Nuclear Instruments and Methodsin Physics Research B66(1992) 230-236 North-Holland
Nuclear Instruments &Methods MPhysics Research Section B
The structure and composition of plasma nitrided coatings on titanium H .J . Brading, P .H . Morton and T. Bell
School of Metallurgy and Materials, The University of Birmingham, Birmingham B15 27T, UK
School of Physics and SpaceResearch, 77íe University of Birmingham, Birmingham B15277, UK Titanium nitride layers on titanium metal have been produced by glow discharge plasma nitriding in nitrogen gas and the structure and composition of the layers has been investigated as a function of gas pressure and substrate temperature. Plasma nitriding is used as a surface treatment to improve the wear resistance of titanium alloys. The development of precise analytical techniques plays an important role in gaining a clearer understanding of the mechanism and kinetics of the coating formation. Light elementanalysis hasbeen carried out using RBS and NRA and nitrogen concentration depth profiles obtained to a depth of 4-5 Wm. The extent of oxygen contamination was measured using NRA and electronprobe microanalysis (EPMA) . X-ray diffraction (XRD) was used to investigate the degree of preferred orientation in the coatings and also provide information on the amount of various nitride phases (TiN, T'2 N) present in the coating. 1. Introduction Titanium alloys have the inherent advantages of light weight and high strength. However since their tribological behaviour is characterised by a high coefficient of friction and poor wear resistance, their usefulness in many engineering applications is impaired . Recently a variety of surface engineering techniques have brought about improvements in these parameters [1,21 by the production of a hard surface coating of'titanium nitride. Such protective coatings are being used increasingly on nnany industrial components - for example cutting tools and aeroengine turbine blades. The glow discharge or plasma nitriding process  makes use of an abnormal glow discharge [41, which is associated with high current and charge densities. The components to be nitrided are electrically isolated and placed or suspended in a vacuum furnace which is evacuated and back filled with the treatment gas. A do voltage is then applied between the components (cathode) and the furnace walls (anode) and the potential difference ionises the treatment gas producing a glow discharge. Positive ions in the treatment gas are accelerated towards the negatively connected components and hit the surface with high kinetic energy giving rise to sputtering of the surface, and heating of the components . In the plasma nitriding of titanium [5,61, nitrogen ions and/or accelerated neutral nitrogen atoms impinge on the surface and react to form a nitrogen rich film which results in both the formation of a compound nitride layer on the surface (S-TiN and
e-Ti ZN phases) and the diffusion of nitrogen into the substrate . This has the effect of providing the titanium with a hard surface coating which adheres well to the substrate. In order to produce titanium nitride coatings tailored to specific engineering tasks, it is necessary to characterise the coatings in terms of structure (thickness) and composition, as this allows for the optimization of the process parameters on which these depend . In this paper a number of surface analysis techniques have been applied to the characterisation of titanium nitride coatings which have been plasma nitrided with different substrate temperatures and plasma pressures. Nitrogen concentration profiles were measured to depths of 4 to 5 Rm, the oxygen concentration in the coatings was investigated and the nitride phases present were identified. 2. Experimental procedures A number of analysis techniques were used on the coatings in order to characterise different aspects of their structure. 2.1. Nitrogen concentrationprofiles RBS and NRA experiments were carried out using the 3 MV Dynamitron accelerator at the School of Physics and Space Research in the University of Birmingham. Alpha particle backscattering was used to ob-
0168-583X/92/$05.00 0 1992 - Elsevier Science Publishers B.V. All rights reserved
H.J. Brading et al. / Plasma-nitrided coatings of titanium
ZAF/4) electron probe microanalyser linked to a computer (LinkSystems series 2model 500 System running SPECTA ZAF-0/FLS Software). The electron microprobe analyser, which was run in wavelength dispersive mode (WDX) with an LDE, diffracting crystal, was used to measure the oxygen concentration in the first micron of the coatings. Measurements were made at several positions on each coating where the layer differed in colour. Counts were measured for the central channel of the oxygen Ka X-raypeak for 100 s and for 50 s at positions either side to account for background continuum radiation. These were compared to similar measurements made on a standard Si02 sample . The beam size was about 1 pm in diameter and had a 1 Rm penetration depth . A probevoltage of 10 kV and a current of 1 x 10-7 A were used . The í60(d, po)t7 0 reaction was also used to obtain an upper limit of the oxygen concentration (averaged over the first 4 to 5 Wm of the coating) by comparing the oxygen pD peak height with the nitrogen p,,2 peak height and adjusting for the difference in reaction cross section for these peaks.
