Energy Dependence of Radiation Damage in Sb-Implanted
Si(100)
Yi-Sheng Lai,aJ. S. Chen,a,
*
,zY. S. Ho,bH. L. Sun,bC. J. Tsai,bT. C. Chen,b Y. F. Ko,bF. S. Lee,band W. M. Youb
a
Department of Materials Science and Engineering, National Cheng Kung University, Tainan, Taiwan b
Taiwan Semiconductor Manufacturing Company, Tainan, Taiwan
Extended defects formed by antimony ion implantation in Si共100兲 are investigated as a function of the implant energy. After implantation, spike annealing and furnace annealing are performed to examine the evolution of defects. The amorphization/ recrystallization of the implanted layer is examined by transmission electron microscopy共TEM兲, photothermal characterization, and Raman spectroscopy. Secondary-ion mass spectroscopy is employed to identify the dopant distribution before and after annealing. Cross-sectional TEM reveals that, at a dose of 1⫻ 1014cm−2, 20 keV Sb implantation is sufficient to induce an amorphous-like layer in Si共100兲. After spike annealing, the amorphous-like layer restores to the crystalline state, but defects are observed when the Sb implantation energy is greater than 50 keV. For 70 keV implantation, extended defects appear at the near-surface and the end-of-range 共EOR兲 regions. It is observed that near-surface defects diminish after spike annealing at temperatures higher than 980°C, while the EOR defects become coarse at 1095°C. A comparison between the spike annealing and the furnace annealing for the sheet resistance and the EOR defect is also addressed.
© 2005 The Electrochemical Society. 关DOI: 10.1149/1.1922907兴 All rights reserved.
Manuscript submitted June 3, 2004; revised manuscript received January 18, 2005. Available electronically May 24, 2005.
In order to achieve the reduced dimensions of state-of-the-art semiconductor devices, ultrashallow implantation in source/drain extensions is used to suppress short-channel effects as well as to meet the requirement for high-current drive capability.1The 90 nm technology node will require the source/drain junction depth to be below 45 nm for the contact and below 25 nm for the extension.2 This requires the use of a low thermal budget cycle to activate the dopant and limit diffusive distances. Since electrical activation of high dopant concentration usually requires a high annealing tem-perature, spike annealing with a nearly zero dwell time at the peak temperature has become the method of choice for activating the dopants. Besides, increasing the ramp rate during the activation an-nealing also leads to the reduction of the junction depth.3,4 Im-planted antimony共Sb兲 exhibits lower loss of total dose and lower sheet resistance than similarly implanted arsenic共As兲, and therefore can be used for source/drain extensions in n-channel metal-oxide-semiconductor devices.5
In the past two decades, extended defects formed by ion implan-tation and the activation anneal have been intensely investigated.6-8 A classification scheme has been developed to group all extended defects into five types.8Among the best known defect type is the end-of-range共EOR兲 defect, which forms beneath the amorphous/ crystalline共a/c兲 interface in the damaged crystalline Si. It is noticed that this type of defect is introduced when an amorphous layer is formed by ion implantation. High-mass Sb ions used for ultrashal-low implantation usually exhibit a ultrashal-low amorphization threshold.9 The density of EOR defects is reported to be influenced by the implant species,10,11 energy,12 ion dose,13,14 dose rate,14,15 temperature,16and solubility.17The defects are very difficult to re-move even under high-temperature annealing.18Moreover, they may have a deteriorating effect, such as enhanced leakage currents, if the space-charge region of the junction encloses the EOR defect. As a result, engineering of the ion-induced damage has become a chal-lenge to ion implantation technology.
Postimplantation annealing ambient and surface chemistry also affect defects clustering or dissolution.19,20During the early anneal-ing stage, Si interstitial atoms have a tendency to aggregate on兵311其 planes and grow preferentially along the具110典 direction.21The gov-erning mechanism is possibly due to a balance between the local stress, binding energy, and dangling bond density.22 Once these 兵311其 planes are formed, further annealing can make them coarse in size and thus reduce in density. To our knowledge, defects induced
by Sb ions as well as their evolution after spike annealing have not been fully investigated. In this work, we study the Sb ion-induced crystal amorphization, and also evolution of defects at various spike annealing temperatures. A comparison between spike annealing and furnace annealing for the sheet resistance and the EOR defect is also demonstrated.
