Thermally induced structural changes in ... - Semantic Scholar
Thermally induced structural changes in nanoporous silicon dioxide from x-ray photoelectron spectroscopy Martin T. K. Soha兲 Woodside Energy Ltd., Perth 6000, Australia
J. H. Thomas III Characterization Facility, University of Minnesota, Minneapolis, Minnesota 55455
Joseph J. Talghader Department of Electrical and Computer Engineering, University of Minnesota, Minneapolis, Minnesota 55455
共Received 10 July 2006; accepted 6 September 2006; published 18 October 2006兲 Morphological changes due to adsorbed gases in nanoporous silicon dioxide thin films are demonstrated using in situ x-ray photoelectron spectroscopy at temperatures in the range 20艋 T 艋 300 ° C. Adsorbed hydrogen bonded water vapor is observed to relax the surface bond strain of low-temperature electron-beam deposited silicon dioxide up to 100 ° C. This was determined by measuring the width of the Si 2p and O 1s photoemission peak full widths at half maximum, which are distinctly smaller for films with adsorbed water vapor than for the same films after vapor has been outgassed by heating above 100 ° C. In situ heating in the range 100⬍ T ⬍ 200 ° C decreases the peak width as the atoms gain sufficient energy to establish a more homogeneous local bonding environment. This process is overshadowed above 200 ° C as thermally induced localized bond strains and charge inhomogeneities at the surface begin to introduce disorder, as demonstrated in the repeatable increase in peak spread with temperature for thermally grown silicon dioxide and quartz. The in vacuo peak width behavior in subsequent thermal cycles is repeatable for the nanoporous thin films. However, if the films are reexposed to atmosphere, the initial increase in peak width is seen again. © 2006 American Vacuum Society. 关DOI: 10.1116/1.2359734兴
I. INTRODUCTION Silicon dioxide 共SiO2兲, in particular, its low-temperature deposited nanoporous polymorphs 共np-SiO2兲, has been studied extensively for device passivation and optical coatings.1 The low-packing density and moisture permeability of such films are well known,2 which has led to alternative deposition schemes to avoid these issues.3,4 While such characteristics are usually undesirable, they imply that one might be able to trap adsorbed gases using encapsulation, thus controlling stress and perhaps other mechanical properties. This capability would be particularly useful in the design of coatings on membranes in micromechanical devices, which makes comprehension of the precise mechanisms behind such characteristics critical to device design and fabrication. In the present work, in situ x-ray photoelectron spectroscopy 共XPS兲 has been used to study the full width at half maximum 共FWHM兲 dependence of Si 2p and O 1s peaks as a function of temperature of electron-beam deposited npSiO2, thermally grown SiO2, and ␣ quartz. It has been found that the FWHM of np-SiO2 during the first thermal cycle in vacuo is due to the desorption of hydrogen bonded water vapor resulting in subsequent bond angle strain relaxation at the surface. This is contrasted with the thermal and quartz samples that do not follow this behavior, that is, the FWHM increases with temperature indicating an increase in bond angle strain. After the initial desorption process of water a兲
Electronic mail: [email protected].
