- Email: [email protected]
Early stages of nucleation and growth of diamond film by AES, SEM, UPS and optical reflectivity techniques: Surface composition
Physica B 185 (1993) North-Holland
94-98
Early stages of nucleation and growth of diamond film by AES, SEM, UPS and optical reflectivity techniques: Surface composition L. Ferrari,
S. Selci, A.C.
Felici,
M. Righini,
M.A.
Scarselli
and A. Cricenti
Istituto di Struttura della Materia, CNR, Frascati, Italy
R. Polini Dipartimento di Scienze e Tecnologie
Chimiche,
Universitci di Tar Vergata Roma, ltaly
We have investigated the morphology, chemical bonds and electronic states of CVD carbon grown on silicon (1 1 1) substrates by means of scanning electron microscopy (SEM), Auger electron spectroscopy (AES), ultraviolet photoemission spectroscopy (UPS) and optical reflectivity. Both AES and UPS techniques show variations in the observed spectra if referred to samples at different stages of growth. The optical reflectivity technique has also been used in order to study the diamond-substrate interface and to quantify the film thickness.
1. Introduction
2. Experimental
Diamond is a technologically and scientifically precious wide-gap material whose electronic, optical and tribological properties have led to a great many efforts to synthesize and characterize it. Nevertheless it is not yet completely obvious which are the mechanisms of diamond nucleation, which are the more suitable substrates for its synthesis, how preliminary treatment influences the film formation and microstructure and if there is formation of carbides or other carbonbase phase formation at the interface between the substrate and diamond. In this paper we report on observations by AES, SEM and UPS surface techniques and by the optical reflectivity interface technique in the early stage of the CVD diamond film prepared on a Si( 1 1 1) substrate.
We prepared diamond films on Si(1 1 1) substrates using the hot-filament chemical vapor deposition technique (HFCVD) [l] in a high-purity gas mixture containing molecular hydrogen and methane (2%) at a total pressure of 76 Torr (total flow = 100 seem). The silicon wafers were previously scratched with a diamond paste of $ pm size and then washed with acetone in an ultrasonic cleaner. A tantalum filament, placed 5 mm from the substrate surface, is kept at a temperature of 2450 K, as monitored by a two-color optical pyrometer, in order to activate the dissociation of reactant gases from which the C-deposition of subprocess takes place. The temperature strate was 1023 K, as monitored by a thermocouple pressed against the sample. We have employed several techniques to investigate the surface and interface composition and structure. Prior to the characterization, each sample was exposed to air.
Correspondence Materia, CNR,
0921-4526/93/$06.00
to: L. Frascati,
0
Ferrari, Italy.
Istituto
1993 - Elsevier
di Struttura
Science
della
Publishers
B.V. All rights
reserved
L. Ferrari et al. I Nucleation and growth of diamond films
Auger spectroscopy is a high surface sensitive technique because of the limited mean free path of electrons, and that makes AES one of the most suitable techniques for investigating the first layers at the surface (-10 A). We performed Auger analysis in an UHV chamber at a pressure better than 5 X lo-‘” Torr, with a double pass cylindrical mirror analyzer (CMA) for electron analysis, with an energy resolution of 0.6%. The Auger dN(E)ldE mode spectra were taken with an incident electron beam energy of 3 keV, and the beam current ranged from 1 PA to 3 FA to reduce possible electron beam damage. A 3 V peak to peak modulation voltage was used. We analyzed the same samples by means of core levels and valence band photoemission spectroscopy, using synchrotron radiation at the Grasshopper beam line of the Adone storage ring in Frascati. The experiment was carried out in a UHV chamber at a pressure of 5 X lo-” Torr. The valence band spectra were taken at a photon energy of 60 eV and pass energy of 25 eV, with a total energy resolution of 0.5 eV. A normal incidence reflectivity spectrum on a CVD diamond sample is obtained with a cross double-beam reflectometer. A light beam from a tungsten lamp is split by a beam splitter to produce signal and reference beams. The two reflected lights are detected by an optical multichannel analyzer (OMA, by EG&G) with a resolution of 0.07 nm. Four shutters allow the intensities of the two beams and their respective background to be measured separately. We measured the ratio between the reflectance from our sample and a reference mirror, after background subtraction, thus removing the features due only to the optical system. We have also observed a diamond sample by scanning electron microscopy (Stereo Scan 260 by Cambridge Instruments Ltd) at a pressure of 1O-6 Torr.
