Synthesis of TiN thin films from titanium imido complexes

Claire J. Carmalt *a, Anne Newport a, Ivan P. Parkin *a, Philip Mountford b, Andrew J. Sealey b and Stuart R. Dubberley b
aDepartment of Chemistry, University College London, Gordon St., London, UK WC1H 0AJ
bInorganic Chemistry Laboratory, Oxford University, South Parks Road, Oxford, UK OX1 3QR

Received 23rd August 2002 , Accepted 21st October 2002

First published on 6th November 2002


Abstract

Thin films have been formed on glass by low pressure chemical vapour deposition from eleven closely related single-source precursors of generic form [TiCl2(NR)(L)x] (where x = 1, L = tridentate N-ligand; x = 2, 3, L = monodentate N-ligand). Most of the precursors formed titanium nitride films, however, bulkier imido complexes and those with chelating ligands tended to produce thin films with significant oxygen and carbon contamination, suggesting incomplete decomposition and post reaction oxidation. The best single-source precursor was found to be [TiCl2(NtBu)(py)3], which gave gold-coloured films of stoichiometric TiN1.0. Despite the coordination environment around the metal being essentially the same and the materials having comparable volatility, significant differences in film quality were observed.


Introduction

Titanium nitride films have very useful characteristics, including excellent thermal stability, resistance to chemical attack and high hardness.1,2 As a result of this combination of properties, they have applications as tribological and protective coatings.3 Titanium nitride (TiN) has a NaCl-type cubic crystal structure (Fm3m) with a bulk lattice parameter a = 4.241 Å (TiN1.0).4 The non-stoichiometric forms, TiN0.42–TiN1.2, also all have the NaCl cubic structure.5 Pure TiN1.0 films are highly reflective and gold in colour, and have found applications in jewellery and optics.6 Synthesis of these golden films is not straightforward, since contamination from carbon and oxygen leads to brown, black or grey colouration. Titanium nitride films grown from single-source precursors such as [Ti(NR2)4] (R = Me , Et) are generally contaminated with up to 40 atom% of carbon and oxygen.7

As a thin film, TiN has uses in microelectronics due to its efficiency in preventing aluminium diffusion into silicon in Al–TiN–Si trilayers at up to 550 °C.8 The resistivity of bulk TiN is 22 µΩ cm, although that of the films is higher.5 The resistivity of TiN films for barrier layer applications should be less than 600 µΩ cm.9 Generally, the resistivity of TiN films depends on microstructure, preferred orientation and stoichiometry. Optimal physical properties are obtained for fully stoichiometric TiN1.00.10 It should be noted that as the electrical resistivity of a film increases, it becomes more prone to the uptake of oxygen.11,12 However, films with low resistivities (TiN1.00) have been found to be air stable.13

Chemical vapour deposition (CVD) has proved to be a useful method for depositing thin films for many applications.14 For TiN, there are two main approaches: single source and dual source.14–17 Single-source processing of TiN involves the decomposition of a titanium complex in which the Ti is directly bonded to at least one nitrogen atom. The nitrogen atom needs to be linked to the rest of the ligand by a readily broken bond—in effect, a molecular Achilles’ heel. The rest of the ligand needs to form effective leaving groups so that carbon contamination is minimised.18,19

Titanium imido complexes have attracted much interest as precursors for single-source CVD due to the presence of the titanium–nitrogen double bond.20–24 It has been shown that the complex [Ti(NtBu)Cl2(py)3] yields conducting, gold-coloured TiN films under low pressure CVD conditions.24 This is a known complex that has been successfully synthesised by several routes.24,25

This report further examines and compares a selection of titanium imido complexes as precursors for TiN thin films using low pressure CVD techniques.

Experimental

Compounds studied

A selection of imido compounds were chosen for study as single-source precursors: [TiCl2(NtBu)(py)3], [TiCl2(NC6F5)(Me3[9]aneN3)], [TiCl2(NiPr)(NHMe2)2], [TiCl2(NC6H3Me2-2,6)(py)3], [TiCl2(NC6H3iPr2-2,6)(py)3], [TiCl2(NPh)(NHMe2)2], [TiCl2(NiPr)(Me3[9]aneN3)], [TiCl2(NC6F4H-4)(NHMe2)2], [TiCl2(NtBu)(Me3[6]aneN3)], [TiCl2(NC6H3Me2-2,6)(NHMe2)2] and [TiCl2(NiPr)(Me3[6]aneN3)].