2147
J. Vac. Sci. Technol. A 24„6…, Nov/Dec 2006
from np-SiO2 occurs, the FWHM indicates a reproducible decrease in bond angle strain and relaxation. These data have been shown to be reproducible run to run and for different depositions. Re-exposure to ambient strongly suggests that any thermally induced bond reconfiguration in np-SiO2 is not permanent. II. EXPERIMENT np-SiO2 thin films of 200 nm thickness were deposited from fused silica granules onto 0.5 mm thick 共100兲 silicon substrates, as well as 0.13 mm thick 共100兲 germanium substrates, in a Varian 3118 electron-beam evaporator at room temperature. Chamber base pressure was 3 ⫻ 10−7 Torr, and the deposition rate was set at 1 nm s−1 using an Inficon quartz-crystal monitor. A 200 nm thick dry thermal oxide was also grown at 1100 ° C on a 共100兲 silicon substrate using a Tylan furnace for comparison. XPS was performed on a Physical Electronics 555 system using a cylindrical mirror analyzer. Survey 共1000艌 E 艌 0 eV, 100 ms/step, 0.5 eV/step, and 200 eV pass energy兲 and high-resolution spectra 共10 eV window, 100 ms/step, 0.05 eV/step, and 25 eV pass energy兲 were obtained as a function of probe temperature. The system binding energy scale was calibrated using a gold 共Au兲 and copper 共Cu兲 standard to yield Au 4f 7/2, Cu 2p3/2, and carbon C 1s photoelectron binding energies of 83.85, 932.68, and 285 eV, respectively. The FWHM of the Au 4f 7/2 peak was 1.18 eV. XPS specimens were mechanically mounted onto a low-temperature heating stage, with the
0734-2101/2006/24„6…/2147/4/$23.00
©2006 American Vacuum Society
2147
2148
Soh, Thomas III, and Talghader: Thermally induced structural changes in nanoporous silicon dioxide
FIG. 1. High-resolution 共a兲 Si 2p and 共b兲 O 1s binding energy 共uncorrected for surface charging兲 photoemission peak profiles 共line兲 and 85% Gaussian curve fit 共dashed line兲.
stage temperature measured from an attached K-type thermocouple. The heater ramp rate was approximately 1 ° C / min up to 300± 1 ° C, and the chamber pressure was ⬍1.5 ⫻ 10−7 Torr. XPS survey spectra of all specimen surfaces yielded Si, O, and C photoemission peaks, with the integrated area of the C 1s peak monotonically decreasing with increasing measurement temperature. An 85% Gaussian curve fit was applied to high-resolution scans of the Si 2p and O 1s peaks after the background was subtracted using the Shirley algorithm 共Fig. 1 for the np-SiO2 thin films兲.5 From Fig. 1, there was no evidence of peak structure. The model curve fitted well to the experimental data, justifying the single peak parameter fitting approach taken. III. RESULTS AND DISCUSSION Figure 2 plots the FWHMs of the Si 2p and O 1s photoemission peaks of the np-SiO2 thin films, thermally grown oxide, and quartz substrate versus temperature. These data suggest a monotonic reduction in Si 2p and O 1s FWHMs to a value of 1.87 and 1.89 eV 共at ⬃200 ° C兲, respectively, for the np-SiO2 thin films after the first heating cycle. This is in J. Vac. Sci. Technol. A, Vol. 24, No. 6, Nov/Dec 2006
2148
FIG. 2. 共a兲 Si 2p and 共b兲 O 1s peak FWHMs vs temperature for the npSiO2 thin films 共square兲, thermal oxide 共circle兲, and quartz substrate 共crossed diamond兲. Typical first heating cycle data for the np-SiO2 thin films are also included 共open square兲. The error bars denote the standard deviation, and the quartz and thermal oxide data points have been displaced for clarity by −10 and 10 ° C, respectively.
contrast to the increase in FWHMs of the thermal oxide and quartz specimens. The O 1s and Si 2p peak binding energy difference was measured over the entire temperature range to be 429.38± 0.01 eV for the np-SiO2 thin films, 429.37± 0.02 eV for the thermally grown oxide, and 429.43± 0.02 eV for the quartz 共Fig. 3兲. This indicates that the O 1s and Si 2p binding energy chemical shifts are not significantly affected by heating 共20艋 T 艋 300 ° C兲. np-SiO2 specimens deposited onto Ge substrates also display the behavior of Fig. 1, and FWHM measurements taken at the beginning and end of a 2 h anneal cycle at all measurement temperatures ⬎60 ° C were found to be the same. Furthermore, the negligible impact of carbon resulting from ambient exposure on the FWHM was confirmed by constant O 1s / Si 2p and C 1s / Si 2p integrated area ratios 共Fig. 4兲. FWHM measurements during temperature cycles were repeatable for all specimens, with the exception of first cycle behavior for the np-SiO2 thin films—observable again postanneal after prolonged exposure to the ambient.
2149
Soh, Thomas III, and Talghader: Thermally induced structural changes in nanoporous silicon dioxide
2149
FIG. 3. O 1s – Si 2p binding energy chemical shifts vs temperature for the 共a兲 np-SiO2 thin films, 共b兲 thermal oxide, and 共c兲 quartz substrate. The error bars denote the standard deviation.
