Nitrogen and hydrogen related infrared absorption in CVD diamond films E. Titus a,*, N. Ali a, G. Cabral a, J.C. Madaleno a, V.F. Neto a, J. Gracio a, P Ramesh Babu b, A.K. Sikder c, T.I. Okpalugo d, D.S. Misra c a Department of Mechanical Engineering, University of Aveiro, 3810-193, Portugal Materials Ireland, Polymer research Centre, School of Physics, Dublin, Ireland c Department of Physics, Indian Institute of Technology (IIT), Bombay, India d Northern Ireland Bio-Engineering Centre, NIBEC, University of Ulster, UK b Abstract In this paper, we investigate on the presence of hydrogen and nitrogen related infrared absorptions in chemical vapour deposited (CVD) diamond films. Investigations were carried out in cross sections of diamond windows, deposited using hot filament CVD (HFCVD). The results of Scanning Electron Microscopy (SEM), Fourier Transform Infrared (FTIR) and Raman spectroscopy carried out in a cross section of self-standing diamond sheets are presented. The FTIR spectra showed several features that have not been reported before. In order to confirm the frequency of nitrogen related vibrations, ab-initio calculations were carried out using GAMESS program. The investigations showed the presence of several C – N related peaks in one-phonon (1000 – 1333 cm 1). The deconvolution of the spectra in the three-phonon region (2700 – 3150 cm 1) also showed a number of vibration modes corresponding to spm CHn phase of carbon. Elastic recoil detection analysis (ERDA) was employed to compare the H content measured using FTIR technique. Using these measurements we point out that the oscillator strength of the different IR modes varies depending upon the structure and H content of CVD diamond sheets. 1. Introduction The extraordinary behaviour of nitrogen in CVD diamond is continuing to attract considerable interest from many researchers working on the CVD of diamond films and coatings. Indeed, the failure to n-doped diamond and a number of experimental results [1 – 3] showing that the single substitutional nitrogen in CVD diamond behaves rather differently than in natural diamond, have intensified this interest. It is widely accepted that nitrogen acts as an impurity in CVD diamond, which strongly affects its optical, electrical and mechanical properties. Nitrogen is one of the major impurities in natural diamond as well and it has been used as a characteristic for the classification of the same. A number of nitrogen related centers in natural as well as CVD diamond have been identified using FTIR technique [4]. In CVD diamond, hydrogen is another impurity that plays a major role in controlling various characteristics of the material. The source of hydrogen in CVD diamond lattice is mainly the high concentration ( 99% by vol.) of H2 gas present in the gas mixture. Such large concentration of H2 is required to generate high concentration of atomic hydrogen (H) during growth that is essential for the synthesis of diamond via CVD routes [5]. The study of IR absorption in C –H stretch region is important not only for identifying various possible modes of vibration, but also for the quantitative estimation of hydrogen bonded to the diamond lattice. As the absorption is due to the superposition of vibrations from different modes where carbon is in different configuration, the quantitative calculation of H content using a unique value of the oscillator strength often proves erroneous. Jacob and Unger [6] pointed out that the constant (A n ) proportional to the oscillator strength does not have a unique value and depends upon the structure of the C:H films. A n for polymer-like films with H / C ratio around 1 increases by a factor of 4 with decreasing of H / C ratio. Hence 202 order to remove the passive oxide layer of silicon. Methane (CH4) and Hydrogen (H2) were used as precursor gases. The flow of methane (1.6 sccm) and hydrogen (200 sccm) was maintained constant during sample preparation. In few cases, N2 was added in controlled manner (100 – 200 ppm) to the gas mixture in order to ascertain the origin of certain IR bands. The deposition pressure in the chamber was varied from 20 to 120 