These samples were prepared by modification of existing methods. Full details of the syntheses are reported elsewhere.26,27 The precursors in this series are all very similar, containing one strong Ti[double bond, length as m-dash]NR multiple bond, two Ti–Cl bonds and two or three Ti–N dative bonds. The geometry around the titanium is either pseudo-octahedral or trigonal bipyramidal.

General procedures

All manipulations were performed under a dry, oxygen-free dinitrogen atmosphere using standard Schlenk techniques or in an Mbraun Unilab glove box. All reagents were obtained from Aldrich and used without further purification. Microanalytical data were obtained at University College London. BDH microscope slides were used for the CVD experiments.

Physical measurements

EDAX/SEM results were obtained on a JEOL 35-CF instrument using ISIS software (Oxford Instruments). Raman spectra were acquired with a Renishaw Raman System 1000 using a helium–neon laser of wavelength 632.8 nm. The Raman system was calibrated against the emission lines of neon. Electron probe data were collected with a JXZ 8600 instrument calibrated using a PET crystal (88.03). TGA of the compounds were carried out at the Thermal Methods Laboratory at Birkbeck College (ULIRS). TGA samples were handled under nitrogen and the measurement made under flowing N2; heating rates were 10 °C min−1. UV/vis spectra were measured on a Helios λ-6 instrument. Reflectance and transmission spectra were collected using a Zeiss minature spectrometer. Sheet resistance measurements were made on a four-probe device and converted into resistivities by taking the film thickness into account.

Low pressure CVD experiments

The same procedure was followed for all the imido compounds. Approximately 0.3 g of compound was loaded into the sealed end of a glass tube (400 mm length × 16 mm diameter) in the glovebox. Glass slides (100 mm × 7 mm × 1 mm) were placed carefully along the inside of the tube. The tube was then placed in a furnace such that 30 cm was inside the furnace and the end containing the sample protruded by 4 cm. The tube was heated to a temperature of 600 °C under dynamic vacuum (0.1 Torr, measured by a Hg vacuum gauge). The tube was slowly drawn into the furnace over a period of a few minutes until the sample started to melt. It was then held at that point until all of the precursor had decomposed. Once this had happened, the furnace was allowed to cool to room temperature. The total time taken to complete the whole process was in the region of 60 min. The films deposited on the glass substrates were analysed by EDAX/SEM, electron probe and visible–NIR reflectance/transmission measurements. Glancing angle X-ray diffraction analysis of the films was carried out using a Siemens D5000 diffractometer in reflection mode (1.5° incidence) using Cu-Kα (λ = 1.5406 Å) radiation.

Results

All the [Ti(NR)Cl2(L)x] (x = 1, L = tridentate nitrogen ligand; x = 2, 3, L = monodentate nitrogen ligand) samples deposited thin films under low pressure (LP) CVD conditions at 600 °C. The films produced were highly reflective with a gold hue, although on specific samples there were large areas that were almost black (see Table 1). All the samples produced films extending over a good proportion of the length of the glass slides; between 15 and 20 cm. Scanning electron microscope (SEM) analysis shows that these films are particulate and consist of either needle-like particles or irregular shapes of a uniform size (Fig. 1). The films were not affected by immersion in acetone, ether, dichloromethane, water or isopropanol. The films were also unaffected by dilute acids. Scratching the films with a metal stylus did not damage them. Thus, the films showed good chemical resistance and mechanical toughness—properties characteristic of TiN films.
SEM image of the film deposited by LPCVD using [TiCl2(NC6H3iPr2-2,6)(py)3] as the precursor.
Fig. 1 SEM image of the film deposited by LPCVD using [TiCl2(NC6H3iPr2-2,6)(py)3] as the precursor.
Table 1 Analytical data for the films obtained by LPCVD using titanium imido complexes as precursors
Precursor Film appearancea Resistivity/µΩ cm XRD EDAX analysis (Ti, N) Calc. mass lost for TiN (%) Measured mass lost at 500 °C (%)
Phases presentb a c
a G gold; GH gold hue; Gr grey; Bl blue; B black; R reflective; D dull; thin <200 nm. b am X-Ray amorphous. c Lattice constant for TiN phase. d Lattice constant uncertain as XRD lines very broad.
[TiCl2(NtBu)(py)3] G, R 200 TiN 4.209 Ti, N 85 79.2
[TiCl2(NC6F5)(Me3[9]aneN3)] G, R thin 200 am   87 66.8
[TiCl2(NiPr)(NHMe2)2] G, R 200 TiN, TiO2 (anatase) 4.3d Ti 83 59.9
[TiCl2(NC6H3Me2-2,6)(py)3] G, B, R 320 am Ti, N 87 76.9
[TiCl2(NC6H3iPr2-2,6)(py)3] G, R 920 am Ti, N 89 72.3
[TiCl2(NPh)(NHMe2)2] G, R 13[thin space (1/6-em)]500 am Ti, N 79 64.0
[TiCl2(NiPr)(Me3[9]aneN3)] GH, Gr, D 370 am   86 76.5
[TiCl2(NC6F4H-4)(NHMe2)2] G, R 350 am   83 70.4
[TiCl2(NtBu)(Me3[6]aneN3)] GH, B, R 4800 TiN, TiO2 (anatase) 4.242 Ti, N 85 62.8
[TiCl2(NC6H3Me2-2,6)(NHMe2)2] Bl, B, R 820 TiN 4.306 Ti, N 81 71.2
[TiCl2(NiPr)(Me3[6]aneN3)] GH, B, D 1300 am Ti, N 83 71.1