The evidence suggests that the FWHM versus temperature observations are a result of surface bonding reconfiguration. According to Grunthaner et al.,6 thermally grown SiO2 consists of a continuous network of Si-centered tetrahedra joined together by oxygen atoms with an average Si– O–Si bond angle, corresponding to a certain O 1s and Si 2p binding energy displacement and ring size 共6 for quartz from pseudo-potential calculations7 and thermally grown SiO2兲. A decrease in ring size will reduce the average Si–O–Si bond angle and increase the O 1s and Si 2p chemical shifts, while an increase in ring size will produce the opposite effect. Since the O 1s and Si 2p binding energy difference was observed to be invariant to temperature 共Fig. 3兲, the average Si–O–Si bond angle6,8 is independent of annealing temperature. However, the increase in FWHM with temperature for the thermal oxide specimen indicates that the distribution of bond angles and ring sizes at the surface broadens with temperature.9 This suggests that thermally induced localized bond strain and charge inhomogeneities on the thermal oxide and quartz surfaces are prevalent at elevated temperatures, making the surface more chemically reactive and vulnerable to etching 共formation of silanol groups10–12兲 and failure 共for
example, in humidity13兲. It is clear that ex situ measurement of Si 2p and O 1s FWHMs after annealing will not reveal this behavior. In contrast, the np-SiO2 Si 2p and O 1s FWHMs versus heat treatment in Fig. 2 exhibit a different first cycle behavior. The np-SiO2 FWHM versus temperature behavior in subsequent temperature cycles is repeatable and rises to a significantly higher value at 20 ° C, which may be a result of 共water兲 impurity desorption. The average bond angle of the np-SiO2 thin films is analogous to those of the thermal oxide and quartz substrates due to similar O 1s and Si 2p chemical shifts. However, the variation in bond angle 共and ring size兲 distributions is much more pronounced in the np-SiO2 thin films because of the larger peak FWHM at 20 ° C. This may be due to lower packing density in thin films deposited at reduced temperatures,2 with low adatom mobility at the surface resulting in highly disordered bonding 共large FWHM兲. Therefore, it is likely that hydration of the np-SiO2 thin film surface serves to stabilize the morphology through interatomic bond strain relaxation—already observed for npSiO2 thin films deposited by electron beam at room temperature and 200 ° C.14 The as-deposited bond strains are rein-
FIG. 4. O 1s / Si 2p vs C 1s / Si 2p area for the 共a兲 np-SiO2 thin films, 共b兲 thermal oxide, and 共c兲 quartz. JVST A - Vacuum, Surfaces, and Films
2150
Soh, Thomas III, and Talghader: Thermally induced structural changes in nanoporous silicon dioxide
stated upon dehydration, increasing the FWHM in the first heating cycle up to 100 ° C. Above 100 ° C, the FWHM will begin to decrease as the atoms gain sufficient energy to establish a more homogeneous ring configuration. However, this process is overshadowed at higher temperatures due to the mechanisms already established for the thermal oxide and quartz. The FWHM monotonically increases upon cooldown 共due to the absence of water adsorption兲 that explains previous observations of higher Si 2p and O 1s peak FWHMs from immediate ex situ XPS post-anneal.15 The ex situ adsorption and thermal desorption of water into and out of np-SiO2 thin films have been postulated by many workers through infrared absorption studies.1,3,16 The permanent reduction in Si–OH 共stretching兲 infrared absorption intensity around 3350– 650 cm−1, as well as peak shift and FWHM reduction of the Si–O 共stretching兲 absorption mode around 1070 cm−1 from annealing,17 has been attributed to the removal of physisorbed silanol groups 共through Novak condensation18兲 and thermally induced morphological changes. However, the XPS peak FWHM first cycle and temperature characteristic of the np-SiO2 thin films 共Fig. 2兲 were reproducible after re-exposure to the ambient, suggesting that any thermally induced bonding reconfiguration 共for T 艋 300 ° C兲 is not permanent. Given