CHAPTER 4 THE STUDIES OF THE CVD GROWTH PROCESS FOR EPITAXIAL DIAMOND (100) FILMS USING UHV STM

Transcription

1 CHAPTER 4 THE STUDIES OF THE CVD GROWTH PROCESS FOR EPITAXIAL DIAMOND (100) FILMS USING UHV STM 4.1 Introduction This chapter presents studies of the CVD diamond growth process using UHV STM. It has been observed that a large amount of atomic hydrogen relative to hydrocarbon gas is needed for CVD diamond growth. It has been hypothesized that atomic hydrogen etches sp 2 graphitic bonds and/or converts non-diamond carbon on the growth surface to diamond [1-3]. In order to study the CVD growth mechanism of diamond films using ultrahigh vacuum (UHV) scanning tunneling microscopy (STM), the problem was simplified by growing epitaxial films on type 2b diamond (100) substrates. Previous works have shown that CVD films grown on diamond substrates are epitaxial and much smoother than films grown on other materials [2]. For example, Silicon (Si), a popular substrate for CVD diamond growth, tends to produce polycrystalline diamond. The single crystal surface of epitaxial films has been useful for studying the growth mechanism of materials such as Si and GaAs at the atomic scale using UHV STM [4]. In this chapter, UHV STM studies of the effects of atomic hydrogen on the CVD growth process of epitaxial diamond (100) films are reported [5]. These studies involved imaging epitaxial diamond (100) films after growth with very low-level post-growth exposure to atomic hydrogen, lower exposure than the 2 minute exposure used in the studies 37

2 discussed in the previous chapter. Large area STM images show that after growth these films consist of oriented (100) planes parallel to the substrate surface. Atomic resolution images show that the surface of the (100) planes is amorphous. After approximately 2 minutes of exposure to atomic hydrogen at 30 Torr, the surface was observed to consist of a mixture of amorphous regions and (2x1) dimer reconstructed regions. After 5 minutes of exposure to atomic hydrogen, the surface was observed to consist mostly of (2x1) dimer reconstructed regions. These results support a recently proposed model for CVD diamond growth that involves the formation of a carburized layer on the diamond substrate that is converted to diamond by atomic hydrogen [3]. Also reported in this chapter is tunneling current versus voltage (I-V) spectroscopy of undoped and boron doped epitaxial CVD grown diamond (100) films. 4.2 Experiment In these experiments, the epitaxial films were grown on polished 0.25 x 1.5 x 1.5 mm 3 type 2b (100) diamond substrates purchased from Harris Corporation. The films were JURZQ WR D WKLFNQHVV RI DSSUR[LPDWHO\ P LQ WKH KRW-tungsten filament CVD reactor described in Chapter 2. The diamond films were deposited for 120 minutes at a pressure of 30 Torr using hydrogen and methane gases with flow rates of 200 sccm and 1 sccm, respectively. The substrates were heated to 800 o C using a tantalum foil as the heating element. The tungsten filament temperature was 2200 o C, as measured using a disappearing filament optical pyrometer. The growth process was terminated by first shutting off only the methane flow while maintaining the sample, filament, and H 2 settings for approximately half a minute. In contrast, the experiments in Chapter 3 had a post-growth hydrogen exposure of 2 38