Torr in order to grow the samples with desired quality. The temperature of the substrates was maintained at 880 -C and measured using a K-type thermocouple. The thickness of the films was in the range: 30– 32 Am. A window of diameter 6 –7 mm was cut in the center of the diamond sheets deposited on silicon substrates by chemical etching [8]. This facilitates to record the IR spectra on self-standing sheets and the unwanted interference from the substrates could be easily avoided. Several sets of sheets were deposited and subjected to FTIR spectroscopy in order to check the reproducibility of the data and methodology. The IR spectra were recorded in the range 400 –4000 cm 1 with a resolution of 4 cm 1 on a Nicolet Fourier transform spectrometer in transmission mode. The Raman spectra were recorded at room-temperature using a Renishaw micro-Raman system with an excitation wavelength of 514.5 nm from an Ar+ ion laser. The diamond films were irradiated with 2.4 MeV He2+ for ERDA analysis. Depth profiles were obtained by deconvoluting the energy lost as the ions travel into and exit from the sample. for the determination of A n , the exact concentration of H atoms bonded to carbon using an independent method is required. In this paper, we report on results obtained from sample characterizations using scanning electron microscopy (SEM), elastic recoil detection analysis (ERDA), Raman and Fourier Transform Infrared (FTIR) spectroscopic techniques. The samples were cross sections of self-standing diamond sheets deposited at various pressures (20 – 120 Torr). The FTIR spectra showed several features that have not been reported earlier. In order to confirm the frequency of nitrogen related vibrations, ab-initio calculations were carried out using GAMESS program. The quality of the films was studied by the determination of its H content in the sheets. Using different hydrogen measurement technique, we point out that the oscillator strength of the different IR modes varies depending upon the structure and H content of CVD diamond sheets. 2. Experimental Diamond sheets of the thickness varying from 10 –50 Am were deposited using a HFCVD [7]. Boron doped silicon wafers with (100) orientation were used as substrates for the deposition diamond films. The substrates were polished using 2 Am diamond powder and subsequently ultrasonically cleaned in acetone. Prior to the deposition, the substrates were dipped in HF followed by de-ionized water cleaning in 3. Results and discussion Fig. 1 shows the SEM micrographs of diamond films deposited in the absence (a) and presence (b) of nitrogen. It is evident from these micrographs that both the films contain dominantly (111) oriented diamond grains. However, the nitrogen added film exhibited uniform and smaller grains below micron scale. Fig. 2 shows the baseline corrected FTIR spectra of the diamond sheets (with 200 ppm nitrogen doping at 20 Torr chamber pressure) in one phonon, two phonons and three phonon Intensity (a.u.) Fig. 1. SEM micrographs of the diamond films deposited in the absence (a) and presence (b) of nitrogen. 1000 2000 3000 -1 Wavenumbers (cm ) Fig. 2. Baseline corrected FTIR spectra of the diamond sheets (with 200 ppm nitrogen doping and 20 Torr chamber pressure) in one phonon, two phonons and three phonons regions (below 1333, 1333 – 2666 and 2667 – 3100 cm 1, respectively). 203 Table 1 Characteristic vibration frequencies observed in our films Wave number (cm 1) Mode of vibration 1109 1332 1339 2027 2161 2820 2832 2850 2880 2920 2960 3025 Nitrogen related Isolated nitrogen atoms C_C C_C C_C N – CH3 O – CH3 Sym. SP3CH2 Sym. SP3CH3 Asym. SP3CH2 Asym. SP3CH2 SP2 CH regions (below 1333, 1333 – 2666 and 2667 – 3100 cm 1, respectively). The summary of the various modes of vibrations and their frequencies in CVD diamond sheets is presented in Table 1. The IR absorption in the region below 1333 cm 1 is forbidden by symmetry in pure diamond crystal. However, due to the presence of nitrogen and hydrogen related defects in natural and CVD diamond, the IR absorption bands are visible in this region as well. In our samples one phonon region was typically