess of 6 h. The appearance of the diamond particles after 2 h of growth suggests that there may be a critical thickness of graphite necessary to shield the incoming precursor gases from the catalytic surface effects of the iron. However, another possibility is that the Fe concentration at the surface is detrimental for diamond growth, and that diamond will only nucleate once the con- centration of Fe has fallen below a certain critical value that occurs once the deposited C layer is sufficiently thick.

In the absence of any TiN coating, the C diffuses deep into the Fe substrate, but after 7i min of CVD deposition on the adjoining 250 A of TiN coated region of the Fe, there is no increase in the C concentration within the Fe. Interestingly, comparison of the AES depth profiles of the 250 A TiN layers before and after 76 min of exposure to the CVD plasma, shows that the TiN layer is thinner after CVD deposition (requires two thirds-of the AES argon ion sputtering time to penetrate). Hence, the presence of even a thin layer of 250 A of TiN is remarkably effective at inhibiting C diffusion despite the fact that there is some removal of the TiN in the plasma. Similar results were obtained for the 500 A TiN layer after 2 h of CVD depo- sition, in that there was no diffusion of C through the TiN into the bulk and diamond particles were formed on the TiN.

CONCLUSION

The growth of CVD diamond on Fe in a microwave plasma is complicated by the formation of a thick graphitic film on the surface during the first few minutes of exposure to the plasma. During this time, considerable diffusion of C into the Fe substrate occurs. A thin coating of TiN (250 A) on the Fe was found to inhibit the formation of the graphitic layer and prevent C diffusion into the Fe. For longer deposition times (>2 h), diamond nucleates and grows on both the TiN and on this thick graphitic layer. However, even after 2 h CVD deposition time, no C dif- fused past the TiN layer. The results show that TiN is very effective as a diffusion barrier for diamond deposition on Fe, as well as being a surface upon which good quality diamond will nucleate and grow.

ACKNOWLEDGMENTS

The authors would like to acknowledge the support of the APIRA award held by PSW in conjunction with the support of CRA-Advanced Technical Development. We would also like to thank Peter Jewsbury and Chris Townsend of the Materials Research Laboratory (Defence Science and Technology Organization) for their assistance with the TiN coatings. The assistance of Terry Mernagh

4646 J. Appl. Phys., Vol. 72, No. 10, 15 November 1992 Weiser et al. 4646

Downloaded 31 Dec 2011 to 140.112.24.30. Redistribution subject to AIP license or copyright; see http://jap.aip.org/about/rights_and_permissions

(The Bureau of Mineral Resources, Canberra) with the Raman measurements is gratefully acknowledged.

’ N. Kikuchi, T. Komatsu, and H. Yoshimura, Mater. Sci. Eng. A 105/6, 525 (1988).

‘R. C. McCune, D. W. Hoffman, T. J. Whalen, and C. 0. McHugh, in Thin Films: Stresses and Mechanical Properties (Materials Research Society, Pittsburgh, PA, 1989), pp. 261-266.

3S. Jahanmir, D. E Deckman, L. K. Ives, A. Feldman, and E. Fara- baugh, Wear 133,73 (1989).

4P.-0. Jo&au, R. Haubner, and B. Lux, J. Ref. Hard Met. 7, 186 (1988).

‘T. P. Ong and R. P. H. Chang, Appl. Phys. Lett. 58, 358 (1991). 6J. 0. Olowolafe, J. Li, B. Blanpain, and J. W. Mayer, Appl. Phys. Lett. 57, 1307 (1990).

‘S. Prawer, A. Hoffman, S.-A. Stuart, R. Manory, P. Weiser, C. S. Lim,

,_.

. . --

J. M. Long, and F. Ninio, J. Appl. Phys. 69, 6625 (1991). *A. Hoffman, P. J. K. Paterson, and S. Prawer, Nucl. Instrum. Methods.

Phys. Res. B 51, 226 (1990). i L. .

9L. E. Davies, N. C. MacDonald, P. W. Palmberg, G. E. Riach, and R. E. Weber, Handbook of Auger Electron Spectroscopy (Physical Elec- tronics Industries, Edina, 1976), pp. 51-60.

“M C Salvadori, J. W. Ager III, I. Appl.‘Phys. Lett. 59 2386 (1991).

G. Brown, and K. M. Krishnan,

“M. Y. Al Jaroudi, H. T. G. Hen&elf, S. Gong, and A. Bengston, Thin Solid Films 195, 63 (1991).

“0. Kubbaschewski, Iron-Binaty Phase Diagrams (Springer, Berlin, 1982), p. 24.

I3 W. L. Grube and J. G. Gay, Metall. Trans. A 9, 1421 (1978). 14P. S. Weiser, S. Prawer, A. Hoffman; R. Manory, P. J. K. Paterson, and

S.-A. Stuart, in Wide Band gap Semiconductors (Materials Research Society, Pittsburgh, PA, 1992), pp. 63-68.

~_

.- : r-

I __

I)

___ .I I

. ..

-:

4647 J. Appl. Phys., Vol. 72, No: 10, 15 November 1992 Weiser ef al. 4647

Downloaded 31 Dec 2011 to 140.112.24.30. Redistribution subject to AIP license or copyright; see http://jap.aip.org/about/rights_and_permissions