tain high resolution (-0.01 wm) depth profiles of nitrogen in the first micron of the coating and the "N(d, p) t5N nuclear reaction was used to investigate nitrogen to a depth of 4 to 5 pm, which included the nitrogen diffusion zone. RBS measurements were carried out using 2.0 MeV normally incident a-particles together with a 100 Wm depletion-depth silicon surface barrier detector (collimated to 2 .5 x 10-3 sr) placed at 150 ° to the incident beam direction. Data were accumulated for a total incident charge of 4 pC. The NRA measurements  involved the use of 1.1 MeV normally incident deuterons together with a similar detection system placed at 150° to the incident beam direction (data were accumulated for a total incident charge of 150 wC). Scattered deuterons were eliminated from the detector by the use of a 12 Kin aluminium absorber foil . The NRA experimental conditions were chosen so as to allow detection of protons from the 16 0(d, po)i 70 reaction as well as from the nitrogen (d, p) and (d, a) reactions . The experimental data were analysed using computer codes developed at the School of Physics and Space Research  and cross sections taken from the compilation of Jarjis  and the following references contained therein [10-12].
2.3. X-ray diffraction measurements XRD analysis, which was carried out on a Phillips PW 1050 X-ray spectrometer, linked to a BBC Master computer, using unmonochromated Co Ka radiation, wasused to obtain information about the nitride phases present in the coatings. The specimens were scanned
2.2. Oxygen concentration measurements Electron probe microanalysis measurements were carried out on a JEOL-JXA-840A (MODEL AN 10000
.. ... .
RAW DATA - SINUWTIUN
5600 4900 4200 0 W H T
Fig. 1 . RBS spectrum from sample 1 and the simulation used to obtain a nitrogen depth profile of the first micron (see fig. 3) . Two small surface peaks corresponding to iron and copper can also be seen . (The energy width of the channels is 3.60keV.) Ill. CONTRIBUTED PAPERS
HJ Brading et al. / Plasma-nitrided coatings oftitanium
for 20 angles between 30° and 100° and the resulting X-ray diffraction peaks were identified using the JCPDS powder index files.
tions between these ratios of Ti : N it was assumed that the Ti and N number densities changed linearly . It is, however, frequently found that thin films have densities which differ significantly from that of the bulk compound  . The effect of this would be to increase the thickness of layers with a lower number density.
2.4. Analysis of RBS and (d, p) spectra Fig. 1 shows the (Y.-particle RBS spectrum obtained from sample 1 together with a simulation generated from an assumed nitrogen profile in the titanium. The process of producing a simulation is iterative. An initial nitrogen depth profile, in the form of titanium and nitrogen number densities changing with depth, is used to simulate the RBS curve ; this is compared with the raw data, and the simulation is then changed by altering the nitrogen depth profile until a good match between simulation and raw data is achieved. The emitted! particle energy spectrum resulting from deuteron bombardment of sample 4 is shown in fig . 2. The choice of results for analysis by simulation was limited to the nitrogen p, .2 peak as this was free from interference due to other peaks and cross section data for the reaction was available at this angle . The cross section used was that for the summed p, and p2 groups. The depth scale for this type of analysis depends on how well the absolute number densities of the elements within the coating are known. The number densities of the compounds TiN and Ti 2 N and of the a-titanium substrate were known , and for composi-
Sample 1 Sample 2 Sample 3 Sample 4 Sample 5 Sample 6 Sample 7 Sample 8
Table 1 Processing parameters for the eight samples investigated . At a pressure of 1 mbar the maximum temperature achievable was 710°C
All of the coatings investigated were produced by glow discharge plasma nitriding of "commercially pure"
3. Sample investigated
Gas Pressure [mbar]
Applied voltage [V]
Treatment time [h]
700 775 850 700 775 850 700 710
5 5 5 3 3 3 1 1
425 500 520 520 665 680 920 950
10 10 10 10 10 10 10 42
. .. . .. MW DATA st.M. TION
Substrate temperature [°C]
335 335 CHANNEL NUMBER
Fig. 2. NRA spectrum of sample 4 showing peaks from lead, p)170, 14 N(d, p)15N, 1 ZC(d, p) 13C and 14 N(d, ol) 12 C reactions . The N pl .2 peak has been expanded to show the simulation of the nitrogen depth profile. A small O pD peak is also visible and gives an upper limit to oxygen contamination. (The energy width of the channels is 10.61 keV.)