masked due to the presence of the intense interference fringes. Therefore, the comparison of the sheets with different nitrogen concentration was a difficult task. However, in few of the samples where the fringes were not so strong, certain unique features in one phonon have been observed and discussed in some detail below. The expanded FTIR spectra of nitrogen un-doped (a) and doped (b) in one phonon region are shown in Fig. 3. In Fig. 3(a), it shows a dominant peak at 1332 cm 1 and a shoulder at 1109 cm 1. The intensity of the shoulder at 1109 cm 1 increased significantly (Fig. 3b) in the sheets doped with nitrogen. It appears that these two bands are related with the N impurity centers in the sheets. In natural diamond, the 1332 cm 1 band has been assigned to isolated nitrogen [9]. Some aliphatic chemical compounds [10] also show the C –N stretch mode at 1332 cm 1. However, the band at 1109 cm 1 has not been identified in diamond (CVD or natural). As the peak becomes stronger with nitrogen doping, it is likely that this Intensity (a.u.) (b) (a) 1100 1200 1300 1400 -1 Wavenumber (cm ) Fig. 3. Expanded FTIR spectra of nitrogen undoped (a) and doped (b) films in one phonon region. peak is nitrogen related. In order to check if the frequency of C –N stretching mode lies around 1332 cm 1 and reduces due to the increase in number of nitrogen atoms around carbon, abinitio calculations were carried out using GAMESS program. The Hartree Fock calculations were carried out using 6-31G (d, p) basis set. Each geometry was first optimized to obtain minimum energy configuration and then all the normal modes were obtained for the optimized geometry. The calculated frequencies (Table 2) are not only from pure C – N stretching vibrations but also have the contributions from other stretching and bending vibrations. It is observed that with increase in the number N atoms of the parent molecule N –CH3, the C – N stretching frequency decreases appreciably. Also, the environment around C atom shifts the value of C –N frequency. Thus the peak at 1109 cm 1 may be either due to the different environment around C –N-species, or larger number of N atoms bonded to C – N configuration. There is no evidence of N –O, O – H or C – O transitions indicating that O impurities are negligible in the sheets. Our ab-initio calculations were comparable with literature values and experimental data obtained in this study (Table 2). Nitrogen related centers have been reported in natural and CVD diamond by Colling et al. [11]. Major peaks corresponding to N centers, in one phonon region, are at 1280 cm 1 (a pair of nearest neighbor substitutional nitrogen atoms known as A center), 1175 cm 1 (due to an aggregate of 6– 8 N atoms known as B center), 1135 and 1344 cm 1 (due to isolated nitrogen). A few other nitrogen related defects due to platelets are also reported by Davies [12]. Table 2 C – N stretching frequencies for various geometries obtained from ab-initio calculation 204 Fig. 4. Typical FTIR spectra in CH stretch region of undoped diamond sheets grown at 120 Torr. Two phonon absorptions are intrinsic to pure diamond. The IR bands observed in the experimental spectrum (Fig. 2) at 1978, 2028 and 2161 cm 1 are due to C – C coupling. Similar bands in CVD diamond have also been reported by other workers [13]. As the absorption in the CH stretching region is very significant, the three phonon region (2700 to 3100 cm 1) has been studied in detail. The vibrations of the carbon –hydrogen bonds in N –CH3 group are also expected in this region. The spectrum in this region is superposition of the CH vibrations from spm CHn , where m, n = 1, 2, 3. Sharp absorption peaks in the CH region indicate strongly bonded hydrogen. The experimental spectrum was fitted using Gaussian peaks after background correction. Analysis of the spectra was carried out on the basis of peak assignment available in the literature [14]. The half width and amplitude of the bands were taken as fitting parameters. The frequency of the vibration shifted slightly for C –H bonds in different local environments, which enabled us to differentiate between various