that the first cycle behavior only extends to T = 100 ° C, the XPS analysis indicates that the mechanism of bond strain relaxation hydration in np-SiO2 is through much weaker hydrogen bonding. IV. CONCLUSIONS In situ XPS was used to study the Si 2p and O 1s FWHM temperature dependence of np-SiO2 thin films, thermal oxide, and quartz. It was found that np-SiO2 bond distributions go through a reversible change with temperature where it is proposed that water vapor is desorbed in vacuo for T ⬍ 100 ° C. After this process occurs, the material continues to show a relaxation in bond angle distribution 共through reduced peak FWHM兲. This was compared with both the thermal oxide on silicon and quartz, which exhibit an increase in bond angle distributions to T = 300 ° C. Thermal cycling in vacuo of these materials is reproducible. With ambient exposure, the np-SiO2 film is reexposed to moisture and the initial
J. Vac. Sci. Technol. A, Vol. 24, No. 6, Nov/Dec 2006
2150
cycle is reproduced as well as the higher temperature dependence 共due to bond angle relaxation兲. ACKNOWLEDGMENTS The authors thank N. Mohan of the Department of Electrical and Computer Engineering for loaning the power supply used to heat the specimens during XPS and N. Savvides of CSIRO Industrial Physics 共Lindfield 2070, Australia兲 for constructive comments and suggestions. One of the authors 共M.T.K.S.兲 is grateful to the Australian-American Fulbright Commission 共Deakin 2600, Australia兲 and Clough Ltd. 共Perth 6000, Australia兲 for sponsoring his research fellowship in the United States. This work was financially supported by the United States Air Force Office of Scientific Research 共Contract No. FA9550-05-1-0399; Program Monitor: C. Lee兲. One of the authors 共M.T.K.S.兲 was a 2005-6 Fulbright Fellow at the University of Minnesota. W. A. Pliskin, J. Vac. Sci. Technol. 14, 1064 共1977兲. H. K. Pulker, Coatings on Glass 共Elsevier, Amsterdam, 1984兲. 3 J. A. Theil, D. V. Tsu, and G. Lucovsky, J. Electron. Mater. 19, 209 共1990兲. 4 N. Chand, R. R. Kola, R. L. Opila, R. B. Comizzoli, H. Krautter, A. M. Sergent, and W. T. Tsang, J. Appl. Phys. 78, 3315 共1995兲. 5 D. A. Shirley, Phys. Rev. B 5, 4709 共1972兲. 6 F. J. Grunthaner, P. J. Grunthaner, R. P. Vasquez, B. F. Lewis, J. Maserjian, and A. Madhukar, Phys. Rev. Lett. 43, 1683 共1979兲. 7 J. R. Chelikowsky and M. Schülter, Phys. Rev. B 15, 4020 共1977兲. 8 M. G. Tucker, D. A. Keen, M. T. Dove, and K. Trachenko, J. Phys.: Condens. Matter 17, 567 共2005兲. 9 S. Hofmann and J. H. Thomas III, J. Vac. Sci. Technol. A 1, 43 共1983兲. 10 K. M. Davis and M. Tomozawa, J. Non-Cryst. Solids 201, 177 共1996兲. 11 G. Vigné-Maeder and P. Sautet, J. Phys. Chem. B 101, 8197 共1997兲. 12 Y. D. Kim, T. Wei, J. Stultz, and D. W. Goodman, Langmuir 19, 1140 共2003兲. 13 I. Whitney, J. W. Johnson, and B. A. Proctor, Nature 共London兲 210, 730 共1966兲. 14 H. Leplan, J. Y. Robic, and Y. Pableau, J. Appl. Phys. 79, 6926 共1996兲. 15 M. S. Haque, H. A. Naseem, and W. D. Brown, J. Electrochem. Soc. 142, 3864 共1995兲. 16 S. Robles, E. Yieh, and B. C. Nguyen, J. Electrochem. Soc. 142, 580 共1995兲. 17 M. S. Haque, H. A. Naseem, and W. D. Brown, J. Appl. Phys. 82, 2922 共1997兲. 18 A. Novak, Structure and Bonding 共Springer-Verlag, Berlin, 1974兲, Vol. 18, p. 194. 1 2
J. H. Thomas III Characterization Facility, University of Minnesota, Minneapolis, Minnesota 55455
Joseph J. Talghader Department of Electrical and Computer Engineering, University of Minnesota, Minneapolis, Minnesota 55455
共Received 10 July 2006; accepted 6 September 2006; published 18 October 2006兲 Morphological changes due to adsorbed gases in nanoporous silicon dioxide thin films are demonstrated using in situ x-ray photoelectron spectroscopy at temperatures in the range 20艋 T 艋 300 ° C. Adsorbed hydrogen bonded water vapor is observed to relax the surface bond strain of low-temperature electron-beam deposited silicon dioxide up to 100 ° C. This was determined by measuring the width of the Si 2p and O 1s photoemission peak full widths at half maximum, which are distinctly smaller for films with adsorbed water vapor than for the same films after vapor