The thermogravimetric analysis (TGA) under N2 at atmospheric pressure shows that at 500 °C, the precursors lost between 60 and 80% in weight, depending on the sample (Table 1). The precursor [TiCl2(NtBu)(py)3] gave a mass loss of 79.2%, with clean decomposition. This sample showed the greatest percentage mass loss and gave the best deposition characteristics. The other samples exhibited mass losses between 10 and 20% below the calculated values for production of TiN1.0, suggesting that carbon or chlorine is retained in the residue. It is possible that the required weight loss would be attained by 600 °C, although the TGA set-up allows analysis only up to 500 °C (see Table 1 and Fig. 2). Note that TGA at atmospheric pressure only gives an indication of the potential of the precursors to form bulk TiN. Any sublimation of the precursor, a desirable characteristic for low pressure CVD studies, would register as a further mass loss. Nevertheless, the TGA measurements show that all of the complexes decompose at the temperature of the CVD experiment to form a solid that is predominantly TiN, but with significant impurities in some cases.


Thermogravimetric analysis of [TiCl2(NC6H3iPr2-2,6)(py)3].
Fig. 2 Thermogravimetric analysis of [TiCl2(NC6H3iPr2-2,6)(py)3].

All of the films produced are electrically conductive. The conductivity varied along the length of the deposit. The lowest resistivity values recorded for the majority of the films were very low (less than 400 µΩ cm, see Table 1), as expected for TiN. These values were recorded in the central areas of the deposits, which coincide with the optimum deposition temperature.

The contact angles for water droplets on the surfaces of the films are excess of 100°, indicating that the films are hydrophobic. These contact angles are larger than that for the untreated glass and equivalent to the contact angles observed by us for TiN films formed by atmospheric pressure CVD of TiCl4 and NH3.

The reflectance and transmission spectra show that all the films are reflective in the near infrared. This is a known and desirable property of TiN coatings and has found an application in solar control coatings.26 The reflectance in the visible region for the majority of the films shows that visible light is transmitted. This is confirmed by the transmission spectra, which show good transmission in the visible light frequency range (Fig. 3).


Visible–NIR transmission spectrum of the film deposited by LPCVD using [TiCl2(NC6H3iPr2-2,6)(py)3] as the precursor.
Fig. 3 Visible–NIR transmission spectrum of the film deposited by LPCVD using [TiCl2(NC6H3iPr2-2,6)(py)3] as the precursor.

Compositional analysis by energy dispersive X-ray analysis (EDAX) shows the presence of titanium and nitrogen in the films. The presence of oxygen cannot be accurately determined as the films are deposited on silica glass and the accelerating voltages used give X-ray emissions from 0.5–4 µm depth from the surface. The typical film thickness produced in these LPCVD experiments was 0.5–2 µm (SEM measurements). Electron probe analysis of selected films confirmed the presence of Ti and N. However, it was clear that the nitrogen content varied from stoichiometric using [TiCl2(NtBu)(py)3] (TiN1.0) to non-stoichiometric when using [TiCl2(NPh)(NHMe2)2] and [TiCl2(NiPr)(Me3[6]aneN3)] (TiN0.2).

X-Ray photoelectron spectroscopy (XPS) analysis of selected films (Table 1) revealed the presence of titanium, nitrogen, carbon and oxygen, with the exception of the film produced from [TiCl2(NtBu)(py)3], which contained TiN1.00 as the sole phase. The N 1s binding energy shift ranges from 458.0 to 459.0 eV and that for Ti 2s ranges from 396.2 to 400.3 eV, depending on the film. These results are in agreement with the values reported for TiNx films and show a variation with stoichiometry.24 The presence of oxygen and carbon was detected in some samples, suggesting incomplete decomposition of the precursors during the LPCVD process, thereby leading to trapped impurities.