Experimental
Single-crystal silicon 共100兲 wafers 共Czochralski grown 8-12⍀ cm, p-type兲 with 26 Å thick screen oxide were implanted using an Applied Materials XR80S high-current implanter by vary-ing the implant energy from 10 to 70 keV at a dose of 1 ⫻ 1014cm−2. The 26 Å thick screen oxide was grown by rapid ther-mal oxidation. Singly-charged ions共Sb+兲 were implanted with en-ergies of 10, 20, 30, 40, and 50 keV, while doubly-charged ions 共Sb2+兲 were implanted with energies of 50, 60, and 70 keV. The beam current was⬃1.25 mA for Sb+ species, and⬃2.50 mA for Sb2+species. The Sb used in the ion source was the elemental metal, which was heated to 420°C for vaporization. The tilt/twist angle settings were 0°/22°. A water-cooling system was employed in the wafer holder to reduce the heat generated during implantation. After implantation, spike annealing was performed using an Applied Ma-terials Radiance Centura chamber in a nitrogen ambient for dopant activation at 950, 980, 1050, and 1095°C. Samples were first ramped from 350 to 650°C at a rate of 15°C/s and then ramped to the target temperature at a rate of 250°C/s. Furnace annealing was also conducted at 900°C for 15, 30, 45, and 60 min with 100 sccm flowing N2.
Two nondestructive methods were used to characterize the lattice damage. Implanted wafers, before and after spike annealing, were measured by a Therma-Probe 500. A pump laser with a wavelength = 532 nm was modulated at a frequency of 1 MHz. At this fre-quency, the thermal diffusion length was about 5.3m in silicon 共thermal diffusivity of silicon = 0.9 cm2/s兲.23
The intensely modu-lated light beam on the sample surface and the subsequent energy dissipation creates a time-dependent response field. A probe laser at = 670 nm was focused on the sample surface to measure the ther-mal wave共TW兲 induced changes in the reflectivity. The magnitude and a phase shift of the TW are sensitive to the doping level and lattice damage.24The crystallinity of Si was characterized at room temperature by a Coherent Innova 90 Raman spectroscopy using the 514.5 nm line of an argon-ion laser. The signal of Raman spectros-copy was from the nonlinear scattering of laser beams. The probing depth is about 770 nm beneath the Si surface with the Ar+ laser, which in our case was sufficiently deep to detect the shallow junc-*Electrochemical Society Active Member.
tion. The signal of Raman spectroscopy at 521.0 cm−1is related to the defect-free Si single crystal. The attenuated intensity was related to lattice damage as well as the dopant substitution. Laser power and focusing condition were kept the same during the measurement. Four-point probe using a Tencor OmniMap RS75 apparatus quanti-fied the dopant activation after postimplantation annealing. Depth profiles of the Sb121and Sb123isotopes were measured with second-ary ion mass spectrometry共SIMS兲 using an Atomika 4000 instru-ment. O2+ions were used as the primary ion beam for SIMS depth profiling. A Phillip TECNAI F20 transmission electron microscope 共TEM兲 operating at 200 kV was employed to characterize the crystal defects in cross-sectional view.