has been outgassed by heating above 100 ° C. In situ heating in the range 100⬍ T ⬍ 200 ° C decreases the peak width as the atoms gain sufficient energy to establish a more homogeneous local bonding environment. This process is overshadowed above 200 ° C as thermally induced localized bond strains and charge inhomogeneities at the surface begin to introduce disorder, as demonstrated in the repeatable increase in peak spread with temperature for thermally grown silicon dioxide and quartz. The in vacuo peak width behavior in subsequent thermal cycles is repeatable for the nanoporous thin films. However, if the films are reexposed to atmosphere, the initial increase in peak width is seen again. © 2006 American Vacuum Society. 关DOI: 10.1116/1.2359734兴
I. INTRODUCTION Silicon dioxide 共SiO2兲, in particular, its low-temperature deposited nanoporous polymorphs 共np-SiO2兲, has been studied extensively for device passivation and optical coatings.1 The low-packing density and moisture permeability of such films are well known,2 which has led to alternative deposition schemes to avoid these issues.3,4 While such characteristics are usually undesirable, they imply that one might be able to trap adsorbed gases using encapsulation, thus controlling stress and perhaps other mechanical properties. This capability would be particularly useful in the design of coatings on membranes in micromechanical devices, which makes comprehension of the precise mechanisms behind such characteristics critical to device design and fabrication. In the present work, in situ x-ray photoelectron spectroscopy 共XPS兲 has been used to study the full width at half maximum 共FWHM兲 dependence of Si 2p and O 1s peaks as a function of temperature of electron-beam deposited npSiO2, thermally grown SiO2, and ␣ quartz. It has been found that the FWHM of np-SiO2 during the first thermal cycle in vacuo is due to the desorption of hydrogen bonded water vapor resulting in subsequent bond angle strain relaxation at the surface. This is contrasted with the thermal and quartz samples that do not follow this behavior, that is, the FWHM increases with temperature indicating an increase in bond angle strain. After the initial desorption process of water a兲
Electronic mail: [email protected].
2147
J. Vac. Sci. Technol. A 24„6…, Nov/Dec 2006
from np-SiO2 occurs, the FWHM indicates a reproducible decrease in bond angle strain and relaxation. These data have been shown to be reproducible run to run and for different depositions. Re-exposure to ambient strongly suggests that any thermally induced bond reconfiguration in np-SiO2 is not permanent. II. EXPERIMENT np-SiO2 thin films of 200 nm thickness were deposited from fused silica granules onto 0.5 mm thick 共100兲 silicon substrates, as well as 0.13 mm thick 共100兲 germanium substrates, in a Varian 3118 electron-beam evaporator at room temperature. Chamber base pressure was 3 ⫻ 10−7 Torr, and the deposition rate was set at 1 nm s−1 using an Inficon quartz-crystal monitor. A 200 nm thick dry thermal oxide was also grown at 1100 ° C on a 共100兲 silicon substrate using a Tylan furnace for comparison. XPS was performed on a Physical Electronics 555 system using a cylindrical mirror analyzer. Survey 共1000艌 E 艌 0 eV, 100 ms/step, 0.5 eV/step, and 200 eV pass energy兲 and high-resolution spectra 共10 eV window, 100 ms/step, 0.05 eV/step, and 25 eV pass energy兲 were obtained as a function of probe temperature. The system binding energy scale was calibrated using a gold 共Au兲 and copper 共Cu兲 standard to yield Au 4f 7/2, Cu 2p3/2, and carbon C 1s photoelectron binding energies of 83.85, 932.68, and 285 eV, respectively. The FWHM of the Au 4f 7/2 peak was 1.18 eV. XPS specimens were mechanically mounted onto a low-temperature heating stage, with the
0734-2101/2006/24„6…/2147/4/$23.00
©2006 American Vacuum Society
2147
2148
Soh, Thomas III, and Talghader: Thermally induced structural changes in nanoporous silicon dioxide
FIG. 1. High-resolution 共a兲 Si 2p and 共b兲 O 1s binding energy 共uncorrected for surface charging兲 photoemission peak profiles 共line兲 and 85% Gaussian curve fit 共dashed line兲.