3 minutes. The filament was then turned off followed by the sample heater and then the H 2 flow being turned off. After the growth process was terminated, the CVD reactor was evacuated to 1 x 10-8 Torr and the samples were transferred to the UHV STM chamber via the linear translator without exposing them to air. The images were obtained in UHV at a pressure of 1.0 x Torr. The STM tips were constructed of 20 mil tungsten wire and were electrochemically etched in a KOH solution using a manual shut-off. The probes were cleaned in UHV by heating them using the electron field emission current from a negatively biased hot-tungsten filament. Tunneling currents of 1.0 to 3.0 na with a sample bias of -800 to mv were used to obtain atomic resolution. I-V spectroscopy in UHV was obtained using the same image scan parameters. 4.3 Results and Discussion Figure 4.1 is a large area STM topograph of the epitaxial CVD diamond film after 120 minutes of growth and half a minute of post growth exposure to atomic hydrogen. The orientation of the film can easily be seen from the oriented planes and steps that are parallel to the diamond (100) substrate. These planes are similar to (100) planes observed using scanning electron microscopy [2]. At this scale of 500 nm, atomic structure is not visible but it is obvious that the CVD-grown film is uniform and oriented with respect to the diamond (100) substrate. Figure 4.2 is a topograph of the same region but at a higher resolution. The surface of the film does not display any evidence of a (2x1) dimer reconstruction; however, the surface is covered with a disordered layer of particles. This disordered layer was observed over the entire surface of the film. Figure 4.3 shows a higher resolution image of the disordered layer showing groups of disordered particles and linear rows, as indicated by the 39

4 125 nm Figure 4.1 Large area 0.5x0.5 µm 2 UHV STM image of a CVD-grown epitaxial diamond (100) film taken after the growth process and a 0.5 minute post-growth atomic hydrogen exposure. Oriented parallel planes stacked on one another and parallel to the diamond (100) substrate were observed. The sample bias and tunneling current were 3.0 V and 2.0 na, respectively. arrow. In Figure 4.3, the spacing between the particles in the linear chains were measured to be 0.35"0.01 nm and 0.26"0.01 nm along the long and short axis, respectively. These measurements are in excellent agreement with those of linear chains in amorphous carbon on diamond (100) surfaces reported in Chapter 3 [7]. The linear chains were identified by Cho et al. [6] as amorphous graphite in a random network of sp 2 and sp 3 bonded carbon atoms. Therefore, this layer is identified as amorphous carbon. Random networks of sp 2 and sp 3 bonded carbon atoms have been conjectured to form in polycrystalline diamond films based on Raman spectroscopy [7,8]. It was not possible to determine directly the thickness of this 40

5 14 nm Figure 4.2 High-resolution 55.0x55.0 nm 2 UHV STM image of the same region shown in Figure 4.1 with sample bias and tunneling current of 1.0 V and 1.5 na, respectively. The surface of the film is covered with a disordered layer of particles. The step heights are measured to be 0.09 nm, in agreement with the reported spacing between diamond (100) planes of nm. amorphous layer shown in Figure 4.2. However, the step heights in Figure 4.2 were measured to be approximately 0.09 nm, in agreement with the reported spacing between diamond (100) planes of nm. Since the single atomic steps of the (100) film can still be seen through the layer, this implies that the amorphous layer is on the order of a few monolayers thick. The sample was transferred back to the CVD reactor using the linear translator and the film was exposed to atomic hydrogen at 30 Torr for an additional two minutes. The substrate temperature was 600 o C. The CVD reactor was then evacuated to 1 x 10-8 Torr and the film was translated back to the UHV STM system where imaging was resumed. Large 41

6 1 nm Figure 4.3 Atomic resolution 7.0x7.0 nm 2 UHV STM image of the disordered layer shown in Figure 4.2 with sample voltage and tunneling current of 0.3 V and 2.0 na, respectively. Disordered groups of particles and linear rows are observed. The distance between particles is 0.25 nm. These structures and distance between particles are similar to those reported for amorphous carbon. area topographs of the surface of the film showed that the steps are still clearly visible. This shows that the hydrogen exposure did not change the macroscopic structure of the film. At higher magnification, Figure 4.4 shows that the once amorphous layer now has some regions that contain steps covered with parallel rows at equal distance along two orthogonal directions. These rows have a spacing of 0.51 nm that is twice the diamond (100) (1x1) lattice constant of 0.25 nm. As discussed in Chapter 3, previous works have used reflection high-energy electron diffraction to demonstrate that the (100) diamond surface dimerizes with a (2x1) reconstruction when exposed to atomic hydrogen [9]. The (2x1) reconstruction 42