3. Results and discussion Figure la shows a SEM image of a sample grown for 5 min and fig. lb shows a micrograph of a similar sample after a 20-min growth. Here
95
Fig. la. The surface of a sample 5 min after the beginning of the C deposition process, as viewed by SEM.
Fig. lb. The surface of a similar sample after a 20-min growth as viewed by SEM.
one can check the evolution of the surface morphology of the film during the synthesis. In the first case, well-separated diamond crystallites (0.1 urn in the mean diameter) are just formed on the silicon substrate with a density of 5 x lo8 nuclei/cm’. In the following case (fig. lb) we observe that 20 min after the beginning of the process the crystals increase their size (0.5 pm)
L. Ferrari et al. I Nucleation and growth of diamond films
:
thermodynamic conditions. Two peaks at 25.5 and 248 eV with equivalent intensity are recorded. The difference between the two samples can be related to a change in the density of surface state of diamond, resembling AES lines for different diamond surface reconstruction [2]. The valence band electronic states distribution of the two diamond samples, examined by synchrotron radiation at a photon energy of 60 eV (figs. 4(a),(b)), shows that the valence band is typical [3,5,6] for diamond (1 1 1); for the sample grown for 20 min (fig. 4(b)) we observe a
:
: : : i, ; ‘3
zc!ii KineticEnergy Fig. 2. KLL carbon sample surface.
Auger
spectrum
[ev] of the 5-min deposited
until the silicon substrate is almost entirely covered. Some AES spectra of the samples grown for 5 min are reported in fig. 2, where the KLL C lines reveal two strong features at 258 and 248 eV. The lines shapes and the relative intensities are very close to those expected for a clean (1 1 1) natural diamond surface [2-41. So, just 5 min after the beginning of the carbon deposition process, we detect the diamond phase occurrence by AES. We observe the AES carbon line shape, shown in fig. 3, 20min after the beginning of the growth under the same
: : : : : : : : :: :: ii 22u
240
Kinetic Energy Fig. 3. KLL carbon Auger
growth
chamber
for 20 min.
-15
-5
0 = VBM
INITIAL STATE ENERGY (eV)
lb) -25
-15
-5
I
= VBM
INITIAL STATE ENERGY (eV)
aa
280
-25
[eVl
line of the sample
exposed
in the
Fig. 4. Photoemission electron distribution curve of the valence band from a diamond surface for the S-min deposited sample (a) and for the sample exposed in the growth chamber for 20min (b), for photon energy hu = 60eV. Energies are referenced to the valence band maximum.
L. Ferrari et al. I Nucleation and growth of diamond films
shift in energy position (0.6 eV) that can signify, for instance, an increase in the band bending at the surface due to a different reconstruction 13961. In order to exclude that this shift is due to sample charging, we have monitored the Si 2p core level peaks for different photon energies and different slit widths of the Grasshopper monochromator, observing no energy shift for the core line. Si 2p spectra recorded at a photon energy of 150 eV indicate the presence of hydrated silica (Si-OH) on the substrate surface, at a binding energy of 105 eV [7]. In fig. 5, we show the Si 2p spectrum from our diamond sample grown for 5 min. Some results of normal incident reflectance applied to the CVD diamond sample grown for 20 min are also reported. In fig. 6, we report the optical reflectivity spectrum obtained with our cross double beam reflectometer in the 1.73.7 eV spectral range. In this range of photon energy, diamond is transparent and we observe interference peaks due to the presence of the silicon-diamond interface and of the nearly homogeneous diamond film. In order to explain this spectrum, we have calculated, using the Abel& matrix derived method [S-lo], the reflectance behavior of a system composed of one layer of diamond on a Si substrate. The model computed reflectance, obtained using a layer of
1
97
-BEST FIT CVD DIAMOND
, 1.7
2.2
2.7
3.2
PHOTON ENERGY
Fig. 6. Reflectivity of diamond (dotted line) and the calculated and a 25OOA thick diamond
\J
1 3.7
(ev)
film on a silicon substrate spectrum for a Si substrate
layer
(solid line).