stage temperature measured from an attached K-type thermocouple. The heater ramp rate was approximately 1 ° C / min up to 300± 1 ° C, and the chamber pressure was ⬍1.5 ⫻ 10−7 Torr. XPS survey spectra of all specimen surfaces yielded Si, O, and C photoemission peaks, with the integrated area of the C 1s peak monotonically decreasing with increasing measurement temperature. An 85% Gaussian curve fit was applied to high-resolution scans of the Si 2p and O 1s peaks after the background was subtracted using the Shirley algorithm 共Fig. 1 for the np-SiO2 thin films兲.5 From Fig. 1, there was no evidence of peak structure. The model curve fitted well to the experimental data, justifying the single peak parameter fitting approach taken. III. RESULTS AND DISCUSSION Figure 2 plots the FWHMs of the Si 2p and O 1s photoemission peaks of the np-SiO2 thin films, thermally grown oxide, and quartz substrate versus temperature. These data suggest a monotonic reduction in Si 2p and O 1s FWHMs to a value of 1.87 and 1.89 eV 共at ⬃200 ° C兲, respectively, for the np-SiO2 thin films after the first heating cycle. This is in J. Vac. Sci. Technol. A, Vol. 24, No. 6, Nov/Dec 2006
2148
FIG. 2. 共a兲 Si 2p and 共b兲 O 1s peak FWHMs vs temperature for the npSiO2 thin films 共square兲, thermal oxide 共circle兲, and quartz substrate 共crossed diamond兲. Typical first heating cycle data for the np-SiO2 thin films are also included 共open square兲. The error bars denote the standard deviation, and the quartz and thermal oxide data points have been displaced for clarity by −10 and 10 ° C, respectively.
contrast to the increase in FWHMs of the thermal oxide and quartz specimens. The O 1s and Si 2p peak binding energy difference was measured over the entire temperature range to be 429.38± 0.01 eV for the np-SiO2 thin films, 429.37± 0.02 eV for the thermally grown oxide, and 429.43± 0.02 eV for the quartz 共Fig. 3兲. This indicates that the O 1s and Si 2p binding energy chemical shifts are not significantly affected by heating 共20艋 T 艋 300 ° C兲. np-SiO2 specimens deposited onto Ge substrates also display the behavior of Fig. 1, and FWHM measurements taken at the beginning and end of a 2 h anneal cycle at all measurement temperatures ⬎60 ° C were found to be the same. Furthermore, the negligible impact of carbon resulting from ambient exposure on the FWHM was confirmed by constant O 1s / Si 2p and C 1s / Si 2p integrated area ratios 共Fig. 4兲. FWHM measurements during temperature cycles were repeatable for all specimens, with the exception of first cycle behavior for the np-SiO2 thin films—observable again postanneal after prolonged exposure to the ambient.
2149
Soh, Thomas III, and Talghader: Thermally induced structural changes in nanoporous silicon dioxide
2149
FIG. 3. O 1s – Si 2p binding energy chemical shifts vs temperature for the 共a兲 np-SiO2 thin films, 共b兲 thermal oxide, and 共c兲 quartz substrate. The error bars denote the standard deviation.