CHn groups. Since each individual spm CHn configuration is characterized by a specific IR absorption peak, one can use these spectral peaks to analyze the relative hybridization of the carbon atoms. The typical IR spectra in CH stretch region of un-doped diamond sheets grown at 120 Torr are shown in Fig. 4. The deconvolution of the CH band shows mainly seven peaks in this region (Table 1). Closer examination of the CH stretch region shows that diamond sheets deposited at 120 Torr also contain hydrogen bonded to sp2 carbon (3025 cm1). This is, however, considerably small compared to hydrogen bonded to sp3 carbon and is not prominent in 20 Torr sample. The dominant absorption at 2850 and 2920 cm 1 are indicative of the symmetric and asymmetric stretch bands of sp3 CH2 group, respectively. The bands at 2880 and 2960 cm 1, on the other hand, are due to symmetric and asymmetric stretch modes of sp3 bonded CH3 groups. The fitting procedure in the CH region was repeated for spectrum of samples intentionally doped with nitrogen. In addition to symmetric and asymmetric stretch bands of sp3 CH2,3 groups, peaks at 2820 appeared in the fitting procedure for nitrogen doped samples. The peak at 2820 is related to N center and there are CH stretch vibrations of the C –H bond in the N –CH3 group. This band has also been assigned to H terminated diamond (111) surface [15]. However, it is observed that the intensity of the band at 2820 cm 1 increased in the samples deposited with N2 doping. Moreover, the stretch mode of C – N bond observed at 1332 cm 1 in this experiment also supports the above statement. A less intense peak appears at 2832 cm 1 which is identified as oxygen related which appears due to the oxygen impurity. The H contents in CVD diamond sheets were measured using comparison and integration method. These methods has been established for the calculation of H content in the diamond sheets and the details of this techniques are reported elsewhere [16]. The H contents were also calculated using ERDA in few of the samples, which is a primary technique for quantifying and depth profiling of hydrogen in thin films. It has been used to depth profile, bonded and unbonded hydrogen in diamond thin films. The results of the H content in the CVD diamond sheets obtained using different methods are presented in Table 3. The significant observation is that there is a substantial difference in the value of H content obtained using comparison method and ERDA. The difference widens in the sheets grown at higher pressures. As discussed earlier, ERDA measures the total (bonded and unbonded) H content whereas the IR absorption gives only the bonded H content. This implies that there is significantly high concentration of unbonded H in diamond sheets. It mainly lies at interstitials, grain boundaries, internal voids, vacancy clusters etc. and can be easily dislocated from the diamond samples upon heating. The unbonded H has been recognized as a possible cause for the change in electrical conductivity of the CVD diamond sheets upon annealing [17]. The integrated absorbance of each band can be used to estimate the hydrogen concentration in a particular mode as the Table 3 Hydrogen content in the sheets calculated from different methods Sample Pressure (Torr) Comparison method (at.%) Integration method (at.%) ERDA (at.%) 1 2 3 4 5 6 20 40 60 80 120 140 0.051 0.032 0.055 0.056 0.064 0.077 0.072 0.025 0.188 0.258 0.329 0.595 0.18 0.24 0.36 0.64 – – 205 Table 4 Variation of proportionality constant with different modes Intensity (a.u.) 1. 20 Torr 2. 40 Torr 3. 60 Torr 4. 80 Torr 5. 