Results and Discussion
Amorphization and dopant distribution in Sb-implanted Si.— The threshold for amorphization depends on the mass of the implanted species, the implant energy, the dose and dose rate, as well as on the sample temperature.25-28 Since antimony has a larger ion mass than silicon, the ion-solid interaction will lead to a dense collision cascade, which is an efficient process of amorphization. Hence, the a/c interface is expected to form at a depth beyond the projected range. Figure 1 shows the cross-sectional TEM view of 20 and 70 keV Sb-implanted samples. Both micrographs show an amorphous-like layer共a-Si兲, a transition region 共TR兲, and the undamaged crystalline Si 共c-Si兲. Figure 1 suggests that 20 keV Sb implantation is sufficient to produce an amorphous-like layer at the surface of Si共100兲. Note that the amorphization may be incomplete near the Si surface for the 70 keV implantation, and the incomplete amorphization will lead to the near-surface defects upon annealing. This is discussed in the section on Defects evolution with postimplantation annealing. The amorphous-like layer extends from the top of the transition region to the Si surface and the thicknesses are ⬃177 Å for the 20 keV implantation and⬃473 Å for the 70 keV implantation. The widths of the transition regions are ⬃91 Å for the 20 keV implantation and⬃211 Å for the 70 keV implantation, respectively. The roughness at the a-Si/TR/c-Si interface increases with the increasing implant energy. In consequence, for the same dose, higher energy implants are shown to have a thicker amorphous-like layer and a thicker transition region.
Figures 2a and b show SIMS depth profiles of 20 and 70 keV Sb implanted wafers before and after spike annealing at 1095°C. The projected ranges共Rp兲 of the 20 and 70 keV Sb implantations are around 138 and 371 Å, respectively. The peak concentrations of the 20 and 70 keV Sb implantations are about 6.75⫻ 1019 and 2.74 ⫻ 1019atom/cm3. SRIM29
simulations predicted that Rp equals 159 Å for the 20 keV, and 402 Å for the 70 keV Sb implantation, respectively. The peak concentrations were calculated to be 7.82 ⫻ 1019atom/cm3for the 20 keV and 3.19⫻ 1019atom/cm3for the 70 keV implantation, respectively. In consequence, the Rpand peak concentration values predicted by SRIM were both found to be larger than those from the SIMS results.
Figures 2a and b also show SIMS depth concentration profiles of Sb dopants for 20 and 70 keV Sb implanted wafers before and after spike annealing at 1095°C. It is found that spike-annealing treatments do not change the dopant profile very much. However, dopant diffusion still occurs after the short-duration thermal cycle. The peak concentration decreases from 6.75⫻ 1019 to 4.65 ⫻ 1019atom/cm3 共⬃31% reduction兲 in Fig. 2a and from 2.74 ⫻ 1019to 2.55⫻ 1019atom/cm3共⬃7% reduction兲 in Fig. 2b. The high peak concentration in the sample implanted at 20 keV exhibits a more significant reduction than the one implanted at 70 keV. This implies that the dopant diffusion increases as the concentration increases.30
Figure 1. TEM images of Sb-implanted Si with implant energies of共a兲 20
and共b兲 70 keV.
Figure 2. SIMS depth profiles of 共a兲 20 and 共b兲 70 keV Sb-implanted
samples, before and after spike annealing at 1095°C.
G512 Journal of The Electrochemical Society, 152共7兲 G511-G516 共2005兲
Effect of implant energies on the formation of EOR defects.— The defects in the samples implanted with Sb at energies ranging from 10 to 70 keV were characterized by using cross-sectional TEM. Previous studies on the nature of dislocation loops produced in the ion-implanted Si after annealing concluded that the dislocation loops were generated by the recoil of Si interstitials in the amorphous layer.8 In the present study, given the implant dose of 1⫻ 1014cm−2and the spike annealing temperature at 1095°C, we do not observe extended defects in the Si wafers implanted at energies of 50 keV or lower. Figure 3a and b show that there are no conspicuous defects in either 50 keV singly-charged or 50 keV doubly-charged Sb-implanted samples. However, a defect density of 1.86 defects/m, which were counted under TEM observation in a zone of 8m in length, is observed in the 60 keV Sb-implanted samples, as shown in Fig. 3c. The average depth of EOR defects is around 69 ± 5 nm, and their diameter is less than 32 nm. Accordingly, the threshold energy for defect formation should be between 50 and 60 keV.