Cil t7 ro a ro
-1--,, 3 4
0 ( o
Ilepll, (pn ,)
Sample 5 (3
Saugdc8 (I I-, 710YJ, 42huurs, 950V)
bu, 775`C, 665V)
Samplc 6 (3mbar, 851M, 680V)
Sample 3 (5mbar, SSOT, SIOV)
Fig. 3. Nitrogen depth profiles for s mples 1 to 8 . These are a combination of the RBS profiles (first micrometer) and the NRA profiles.
1 Ucplh (I,nti
H.J. Brading et al. / Plasma-nitrided coatings of titanium
titanium in a nitrogen atmosphere, the nitriding being carried out using a 40 kW (GZM40) Kldckner lotion unit, designed specifically for the treatment of titanium components . The specimen support and current lead were made of titanium to reduce contamination of the atmosphere and specimen surface by sputtering from these components . The specimens were first heated to the treatment temperature in an inert argon atmosphere, all the heat being supplied by the glow discharge, and once the treatment temperature was achieved the argon was replaced by nitrogen at the treatment pressure. Theplasma unit used did not allow direct control of the voltage and current of the glow discharge, V and I varying with the temperature and pressure settings. The samples produced and the treatment conditions are listed in table 1. 4. Results and discussion 4.1. Nitrogen depth profiles The RBS and (d, p) spectra were measured for all the samples and analysed using the above method. The resulting RBS and (d, p) nitrogen depth profiles were then combined for each sample to give a nitrogen depth profile 4 to 5 pm deep from the(d, p) data, with a good near surface resolution (first micrometer) provided by the RBS data . The composite depth profiles obtained for the coatings are shown in fig. 3. All the profiles show the same general shape, an initial steep region where the nitrogen concentration is dropping rapidly, suggesting a thin TiN layer, a flattened region containing about 33 at .% nitrogen corresponding to the Ti ZN phase, followed by an exponential fall off in nitrogen corresponding to the region of nitrogen diffusion into (x-titanium. The solubility quoted for nitrogen in the S-TiN structure is in the range 55 to 30 at .%  . The e-Ti ZN phase exists only over a very small range of solubilities about 33 at .% of nitrogen. Nitrogen is soluble in atitanium up to 22 at.% . Several trends can be seen in these profiles. The width of the TiN layer is increased by raising the temperature and by lowering the pressure . Surface nitrogen concentrations of 59 to 63 at .% were observed on all of these samples, well above the previously assumed value corresponding to the stoichiometric nitrogen concentration for TiN (50 at.%) and also above the range of solubility quoted for nitrogen in TiN. On samples 7 and 8, thick layers (121 to 2 pm) of over stoichiometric nitride with a constant titanium :nitrogen ratio were seen. The width of the Ti ZN layer and the extent of nitrogen diffusion into the substrate both increase with the treatment temperature, but do not appear to be
11S01235mbor 775 P 2J
7 8 o-a 710 1 .-
Applied voltage (Volts)
Fig. 4. Voltage and current characteristics of the plasma unit for the substrate temperature and pressure settings used for nitriding . significantly effected by treatment pressure. The effect of temperature is to increase the rate of nitrogen diffusion through the coating, which increases the thickness of all the layers in the coating. The increase in the TiN layer thickness due to decrease in pressure and, in particular, the amount of over stoiciometric nitride, is probably dependent on the availability and energy of nitrogen ions arriving at the specimen surface. Ions and accelerated neutrals at the surface cause physical sputtering, which increases with decreasing pressure and increasing ion current (current increases with temperature). They also hit and penetrate the top most atomic layers and have an energy that depends on the applied voltage. Fig. 4 shows the current and voltage conditions that were observed forthe temperatures and pressures used in the treatments . From this it can be seen that, current increases with increasing temperature and pressure, and that voltage increases with increasing temperature and decreasing pressure . Examination of the data in fig. 3 in the light of this information indicates that it is the higher applied voltage, required to operate at lower pressures that increases the width of the TiN layers ; extra nitrogen appears to be forced into the titanium lattice by the bombardment of higher kinetic energy nitrogen ions. In fig. 3 an increase in voltage (samples 1 and 4, and again samples 2 and 5) gives rise to increased TiN layer thickness . This effect is most dramatic forsamples 7 and 8 where the applied voltage is over 900 V and where a uniform layer over 1 pm thick is produced . The overstoichiometric TiN layer on sample 7 contains 59 at .% nitrogen whereas on sample 8 it is 63 at .%. This increase in the atomic percentage of nitrogen may be due to the increase in the applied voltage measured for sample 8 or alternatively to the longer treatment time of 42 h. The RBS spectra show two small peaks due to surface contamination by iron and copper. This was temperature dependent and was therefore probably picked up during processing from sputtering of the
H.J. Bradinget al. / Plasma-nitrided coatings of titanium structure of the vacuum furnace, which contains both these elements . 4.2. Oxygen contamination measurements Oxygen concentrations measured with the electron microprobe varied between 0.14 and 1.45 wt.%, but this variation appeared to be random, having no obvious correlation with processing temperature or pressure . An average was made of the measured oxygen concentrations (0.6 t 0.4 wt .%) and this was converted
to an atomic percent by assuming the ratio of titanium to nitrogen in the first micrometer of the coatings to be 3 :2 . The average oxygen concentration was = 1.2 t 0.8 at .% and the highest was 3.1 at .%. The heights of the í60(d, pa) peaks seen on the NRA spectra also indicated an upper limit of 2 to 3 at.% for the oxygen concentration averaged over the . of the coating. the first 4 to 5 win There are several sources of this oxygen . The alloy used (commercially pure titanium - IMI 125)) contains 0.4 at .% oxygen . Also, the metal forms a surface oxide layer at room temperature . Some of this will be physically removed by sputtering in the plasma during processing. However some will also diffuse into the surface at the treatment temperature used . After processing, TiN is also known to form a Ti0Z oxide on its surface when exposed to the atmosphere at room temperature which is estimated to extend five to ten atomic layers into the surface  . 4.3. Phase identification using XRD The XRD analysis confirmed the presence of the S-TiN, e-Ti ZN and a-titanium phases in the coating (fig. 5). The diffraction patterns show an enormous increase in the height of the Ti ZN (002) peak compared with the other Ti ZN peaks with increased treatment temperature [fig. 5 (sample 3)] which indicates that there is a preferred crystalographic orientation for the Ti ZN layer . In this case, the 002 planes of the Ti Z N lining up with the surface . Fig. 5 (sample 7) shows the effect of a thick layer of over stoichiometric TiN on the diffraction patterns . The TiN peaks are broad in comparison with those for sample 1 (fig. 5) . This is most likely to be caused by the extra nitrogen atoms being forced into the Ti lattice (which has a NaCl structure). These extra nitrogen atoms may fill tetrahedral sites  rather than the octahedral sites usually filled in TiN, or they may occupy vacant sites on the titanium lattice. 5. Conclusions
Fig. 5. X-raydiffraction traces for samples 1, 3 and 7. Sample 1 shows peaks from TiN, TiZ N and a-Ti phases, sample 3 shows significant growth of the TiZN (002) peak due to the preferred crystallographic orientation in the Ti ZN layer, and sample 7 shows broadening of TiN peaks due to the presence of an over stoichiometric nitrogen concentration in this layer.