120 Torr Frequency (cm 1) A n 1019 1 2 3 4 5 1300 1400 1500 1600 1700 1800 Raman shift (cm-1) Fig. 5. Raman spectra recorded in the full range of one set of diamond sheets grown with different deposition pressures ( P d). concentration of the oscillating species is proportional to the integrated intensity of the absorption band. The total hydrogen concentration (Hc) is given by the following equation: Z aðxÞd ðxÞ Hc ¼ An dx where A n is the proportionality factor and a(x) is the absorption coefficient at frequency x. A n is proportional to the inverse of oscillator strength. The difference in the value of hydrogen concentration therefore signifies the importance of the uncertainty in the value of A n. According to this measurement by the integration method, the oscillator strength should decrease as the H content in the sheets increases. The details of the above equation and calculations are discussed in detail and published elsewhere [16]. Raman spectra were recorded on all the sheets which were deposited at various conditions in order to estimate the nondiamond content in the sheets. Spectra were recorded with the use of Ar+ laser (k = 514.5 nm and power 50 mW ) in the range 1200 –1700 cm 1 with a step size of 2 cm 1. However, to resolve the fine structure of Raman diamond line, in the range of 1280 –1380 cm 1, a step size of 0.5 cm 1 was used. The spectra recorded in the full range of one set of sheets that were Fig. 6. FTIR spectra in CH stretch region of nitrogen doped diamond sheets grown at 20 Torr. 2820 2.045 2832 4.075 2850 1.085 2880 1.263 2920 6.899 2960 1.215 3020 4.645 grown with different deposition pressures ( P d), are shown in Fig. 5. A sharp Raman line at 1332.5 cm 1 was observed in all the sheets, which implies that all the sheets contain good crystalline diamond [18]. A broad band corresponding to the non-diamond impurities [19] also appeared at around 1500 cm 1 in the sheets. It was observed that the nondiamond carbon components in the sheets increased systematically with the increase of P d. It can be explained with the rate of etching and growth of sp2 and sp3 bonded carbon in the CVD environment. Atomic H is known to etch H from the growing surface and to stabilize sp3 precursors for further growth. Therefore, a continuous and sufficient supply of impingement flux density of atomic hydrogen (IFDH) on the growing surface is required in CVD process for depositing diamond. Insufficient IFDH will leave a few C – H bonds intact and hence hydrogen may get incorporated in the diamond lattice. Inside the diamond lattice, the termination of sp3 carbon bond with H may give rise to the sp2 bonding in the surrounding environment. This implies that the bonded H in the diamond films, which is a result of insufficient IFDH, will give rise to more hydrogenated carbon impurities. In HFCVD process, the dissociation of H2 into H atoms takes place in the vicinity of the hot filament. The recombination of H atoms occurs during its movement towards the substrate. The recombination rate will increase because the mean free path of H atoms decreases with increase in P d. This will result in lower IFDH at higher P d and higher non-diamond carbon as well as higher H concentration in the sheets. This hypothesis, although a simplified picture of a complex situation, can be used to explain the well observed correlation between the H content and the non-diamond carbon impurities reported by Windischmann et al. [20]. However, we have observed higher graphitic content in the intentionally nitrogen doped samples (20 Torr) despite of its sharp and well defined diamond peak (Fig. 6). The poor quality of the Fig. 7. Variation of the values of A n for different diamond sheets grown at different deposition pressures. 206 nitrogen doped CVD diamond films is also reported by Adhikari et al. [21]. It is known that Raman scattering coefficient is significantly higher for graphite than diamond [19]. Thus very small concentration of sp2 phase could be easily detected using Raman spectroscopy. This also implies that the sheets deposited without nitrogen at low pressure (< 60 Torr) where sp2 bonded phase is detected to be very weak are of high purity. On the other hand, sheets grown at high pressure show substantial concentration of sp2 phase of carbon. At the same time a strong diamond line is also evident in high pressure grown sheets. These sheets can, therefore, be treated as a composite material; a mixture of sp2 and sp3 phase of carbon [22]. As mentioned earlier, the calculation of the value of the proportionality constant A n for different modes of vibrations is quite difficult unless there is an independent technique which can