2500 A of diamond on a silicon substrate, reproduces the experimental spectrum in a satisfying way. Normal incident reflectance offers information about the interface between the diamond surface and the Si substrate, and this technique allows one to obtain complementary results with respect to the AES and UPS techniques. This model becomes critical in the high energy range, above 4 eV, where a tungsten light source did not allow us to have a good signal/noise ratio. Where we can merely say that the material grown on silicon is a 2500 A thick layer of diamond, the possibility of the presence of a very thin layer of a second material, such as Sic, is under investigation, opening up the possibility of diamond-substrate interface studies using optical probes.
References
9s
100
105
110
BINDING ENERGY
Fig. 5. Si 2p core levels spectrum 5 min.
115
120
(eV)
of a sample
grown
for
(I] P. Ascarelli, E. Molinari, R. Polini, V. Sessa, M.L. Terranova, E. Cappelli and S. Fontana, in: Diamond and Diamond-Like Films and Coatings, eds. J.C. Angus, R.E. Clausing, L.L. Horton and P. Koidl (Plenum Press, New York, 1991). [2] L. Ferrari, E. Cappelli, A. Cricenti, S. Selci, R. Polini and G. Chiarotti, Appl. Surf. Sci. 56-58 (1992) 100. 131 B.B. Pate, Surf. Sci. 165 (1986) 83. [4] P.G. Lurie and J.T. Wilson, Surf. Sci. 65 (1977) 476.
98
L. Ferrari et al. I Nucleation and growth of diamond films
[5] F.J. Himpsel, J.F. van der Veen and D.E. Eastman, Phys. Rev. B 22 (1980) 1967. [6] F.J. Himpsel, D.E. Eastman and J.F. van der Veen, J. Vat. Sci. Technol. 17 (1980) 1085. (71 E. Paparazzo, M. Fanfoni, E. Severini and S. Priori, J. Vat. Sci. Technol. A 10 (1992) 2892.
[8] N. Born and E. Wolf, Principles of Optics (Pergamon Press, New York, 1965). [9] C.S. Heavens, Optical Properties of Thin Solid Films (Butterworth, London, 1955). [lo] S. Nannarone and S. Selci, Phys. Rev. B 28 (1983) 5930.
94-98
Early stages of nucleation and growth of diamond film by AES, SEM, UPS and optical reflectivity techniques: Surface composition L. Ferrari,
S. Selci, A.C.
Felici,
M. Righini,
M.A.
Scarselli
and A. Cricenti
Istituto di Struttura della Materia, CNR, Frascati, Italy
R. Polini Dipartimento di Scienze e Tecnologie
Chimiche,
Universitci di Tar Vergata Roma, ltaly
We have investigated the morphology, chemical bonds and electronic states of CVD carbon grown on silicon (1 1 1) substrates by means of scanning electron microscopy (SEM), Auger electron spectroscopy (AES), ultraviolet photoemission spectroscopy (UPS) and optical reflectivity. Both AES and UPS techniques show variations in the observed spectra if referred to samples at different stages of growth. The optical reflectivity technique has also been used in order to study the diamond-substrate interface and to quantify the film thickness.
1. Introduction
2. Experimental
Diamond is a technologically and scientifically precious wide-gap material whose electronic, optical and tribological properties have led to a great many efforts to synthesize and characterize it. Nevertheless it is not yet completely obvious which are the mechanisms of diamond nucleation, which are the more suitable substrates for its synthesis, how preliminary treatment influences the film formation and microstructure and if there is formation of carbides or other carbonbase phase formation at the interface between the substrate and diamond. In this paper we report on observations by AES, SEM and UPS surface techniques and by the optical reflectivity interface technique in the early stage of the CVD diamond film prepared on a Si( 1 1 1) substrate.
We prepared diamond films on Si(1 1 1) substrates using the hot-filament chemical vapor deposition technique (HFCVD) [l] in a high-purity gas mixture containing molecular hydrogen and methane (2%) at a total pressure of 76 Torr (total flow = 100 seem). The silicon wafers were previously scratched with a diamond paste of $ pm size and then washed with acetone in an ultrasonic cleaner. A tantalum filament, placed 5 mm from the substrate surface, is kept at a temperature of 2450 K, as monitored by a two-color optical pyrometer, in order to activate the dissociation of reactant gases from which the C-deposition of subprocess takes place. The temperature strate was 1023 K, as monitored by a thermocouple pressed against the sample. We have employed several techniques to investigate the surface and interface composition and structure. Prior to the characterization, each sample was exposed to air.