Defect evolution with postimplantation annealing.—We exam-ined the evolution of defects after spike annealing at various tem-peratures and the result is summarized in Table I. Figures 4a-d show the defect evolution of 70 keV Sb implantation after spike annealing
from 950 to 1095°C. It is seen in Fig. 4a that defects can be found in two depth ranges. One is just beneath the Si surface共referred to near-surface defects兲, and the other is situated at a depth well below the Si surface共referred to EOR defects兲.
The average depth of the near-surface defects is about 11 ± 2 nm. The number of the near-surface defects decreases as the spike annealing temperature increases. El-Ghor et al. demonstrated that amorphization initially occurs at the interface between the sur-face and the EOR region.31During implantation, the most severely damaged region is around the Rpand the region away from the Rpis
Table I. Defect density, diameter, and depth in 70 keV Sb implant Si, as a function of spike-annealing temperatures, observed by cross-sectional TEM. Energy Spike annealing 共°C兲 Defect density Dislocation loop共nm兲 Defect center from surface 70 keV Sb2+ 950 Continuous TDs and DLs 艋9 69 ± 5 nm 980 Many DLs 艋22 1050 3.70 defects/m 艋23 1095 1.90 defects/m 艋32
Figure 3. TEM images of共a兲 50 keV singly-charged, 共b兲 50 keV
doubly-charged, and共c兲 60 keV Sb-implanted Si after spike annealing at 1095°C.
Figure 4. TEM images of 70 keV implanted Si after spike annealing at
共a兲 950, 共b兲 980, 共c兲 1050, and 共d兲 1095°C. The cross-sectional views of spike-annealed samples show crystal defects in the near-surface and the EOR region.
less damaged. The near-surface defects are likely to originate from an incomplete amorphization near the surface. Besides, they may disappear by dissolution in the vicinity of the oxide/silicon interface at high annealing temperatures. Fair and Kim have demonstrated that the shallow near-surface defects are more easily removed than the deep EOR defects.32The average depth of the EOR defects is around 69 ± 5 nm under the Si surface, which lies deeper than the a/c interface 共about 471 Å in Fig. 1兲. In this region, two defect structures are observed in the 950°C spike-annealed samples. One is a rod-like dislocation line along the 具110典 direction in the 兵111其 lattice plane, referring to threading dislocations共TDs兲.33The others are small circular-shaped dislocation loops共DLs兲, which resulted from the transformation of 共311兲 defects.1 It is found that the samples exhibit no TDs but only DLs共Fig. 4b-d兲 after spike anneal-ing at 980°C and above. The DL densities are 3.70 defects/m for the 1050°C spike-annealed sample and 1.90 defects/m for the 1095°C spike-annealed one. In comparison, the size of the DLs is found to be larger than that of the near-surface defects. On the other hand, it is seen that the size of the EOR defects increases from 9 to 32 nm when the temperature is increased from 950 to 1095°C. From the aforementioned result, it is shown that defect densities are reduced when the annealing temperature increases. Coarsening of defects is observed as the annealing temperature increases, which can be attributed to the Oswald ripening effect.34According to the Gibbs-Thomson equation, the concentration of Si interstitials near a DL decreases as its size increases. Therefore, there would be a con-centration gradient in the Si matrix which drives Si interstitials to diffuse from the small DLs to the large DLs, so that the large DLs grow at the expanse of small ones that shrink. The oversaturation of Si interstitials during annealing will result in transient enhanced dif-fusion of boron or phosphorous.35 For Sb dopants, the enhanced diffusion is very small共see Fig. 2兲, because Sb diffuses mainly via a vacancy-mediated mechanism.