calculate the exact number of H bonded in different modes. Other workers have also emphasized this difficulty in CVD diamond [22]. Using the value of H content calculated for different modes using comparison method, we have estimated the value of A n for various C – H modes in CVD diamond. These values are listed in Table 4. It is observed that A n decreased with increase of H content in the sheets. Fig. 7 shows the variation of the values of A n for different sheets which were grown at different deposition pressures. Inset shows the change in intensity ratio of nondiamond (I nd) and diamond (I d) Raman band with deposition pressure. It can be seen that there is an order of magnitude difference in the value of A n for the sheets with high and low values of I nd / I d. This implies that by knowing the quality of the films in terms of non-diamond/ diamond intensity ratio from Raman spectra, one can accordingly choose a suitable value of A n to calculate the bonded hydrogen in the film. The variation in the values of A n is expected by considering the fact that the environment surrounding each C – H mode varies significantly with deposition pressure and the structure of the sheets. 4. Conclusion Nitrogen doped diamond films exhibited a significant change in grain size and film morphology. FTIR spectroscopy has proven to be a useful tool for the analysis of nitrogen and hydrogen in CVD diamond films. Nitrogen induced films showed several additional peaks which were confirmed as nitrogen related using ab-initio calculations. Proportionality constant (A n ) was calculated for different modes of C – H vibrations and a correlation was made between the hydrogen content and A n. Hydrogen concentration (IR active) of the samples deposited at various conditions has been calculated with comparison and integration method and a comparison of these values with ERDA measurements confirmed the presence of large number of unbound hydrogen present in the CVD diamond sample. Acknowledgements FCT, Portugal and DST, India is highly acknowledged for funding this work. Authors would like to thank Dr. S. Kanagaraj for proof reading the paper. References [1] H.M. Clark, L.M. Raff, H.L. Scollt, Phys. Rev., B 54 (1996) 5914. [2] J. Xing, H.L. Scott, Phys. Rev., B 48 (1993) 4806. [3] B.R. Stones, G.H. Ma, S.D. Wolter, J.T. Glass, Phys. Rev., B 45 (1992) 11067. [4] K. Kobashi, K. Nishimura, Y. Kawate, T. Horiuchi, Phys. Rev., B 38 (1988) 4067. [5] A. Stavrides, J. Ren, M. Ho, J. Chem, Thin Solid Films 335 (1998) 27. [6] W. Jacob, M. Unger, Appl. Phys. Lett. 68 (1996) 475. [7] E. Titus, A.K. Sikder, M.K. Singh, D.S. Misra, Diamond Relat. Mater. 11 (2002) 1403. [8] E. Titus, M.K. Singh, K.N.N. Unni, P.K. Tyagi, D.S. Misra, Diamond Relat. Mater. 12 (2003) 1647. [9] R.M. Chrenko, R.M. Tuft, H.M. Strong, Nature 270 (1977) 141. [10] G. Socrates, Infrared Chareceristic Group Frequencies, Wiley, New York, 1980. [11] A.T. Collins, in: R. Messier, J.T. Glass, J.E. Butler, R. Roy (Eds.), New Diamond Science and Technology, Materials Research Society, Pittsburg, Pennsylvania, 1991, p. 659. [12] G. Davies (Ed.), Properties and Growth of Diamond, INSPEC, London, U.K., 1994. [13] A.K. Claude, M.H. Thomas, J.R. Clifford. 45 (1992) 12854. [14] Aldrich Library of FTIR spectra, Ist ed., edited by C. J. Pouchert, Aldrich Chemical Co., Milwaukee, WI (1985). [15] K.M. MacNamara, B.E. Williams, K.K. Gleason, B.E. Scruggs, J. Appl. Phys. 76 (1994) 2466. [16] E. Titus, D.S. Misra, A. Sikder, P.K. Tyagi, M.K. Singh, Abha Misra, N. Ali, G. Cabral, V.F. Neto, J. Gracio, Diamond Relat. Mater. 14 (2005) 476. [17] K. Hayashi, S. Yamanaka, H. Okushi, K. Kajimura, Appl. Phys. Lett. 68 (1996) 376. [18] H.O. Pierson, Handbook of Carbon, Graphite, Diamond and Fullerenes, Noyes Publications, New Jersey, 1993. [19] N. Wada, S.A. Solin, Physica B 105 (1981) 353. [20] H. Windischmann, G.F. Epps, Y. Cong, R.W. Collins, J. Appl. Phys. 69 (1991) 2231. [21] Sudip Adhikari, Sunil Adhikari, Masayoshi Umeno, Diamond Relat. Mater. (2005) 1824. [22] K.M. McNamara, B.E. Williams, K.K. Gleason, B.E. Scruggs, J. Appl. Phys. 76 (1994) 2466.
© Copyright 2025 Paperzz