7 Figure 4.4 High-resolution 36.5x36.5 nm 2 UHV STM image of the same diamond film shown in Figure 4.2 after an additional 2 min post-growth atomic hydrogen exposure acquired with sample bias and tunneling current of 0.7 V and 2.0 na, respectively. The surface is observed to consist of amorphous regions and rows at 90 to one another. The spacing between the rows is 0.51 nm, in agreement with the lattice constant of diamond of 0.25 nm. The inset shows an atomic resolution image of dimers within the rows. consists of rows of dimers alternating by 90 o from one step height to the next. Thus the surface of the diamond film in Figure 4.4 after an additional two minute atomic hydrogen etch has regions that are consistent with dimerization. At higher magnification, dimers could be resolved as shown in the inset of Figure 4.4. At this point, the film is about 50% amorphous and 50% dimerized. After a total of 5 minutes of exposure to atomic hydrogen, the epitaxial film has fewer amorphous areas and the surface is mostly dimerized, as shown in Chapter 3 and discussed in detail in Chapter 5. 43

8 a) b) Diamond Bulk Diamond Bulk c) d) Diamond Bulk Diamond Bulk Figure 4.5 (a) The proposed diamond growth surface after exposure to more than one monolayer of sputtered carbon. (b) The surface after etching of the disordered carbon film. (c) The surface as surface carbon species are converted to diamond by atomic hydrogen driven reactions. (d) The surface as more than one monolayer of carbon is sputtered onto the newly grown diamond surface and still carburized diamond surface. These results support a recent model by Olson et al. for CVD diamond growth that hypothesizes the formation of a carburized layer one monolayer thick with an overlayer of amorphous carbon several monolayers thick over the entire diamond substrate [3]. This model is based on sequential growth experiments described in Figure 4.5 that step the sample though separate hydrogen and carbon growth stages. In this model, it is then hypothesized that atomic hydrogen etches away the amorphous carbon layers and then converts the carburized layer into diamond. This model involves only surface reactions and no gas phase precursors. The results described in this chapter show that a thin amorphous layer does cover the entire epitaxial film after the growth process. This layer is consistent with amorphous graphite in a random network of sp 2 and sp 3 bonded carbon atoms. Also shown here is that atomic hydrogen changes this layer into a (2x1) dimer reconstructed surface. Raman 44

9 Figure 4.6 (a) Log of I vs V curve for a CVD-grown boron-doped epitaxial diamod (100) film taken with the STM tip over a dimer. Plateaus are observed for positive sample voltages consistent with tunneling into dopant-induced unoccupied states in the valence band. (b) Log of I vs V curve for an undoped CVD-grown epitaxial diamond (100) film taken with the STM tip over a dimer. Plateaus are not observed for positive sample voltages demonstrating that the conduction mechanism is different than that of boron-doped films. spectroscopy of the film after 0.5 minutes of post-growth atomic hydrogen exposure showed that the film does not contain any measurable graphitic carbon, as was previously reported in Chapter 3 for other epitaxial diamond (100) films [7]. This further substantiates that the observed amorphous carbon layer is not incorporated into the bulk during the growth process, but is etched or converted to diamond. The electronic properties of diamond films have attracted increasing interest since the possibility of diamond as a substrate for integrated circuits was realized by CVD diamond growth [2]. UHV STM I-V spectroscopy was used to investigate, with atomic resolution, the electronic structure of CVD grown boron doped and undoped epitaxial diamond (100) films. Undoped epitaxial diamond films have been observed to have a high conductivity that is p 45