The evidence suggests that the FWHM versus temperature observations are a result of surface bonding reconfiguration. According to Grunthaner et al.,6 thermally grown SiO2 consists of a continuous network of Si-centered tetrahedra joined together by oxygen atoms with an average Si– O–Si bond angle, corresponding to a certain O 1s and Si 2p binding energy displacement and ring size 共6 for quartz from pseudo-potential calculations7 and thermally grown SiO2兲. A decrease in ring size will reduce the average Si–O–Si bond angle and increase the O 1s and Si 2p chemical shifts, while an increase in ring size will produce the opposite effect. Since the O 1s and Si 2p binding energy difference was observed to be invariant to temperature 共Fig. 3兲, the average Si–O–Si bond angle6,8 is independent of annealing temperature. However, the increase in FWHM with temperature for the thermal oxide specimen indicates that the distribution of bond angles and ring sizes at the surface broadens with temperature.9 This suggests that thermally induced localized bond strain and charge inhomogeneities on the thermal oxide and quartz surfaces are prevalent at elevated temperatures, making the surface more chemically reactive and vulnerable to etching 共formation of silanol groups10–12兲 and failure 共for
example, in humidity13兲. It is clear that ex situ measurement of Si 2p and O 1s FWHMs after annealing will not reveal this behavior. In contrast, the np-SiO2 Si 2p and O 1s FWHMs versus heat treatment in Fig. 2 exhibit a different first cycle behavior. The np-SiO2 FWHM versus temperature behavior in subsequent temperature cycles is repeatable and rises to a significantly higher value at 20 ° C, which may be a result of 共water兲 impurity desorption. The average bond angle of the np-SiO2 thin films is analogous to those of the thermal oxide and quartz substrates due to similar O 1s and Si 2p chemical shifts. However, the variation in bond angle 共and ring size兲 distributions is much more pronounced in the np-SiO2 thin films because of the larger peak FWHM at 20 ° C. This may be due to lower packing density in thin films deposited at reduced temperatures,2 with low adatom mobility at the surface resulting in highly disordered bonding 共large FWHM兲. Therefore, it is likely that hydration of the np-SiO2 thin film surface serves to stabilize the morphology through interatomic bond strain relaxation—already observed for npSiO2 thin films deposited by electron beam at room temperature and 200 ° C.14 The as-deposited bond strains are rein-
FIG. 4. O 1s / Si 2p vs C 1s / Si 2p area for the 共a兲 np-SiO2 thin films, 共b兲 thermal oxide, and 共c兲 quartz. JVST A - Vacuum, Surfaces, and Films
2150
Soh, Thomas III, and Talghader: Thermally induced structural changes in nanoporous silicon dioxide
stated upon dehydration, increasing the FWHM in the first heating cycle up to 100 ° C. Above 100 ° C, the FWHM will begin to decrease as the atoms gain sufficient energy to establish a more homogeneous ring configuration. However, this process is overshadowed at higher temperatures due to the mechanisms already established for the thermal oxide and quartz. The FWHM monotonically increases upon cooldown 共due to the absence of water adsorption兲 that explains previous observations of higher Si 2p and O 1s peak FWHMs from immediate ex situ XPS post-anneal.15 The ex situ adsorption and thermal desorption of water into and out of np-SiO2 thin films have been postulated by many workers through infrared absorption studies.1,3,16 The permanent reduction in Si–OH 共stretching兲 infrared absorption intensity around 3350– 650 cm−1, as well as peak shift and FWHM reduction of the Si–O 共stretching兲 absorption mode