Selected films were analysed by X-ray diffraction (XRD, see Table 1). The patterns for these films showed the presence of titanium nitride and, in certain samples, anatase (TiO2). The films that contained only TiN were produced by decomposition of [TiCl2(NtBu)(py)3] and [TiCl2(NC6H3Me2-2,6)(NHMe2)2]. An additional anatase phase was detected by XRD in the films derived from [TiCl2(NiPr)(NHMe2)2] and [TiCl2(NtBu)(Me3[6]aneN3)]. The TiN phase present in the films had a lattice parameter ranging from a = 4.209 to 4.306 Å. Bulk TiN has lattice parameters of a= 4.210–4.213 Å for TiN0.42 to TiN1.6, respectively.27

Raman analysis confirmed the XRD results. The presence of TiN was detected by Raman (checked against a powder TiN standard, Aldrich) in those samples that showed TiN as the sole phase present in the XRD. Raman spectra of the other samples showed the presence of TiO2 as the anatase form. This is in agreement with the thermal analysis results that show incomplete decomposition for the majority of the samples by 500 °C, suggesting that although some TiN may be formed, there will be some residual carbon in the final films. It should be noted that with the exception of [TiCl2(NtBu)(py)3], all the samples contained detectable levels of graphitic carbon by Raman analysis.

Discussion

All of the imido samples tested showed thermal decomposition that is commensurate with the production of TiN above 500 °C. These compounds all produced particulate thin films under LPCVD conditions which adhered well to the glass substrates. All the films are electrically conducting and yield reflectance and transmission spectra which are as expected for films of TiN.

The presence of a secondary TiO2 anatase phase in some of the films indicates that low levels of oxygen somehow reached the samples either during the deposition process or on storage in air. This could originate via diffusion of oxygen from the glass substrates, as they did not have a barrier layer, however, XPS depth profiling indicated that the amount of oxide decreased with depth in the sample. This indicates that oxidation most probably takes place after film formation. The degree of oxidation is likely related to film morphology and composition. It has been observed previously that the degree of oxidation of TiN thin films is related to their exact stoichiometry. Indeed, the fully stochiometric film TiN1.0 film obtained from [TiCl2(NtBu)(py)3] showed no evidence of oxidation. The other TiN films were non-stochiometric and, hence, prone to oxidation. The presence of carbon in the samples containing anatase can be attributed to the incomplete decomposition of the precursors. The carbon is present in the form of graphite rather than a titanium carbonitride, according to the Raman analyses.

In this related series of compounds, the question arises as to what makes a good single-source precursor for TiN films? The best precursor for TiN in this study was found to be [TiCl2(NtBu)(py)3]. It is the simplest of these imido complexes, with good leaving groups surrounding the metal centre; the tBu group could conceivably leave as isobutylene and the pyridines are expected to be labile. All of the precursors investigated here have two Ti–Cl bonds, the chlorides are both readily lost during film formation as chlorine contamination of the films was not observed above detection limits (1 atom%). It is not easy to identify a facile decomposition pathway for chlorine loss as no active hydrogen is present within any of the complexes for the loss of HCl, however, the direct loss of Cl2 is possible. The bulkier complexes, especially those with Me[9]aneN3 coordinated ligands, give the poorest TGA results and the most readily identified carbon contamination of the films. This is supported by the XRD, Raman and electron probe results. From the data in Table 1, it is clear that a number of complexes are not suitable for the formation of TiN thin films; namely [TiCl2(NiPr)(Me3[9]aneN3)], [TiCl2(NC6F4H-4)(NHMe2)2] and [TiCl2(NC6F5)(Me3[9]aneN3)]. It is not clear why these samples should be particularly unsuitable precursors compared to the others.

Conclusions

A closely related series of titanium imido complexes were investigated as single-source precursors for the formation of titanium nitride thin films. All of the precursors tested under LPCVD conditions produced thin films on glass substrates at 500 °C. The nature of the films differed widely and was dependent on the precursor used. In most cases, titanium nitride was formed on the glass, however, some of the films had a black appearance due to the co-deposition of graphitic carbon. This was more prevalent in the films prepared from complexes that contained chelating nitrogen donor groups. The best single-source titanium nitride precursor examined in this study was found to be [TiCl2(NtBu)(py)3]. Despite the coordination environment around the metal being essentially the same in this related series of compounds, significant differences in the quality of the films were observed. In any case, the [TiCl2(NtBu)(py)3] complex appears to be a very good single-source precursor for TiN and an improvement over the widely used [Ti(NMe2)4].

Acknowledgement

We thank the EPSRC (GR/R16518) for financial support.