10 type [10]. The conduction mechanism is unknown, but it has been hypothesized that the conductivity is due to hydrogen termination of the surface or graphitic carbon incorporated into the bulk or on the surface [10]. Figures 4.6(a) and 4.6(b) show typical I-V curves measured in UHV with the STM tip over a dimer of boron doped and undoped epitaxial diamond (100) films, respectively. The boron doped films were grown using the same CVD growth parameters as stated above with the addition of 0.5 parts per million diborane to the gas mixture and a 5 minute post atomic hydrogen exposure to remove the amorphous layer. Previous works have shown that this diborane ratio results in p type conductivity [11]. During the I-V measurements, the feedback loop was disconnected so that the tip remains fixed at a constant height while the voltage was ramped and the current was recorded using a 12 bit digitizer. Figure 4.6(a) shows the I-V curve of an epitaxial boron doped film that exhibits plateaus at positive sample bias when tunneling into the unoccupied states. Similar plateaus have been observed for p type GaAs and attributed to tunneling into dopant-induced unoccupied states in the valance band [12]. In contrast, Figure 4.6(b) shows the I-V curve of an epitaxial undoped film. The undoped film does not exhibit plateaus thus demonstrating that the p type conduction mechanism of boron doped CVD films is different than that of undoped films. A band gap is not observed in Figures 4.6(a) and 4.6(b), consistent with the reported high conductivity of CVD-grown diamond films [10]. and the previous UHV STM results of Chapter 3 [7]. 4.4 Conclusion The results described in this chapter show that a thin amorphous layer covers the entire epitaxial film after the growth process. This thin amorphous carbon layer is consistent 46

11 with turbostratic graphite in a random network of sp 2 and sp 3 bonded carbon atoms. Exposure to atomic hydrogen is observed to change this layer into a (2x1) dimer reconstructed surface. Raman spectroscopy showed that the film does not contain any measurable graphitic carbon, as was reported in Chapter 3 for other epitaxial diamond (100) films [7]. These results support a model for CVD diamond growth recently proposed by Olson et al. that involves only surface reactions and no gas phase precursors. I-V spectroscopy of undoped and boron-doped epitaxial diamond (100) films showed different p- type conduction mechanisms. 47

12 REFERENCES 1. W.A. Yarbrough and R. Messier, Current issues and problems in the chemical vapor deposition of diamond, Science 242, 688 (1990). 2. K.E. Spear and M. Frenklach in Synthetic Diamond: Emerging CVD Science and Technology, edited by K.E. Spear, and J.P. Dismukes (John Wiley & Sons, New York, (1993) 3. D.S. Olson, M.A. Kelly, S. Kapoor, and S.B. Hagstrom, A mechanism of CVD diamond film growth deduced from the sequential deposition from sputtered carbon and atomic hydrogen, J. Mater. Res. 9, 1546 (1994). 4. J.A. Strocio and W.J. Kaiser, Scanning Tunneling Microscopy, edited by J.A. Strocio and W.J. Kaiser (Academic, Boston, 1993). 5. R. E. Stallcup II, L. M. Villarreal, S. C. Lim, I. Akwani, A. F. Aviles, and J. M. Perez, Atomic structure of the diamond (100) surface studied using scanning tunneling microscopy, J. Vac. Sci. Technol. B 14, 929 (1996). 6. N.H. Cho, D.K. Veirs, J.W. Ager III, M.D. Rubin, and C.B. Hopper, Effects of substrate temperature on chemical structure of amorphous carbon films, J. Appl. Phys. 71, 2243 (1992). 7. R.E. Stallcup, A.F. Aviles, and J.M. Perez, Atomic resolution ultrahigh vacuum scanning tunneling microscopy of epitaxial diamond (100) films, Appl. Phys. Lett. 66, 2331 (1995). 8. C. Wild, N. Herres, J. Wagner, P. Koidl, and T. Anthony, Electrochem. Soc. Proc. 89, 283 (1989). 9. H. Shiomi, K. Tanabe, and N. Fujimori, Epitaxial growth of high quality diamond film by the microwave plasma-assisted chemical-vapor-deposition method, Jpn. J. Appl. Phys. 29, 34 (1990). 10. H. Shiomi, Y. Nishibayashi, and N. Fujimori, Characterization of boron-doped diamond epitaxial films, Jpn. J. Appl. Phys. 30, 1363 (1991). 11. K. Miyata, K. Kumagai, K. Nishimura, and K. Kobashi, Morphology of heavily B- doped diamond films, J. Mater. Res. 8, 2845 (1993). 48

13 12. R.M. Feenstra and J.A. Stroscio, Tunneling spectroscopy of the GaAs (110) surface, J. Vac. Sci. Technol. B 5, 923 (1987). 49