around 1070 cm−1 from annealing,17 has been attributed to the removal of physisorbed silanol groups 共through Novak condensation18兲 and thermally induced morphological changes. However, the XPS peak FWHM first cycle and temperature characteristic of the np-SiO2 thin films 共Fig. 2兲 were reproducible after re-exposure to the ambient, suggesting that any thermally induced bonding reconfiguration 共for T 艋 300 ° C兲 is not permanent. Given that the first cycle behavior only extends to T = 100 ° C, the XPS analysis indicates that the mechanism of bond strain relaxation hydration in np-SiO2 is through much weaker hydrogen bonding. IV. CONCLUSIONS In situ XPS was used to study the Si 2p and O 1s FWHM temperature dependence of np-SiO2 thin films, thermal oxide, and quartz. It was found that np-SiO2 bond distributions go through a reversible change with temperature where it is proposed that water vapor is desorbed in vacuo for T ⬍ 100 ° C. After this process occurs, the material continues to show a relaxation in bond angle distribution 共through reduced peak FWHM兲. This was compared with both the thermal oxide on silicon and quartz, which exhibit an increase in bond angle distributions to T = 300 ° C. Thermal cycling in vacuo of these materials is reproducible. With ambient exposure, the np-SiO2 film is reexposed to moisture and the initial
J. Vac. Sci. Technol. A, Vol. 24, No. 6, Nov/Dec 2006
2150
cycle is reproduced as well as the higher temperature dependence 共due to bond angle relaxation兲. ACKNOWLEDGMENTS The authors thank N. Mohan of the Department of Electrical and Computer Engineering for loaning the power supply used to heat the specimens during XPS and N. Savvides of CSIRO Industrial Physics 共Lindfield 2070, Australia兲 for constructive comments and suggestions. One of the authors 共M.T.K.S.兲 is grateful to the Australian-American Fulbright Commission 共Deakin 2600, Australia兲 and Clough Ltd. 共Perth 6000, Australia兲 for sponsoring his research fellowship in the United States. This work was financially supported by the United States Air Force Office of Scientific Research 共Contract No. FA9550-05-1-0399; Program Monitor: C. Lee兲. One of the authors 共M.T.K.S.兲 was a 2005-6 Fulbright Fellow at the University of Minnesota. W. A. Pliskin, J. Vac. Sci. Technol. 14, 1064 共1977兲. H. K. Pulker, Coatings on Glass 共Elsevier, Amsterdam, 1984兲. 3 J. A. Theil, D. V. Tsu, and G. Lucovsky, J. Electron. Mater. 19, 209 共1990兲. 4 N. Chand, R. R. Kola, R. L. Opila, R. B. Comizzoli, H. Krautter, A. M. Sergent, and W. T. Tsang, J. Appl. Phys. 78, 3315 共1995兲. 5 D. A. Shirley, Phys. Rev. B 5, 4709 共1972兲. 6 F. J. Grunthaner, P. J. Grunthaner, R. P. Vasquez, B. F. Lewis, J. Maserjian, and A. Madhukar, Phys. Rev. Lett. 43, 1683 共1979兲. 7 J. R. Chelikowsky and M. Schülter, Phys. Rev. B 15, 4020 共1977兲. 8 M. G. Tucker, D. A. Keen, M. T. Dove, and K. Trachenko, J. Phys.: Condens. Matter 17, 567 共2005兲. 9 S. Hofmann and J. H. Thomas III, J. Vac. Sci. Technol. A 1, 43 共1983兲. 10 K. M. Davis and M. Tomozawa, J. Non-Cryst. Solids 201, 177 共1996兲. 11 G. Vigné-Maeder and P. Sautet, J. Phys. Chem. B 101, 8197 共1997兲. 12 Y. D. Kim, T. Wei, J. Stultz, and D. W. Goodman, Langmuir 19, 1140 共2003兲. 13 I. Whitney, J. W. Johnson, and B. A. Proctor, Nature 共London兲 210, 730 共1966兲. 14 H. Leplan, J. Y. Robic, and Y. Pableau, J. Appl. Phys. 79, 6926 共1996兲. 15 M. S. Haque, H. A. Naseem, and W. D. Brown, J. Electrochem. Soc. 142, 3864 共1995兲. 16 S. Robles, E. Yieh, and B. C. Nguyen, J. Electrochem. Soc. 142, 580 共1995兲. 17 M. S. Haque, H. A. Naseem, and W. D. Brown, J. Appl. Phys. 82, 2922 共1997兲. 18 A. Novak, Structure and Bonding 共Springer-Verlag, Berlin, 1974兲, Vol. 18, p. 194. 1 2