Correspondence Materia, CNR,
0921-4526/93/$06.00
to: L. Frascati,
0
Ferrari, Italy.
Istituto
1993 - Elsevier
di Struttura
Science
della
Publishers
B.V. All rights
reserved
L. Ferrari et al. I Nucleation and growth of diamond films
Auger spectroscopy is a high surface sensitive technique because of the limited mean free path of electrons, and that makes AES one of the most suitable techniques for investigating the first layers at the surface (-10 A). We performed Auger analysis in an UHV chamber at a pressure better than 5 X lo-‘” Torr, with a double pass cylindrical mirror analyzer (CMA) for electron analysis, with an energy resolution of 0.6%. The Auger dN(E)ldE mode spectra were taken with an incident electron beam energy of 3 keV, and the beam current ranged from 1 PA to 3 FA to reduce possible electron beam damage. A 3 V peak to peak modulation voltage was used. We analyzed the same samples by means of core levels and valence band photoemission spectroscopy, using synchrotron radiation at the Grasshopper beam line of the Adone storage ring in Frascati. The experiment was carried out in a UHV chamber at a pressure of 5 X lo-” Torr. The valence band spectra were taken at a photon energy of 60 eV and pass energy of 25 eV, with a total energy resolution of 0.5 eV. A normal incidence reflectivity spectrum on a CVD diamond sample is obtained with a cross double-beam reflectometer. A light beam from a tungsten lamp is split by a beam splitter to produce signal and reference beams. The two reflected lights are detected by an optical multichannel analyzer (OMA, by EG&G) with a resolution of 0.07 nm. Four shutters allow the intensities of the two beams and their respective background to be measured separately. We measured the ratio between the reflectance from our sample and a reference mirror, after background subtraction, thus removing the features due only to the optical system. We have also observed a diamond sample by scanning electron microscopy (Stereo Scan 260 by Cambridge Instruments Ltd) at a pressure of 1O-6 Torr.
3. Results and discussion Figure la shows a SEM image of a sample grown for 5 min and fig. lb shows a micrograph of a similar sample after a 20-min growth. Here
95
Fig. la. The surface of a sample 5 min after the beginning of the C deposition process, as viewed by SEM.
Fig. lb. The surface of a similar sample after a 20-min growth as viewed by SEM.
one can check the evolution of the surface morphology of the film during the synthesis. In the first case, well-separated diamond crystallites (0.1 urn in the mean diameter) are just formed on the silicon substrate with a density of 5 x lo8 nuclei/cm’. In the following case (fig. lb) we observe that 20 min after the beginning of the process the crystals increase their size (0.5 pm)
L. Ferrari et al. I Nucleation and growth of diamond films
:
thermodynamic conditions. Two peaks at 25.5 and 248 eV with equivalent intensity are recorded. The difference between the two samples can be related to a change in the density of surface state of diamond, resembling AES lines for different diamond surface reconstruction [2]. The valence band electronic states distribution of the two diamond samples, examined by synchrotron radiation at a photon energy of 60 eV (figs. 4(a),(b)), shows that the valence band is typical [3,5,6] for diamond (1 1 1); for the sample grown for 20 min (fig. 4(b)) we observe a
:
: : : i, ; ‘3
zc!ii KineticEnergy Fig. 2. KLL carbon sample surface.
Auger
spectrum
[ev] of the 5-min deposited
until the silicon substrate is almost entirely covered. Some AES spectra of the samples grown for 5 min are reported in fig. 2, where the KLL C lines reveal two strong features at 258 and 248 eV. The lines shapes and the relative intensities are very close to those expected for a clean (1 1 1) natural diamond surface [2-41. So, just 5 min after the beginning of the carbon deposition process, we detect the diamond phase occurrence by AES. We observe the AES carbon line shape, shown in fig. 3, 20min after the beginning of the growth under the same
: : : : : : : : :: :: ii 22u
240
Kinetic Energy Fig. 3. KLL carbon Auger
growth
chamber
for 20 min.