The 14N(d, p) t5N reaction and RBS a-scattering measurements have been used to measure nitrogen depth profiles to a depth of 4 to 5 Wm with a good resolution over the first micron . These profiles showed the effect of different plasma nitriding treatment parameters. Raising the substrate temperature increases the coating thickness and also increases the thickness of all the layers which make up the coating. The composition of the outermost surface of the outer TiN layer was found in all cases to be over stoichiometric, namely 59 to 63 at .% nitrogen . Lowering the treatment pressure has the effect of increasing the thickness of III. CONTRIBUTED PAPERS
H.J. Brading et al. / Plasma-nitridedcoatings of titanium
the TiN layer, and at the lowest pressure investigated (1 mbar) gave rise to layers of over stoichiometric nitride (over 1 gm in thickness) with constant Ti :N ratios. The beam parameters and detector geometries used allowed an estimate of the oxygen concentration in the coatings to be obtained in the presence of a large quantity of nitrogen. This was found to be in agreement with oxygen contamination measurements made with the electron microprobe. The level of oxygen contamination was found to be low, less than 3 at.% . The presence of the TiN, Ti2N and a titanium phases that make up the coating was confirmed by X-ray diffraction measurements . At the higher substrate temperatures, 775 and 850°C, preferred orientation of the Ti 2 N phase occurred with the (002) planes aligning themselves parallel to the coating surface. The TiN peaks were found to broaden significantly when a large proportion of over stoichiometric TiN was present in the coating. The above surface analysis techniques allow titanium nitride coatings produced by plasma nitriding a titanium substrate to be clearly characterised in terms of nitrogen depth profiles, titanium nitride compounds and the level of oxygen contamination and have revealed a compositional effect at high voltages (low pressures) not previously reported. References [1) M. Thoma, Proc. Conf. on Designing with Titanium (Institute of Metals, Bristol, 1986). [21 T. Bell, Z .L. Zhang, J. Lanagan and A.M . Staines, Surface Treatments for Corrosion and Wear Resistance, eds.
K.N . Strafford, P .K. Datta and C.G. Coogan, Chap. 12, p. 165. [31 B . Edenhofer, Heat Treatment of Metals 1 (1974) 23 . [41 A. Von Engle, Ionised Gases, 2nd ed. (Oxford University Press, London, 1965) p. 217. [51 K.T. Rie and Th. Lampe, Proc. Int. Cord. on Surface Modification of Metals by Ion Beams, Heidelberg, September 1987 . [61 T . Bell, H.W. Bergmann, J . Lanagan, P.H. Morton and A.M . Staines, Surf. Eng . 2 (2) (1986) 133 . [7) J.C . Simpson and L .G. Earwaker, Surf. and Coatings Technol. 27 (1986) 41. [81 J .C .B. Simpson and L .G. Earwaker, Nucl . Instr. and Meth . B15 (1986) 502. [91 R.A. Jadis, Nuclear Cross Section Data for Surface Analysis, vols . 1-3, University of Manchester, Department of Physics Internal Report (1979). [10) G . Debras and G . Deconninck, J. Radioanal . Chem . 38 (1977) 198. [111 G . Amsel and D . David, Phys . Rev. 4 (1969) 383. [121 V . Gomes Porto, N. Veta, R.A . Douglas, O. Sala, D. Wilmore, B.A. Robson and P.E . Hodgson, Nucl. Phys. A136 (1969) 385 . [131 M .H . Lorreto, Electron Beam Analysis of Materials (Chapman and Hall, London, 1984) . [141 Handbook of Chemistry and Physics (CRC Press, Boca Raton, 1986). [151 See, for example, Electrochem. Soc. Ext . Abstr. 3 (1966) and papers therein . [161 H.A. Wriedt and J.L. Murray, Phase Diagrams of Binary Titanium Alloys, ed. J.L . Murray (ASM Metals Park, OH, 1987). [171 M. Wittmer, J. Noser and H . Melchior, J. Appl . Phys . 52 (11) (1981) 6659. [181 R .R. Manory and G . Kimmel, Thin Solid Films 150 (1987) 277.