References

  1. P. Patsalas and S. Logothetidis, J. Appl. Phys., 2001, 90, 4725 CrossRef CAS.
  2. T. Takadoum, H. Houmid Bennani and M. Allouard, Surf. Coat. Technol., 1999, 88, 232 CrossRef.
  3. H. E. Hintermann, Thin Solid Films, 1981, 84, 215 CrossRef CAS.
  4. H. J. Goldschmidt, Interstitial Alloys, Butterworths, London, 1967, p. 296 Search PubMed.
  5. G. H. J. Neville, Encyclopedia of Chemical Technology, Wiley Interscience, New York, 3rd edn., 1984, vol. 23, p. 131 Search PubMed.
  6. A. J. Perry, M. Georgson and W. D. Sproul, Thin Solid Films, 1988, 157, 255 CrossRef CAS.
  7. D. M. Hoffman, Polyhedron, 1994, 13, 1169 CrossRef CAS.
  8. G. I. Grigorov, K. G. Grigorov, M. Stayanova, J. L. Vignes, J. P. Langeron and P. Denjean, Appl. Phys. A, 1993, 57, 195 CrossRef.
  9. I. J. Raaijmakers and J. Yang, Appl. Surf. Sci., 1993, 73, 31 CrossRef CAS.
  10. J. B. Price, J. O. Borland and S. Selbrede, Thin Solid Films, 1993, 236, 311 CrossRef CAS.
  11. A. Intemann, H. Koerner and F. Koch, J. Electrochem. Soc., 1993, 140, 3215 CAS.
  12. G. S. Sandhu, S. G. Meikle and T. T. Doan, Appl. Phys. Lett., 1993, 62, 240 CrossRef CAS.
  13. K. Littau, G. Dixit and R. H. Havemann, Semicond. Int., 1994, 8, 183 Search PubMed.
  14. W. S. Rees, CVD of Non-Metals, VCH, New York, 1996 Search PubMed.
  15. D. M. Hoffman, R. Fix and R. G. Gordon, Chem. Mater., 1990, 2, 235 CrossRef CAS.
  16. W. D. Munz, D. Hoffman and K. Hartig, Thin Solid Films, 1982, 96, 79 CrossRef.
  17. G. S. Elwin, PhD Thesis, University of London, 1999.
  18. C. K. Narula, P. Czubarov and D. Seyferth, Chem. Vap. Deposition, 1994, 1, 51.
  19. C. K. Narula, B. G. Demczyk, P. Czubarov and D. Seyferth, J. Am. Ceram. Soc., 1995, 78, 1247 Search PubMed.
  20. C. J. Carmalt, S. R. Whaley, P. Lale, A. H. Cowley, R. A. Jones, B. G. McBurnett and J. G. Ekerdt, J. Chem. Soc., Dalton Trans., 1998, 553 RSC.
  21. T. S. Lewkebandara, P. H. Sheridan, M. J. Heeg, A. L. Rheingold and C. H. Winter, Inorg. Chem., 1994, 33, 5879 CrossRef CAS.
  22. C. H. Winter, P. H. Sheridan, T. S. Lewkebandara, M. J. Heeg and J. W. Proscia, J. Am. Ceram. Soc., 1992, 114, 1095 Search PubMed.
  23. P. J. McKarns, G. P. A. Yap, A. L. Rheingold and C. H.Winter, Inorg. Chem., 1999, 35, 5968 CrossRef CAS.
  24. C. J. Carmalt, A. Newport, I. P. Parkin, A. J. P. White and D. J. Williams, J. Chem. Soc., Dalton Trans., 2002, 4055 Search PubMed.
  25. P. E. Collier, S. C. Dunn, P. Mountford, O. V. Shishkin and D. Swallow, J. Chem. Soc., Dalton Trans., 1995, 3743 RSC.
  26. C. M. Lappert, Thin Solid Films, 1993, 236, 6 CrossRef CAS.
  27. A. J. Blake, P. E. Collier, S. C. Dunn, W. S. Li, P. Mountford and O. V. Shishkin, J. Chem. Soc., Dalton Trans., 1997, 1549 RSC; P. J. Wilson, A. J. Blake, P. Mountford and M. Schroder, J. Organomet. Chem., 2000, 600, 71 CrossRef CAS; N. Adams, A. R. Cowley, S. R. Dubberley, A. J. Sealey, M. E. G. Skinner and P. Mountford, Chem. Commun., 2001, 2738 RSC.

This journal is © The Royal Society of Chemistry 2003
Click here to see how this site uses Cookies. View our privacy policy here.