-15
-5
0 = VBM
INITIAL STATE ENERGY (eV)
lb) -25
-15
-5
I
= VBM
INITIAL STATE ENERGY (eV)
aa
280
-25
[eVl
line of the sample
exposed
in the
Fig. 4. Photoemission electron distribution curve of the valence band from a diamond surface for the S-min deposited sample (a) and for the sample exposed in the growth chamber for 20min (b), for photon energy hu = 60eV. Energies are referenced to the valence band maximum.
L. Ferrari et al. I Nucleation and growth of diamond films
shift in energy position (0.6 eV) that can signify, for instance, an increase in the band bending at the surface due to a different reconstruction 13961. In order to exclude that this shift is due to sample charging, we have monitored the Si 2p core level peaks for different photon energies and different slit widths of the Grasshopper monochromator, observing no energy shift for the core line. Si 2p spectra recorded at a photon energy of 150 eV indicate the presence of hydrated silica (Si-OH) on the substrate surface, at a binding energy of 105 eV [7]. In fig. 5, we show the Si 2p spectrum from our diamond sample grown for 5 min. Some results of normal incident reflectance applied to the CVD diamond sample grown for 20 min are also reported. In fig. 6, we report the optical reflectivity spectrum obtained with our cross double beam reflectometer in the 1.73.7 eV spectral range. In this range of photon energy, diamond is transparent and we observe interference peaks due to the presence of the silicon-diamond interface and of the nearly homogeneous diamond film. In order to explain this spectrum, we have calculated, using the Abel& matrix derived method [S-lo], the reflectance behavior of a system composed of one layer of diamond on a Si substrate. The model computed reflectance, obtained using a layer of
1
97
-BEST FIT CVD DIAMOND
, 1.7
2.2
2.7
3.2
PHOTON ENERGY
Fig. 6. Reflectivity of diamond (dotted line) and the calculated and a 25OOA thick diamond
\J
1 3.7
(ev)
film on a silicon substrate spectrum for a Si substrate
layer
(solid line).
2500 A of diamond on a silicon substrate, reproduces the experimental spectrum in a satisfying way. Normal incident reflectance offers information about the interface between the diamond surface and the Si substrate, and this technique allows one to obtain complementary results with respect to the AES and UPS techniques. This model becomes critical in the high energy range, above 4 eV, where a tungsten light source did not allow us to have a good signal/noise ratio. Where we can merely say that the material grown on silicon is a 2500 A thick layer of diamond, the possibility of the presence of a very thin layer of a second material, such as Sic, is under investigation, opening up the possibility of diamond-substrate interface studies using optical probes.
References
9s
100
105
110
BINDING ENERGY
Fig. 5. Si 2p core levels spectrum 5 min.
115
120
(eV)
of a sample
grown
for
(I] P. Ascarelli, E. Molinari, R. Polini, V. Sessa, M.L. Terranova, E. Cappelli and S. Fontana, in: Diamond and Diamond-Like Films and Coatings, eds. J.C. Angus, R.E. Clausing, L.L. Horton and P. Koidl (Plenum Press, New York, 1991). [2] L. Ferrari, E. Cappelli, A. Cricenti, S. Selci, R. Polini and G. Chiarotti, Appl. Surf. Sci. 56-58 (1992) 100. 131 B.B. Pate, Surf. Sci. 165 (1986) 83. [4] P.G. Lurie and J.T. Wilson, Surf. Sci. 65 (1977) 476.
98
L. Ferrari et al. I Nucleation and growth of diamond films
[5] F.J. Himpsel, J.F. van der Veen and D.E. Eastman, Phys. Rev. B 22 (1980) 1967. [6] F.J. Himpsel, D.E. Eastman and J.F. van der Veen, J. Vat. Sci. Technol. 17 (1980) 1085. (71 E. Paparazzo, M. Fanfoni, E. Severini and S. Priori, J. Vat. Sci. Technol. A 10 (1992) 2892.
[8] N. Born and E. Wolf, Principles of Optics (Pergamon Press, New York, 1965). [9] C.S. Heavens, Optical Properties of Thin Solid Films (Butterworth, London, 1955). [lo] S. Nannarone and S. Selci, Phys. Rev. B 28 (1983) 5930.