Characteristics of Sb-implanted Si before and after spike annealing.—The sheet resistance共Rs兲 of Sb-implanted Si as a func-tion of the implant energy was measured after spike annealing at 1095°C, as shown in Fig. 5a. It was observed that the Rsdecreases as the implant energy increases. Lower implant energies at the same dose will result in a shallower junction and a higher doping concen-tration. Therefore, the presence of more dopant atoms in the Si lattice will lead to a more intensive scattering. As a result, the sheet resistance increases owing to the reduced mobility of electrical car-riers in the case of low-energy implantation. Other factors might be the dose-loss effect after removal of the screen oxide, and/or the precipitation of Sb共not activated since the solubility is exceeded兲 in low-energy implantation. Figure 5b suggests that the Rsdecreases as the spike annealing temperature increases. The reduction of Rsupon spike annealing is due to the further activation of dopant and the partial elimination of EOR defects. The dopant activation appears to be saturated at 1050°C, and higher annealing temperatures results in nearly the same Rsvalue.
Average TW values as a function of the implant energy before and after spike annealing at 1095°C are shown in Fig. 6a. For as-implanted samples, it is observed that TW values increase with in-creasing implant energies. It can be attributed to the deeper buried amorphous-like layer and the wider transition region, as shown in Fig. 1. The higher the implant energy, the more damage it causes. After spike annealing at 1095°C, the amorphous-like layer is recov-ered to the single-crystalline structure, and TW values are reduced to 300-450. However, the TW value for a virgin Si wafer is only 21-22. Residual defects共such as vacancies and interstitials兲 and electrically activated dopants in Si would alter the thermal and electrical prop-erties, such as thermal conductivity, surface energy state density, and excess carrier lifetime, etc.23Consequently, the depth distribution of dopants and residual defects contribute to the change of the photo-thermal properties in the annealed samples. That is the reason why the TW value cannot recede to the value of a virgin Si wafer.
Figure 6b shows TW values of samples with implant energies of
20 and 70 keV, followed by spike annealing at 950, 980, 1050, and 1095°C. For the 20 keV implanted samples, the TW value does not change much as the spike-annealing temperature increases. How-ever, for the 70 keV implanted samples, the TW value is reduced more significantly when samples are annealed at higher tempera-tures. From TEM micrographs, one can see that the decrease of defect density and the coarsening of defect size are dependent on the annealing temperature. The EOR defect is a poorer thermal conduc-tor than the Si matrix.36Garrido et al. have shown that the amplitude and phase of the thermal wave depend on the density, size, resis-tance, and depth of buried defect spheres.37 Hence, evolution of EOR defects may explain the significant change of TW values in the 70 keV Sb-implanted samples.
Samples with various process parameters were also analyzed by Raman spectroscopy. The c-Si peak at 521.0 cm−1shows nonpolar-ized Si tetrahedral structures, which is the most intense peak in a c-Si sample. If the Si backbone is destroyed or a Si atom is replaced by a foreign atom like Sb, the intensity would be attenuated. Figure 7 presents the normalized peak height obtained from Raman spec-troscopy before and after spike annealing at various temperatures. An unimplanted Si wafer was used as the standard sample for nor-malization. The as-implanted samples manifest a larger amount of
Figure 5. Sheet resistance of Sb-implanted samples共a兲 for various implant
energies共after spike annealing at 1095°C兲, and 共b兲 after spike annealing at various temperatures.
G514 Journal of The Electrochemical Society, 152共7兲 G511-G516 共2005兲
damaged structures; hence, the peak intensities at 521.0 cm−1of the as-implanted samples are much lower than that of the spike-annealed ones. Figure 7 also shows that the 70 keV samples exhibit lower c-Si peak intensities than the 20 keV samples after annealing, which should be related with the EOR defects observed in 70 keV samples.
Spike annealing vs. furnace annealing.—Figure 8 shows values measured for the sheet resistance Rsof 20 and 70 keV implanted Si samples after furnace annealing at 900°C for 15, 30, 45, and 60 min. The Rsof the samples after furnace annealing is found to be lower than after spike annealing共Fig. 5b兲. The observation indicates that Sb dopants are activated to a lesser extent by the short-duration spike annealing than by the isothermal furnace annealing. However, Fig. 8 also indicates that the Rs increases as the annealing time increases, and the rate of the Rs increase with annealing time is higher in the 20 keV implanted sample than in the 70 keV one. Takamura et al. have shown that the Sb deactivation becomes in-creasingly severe as the concentration increases.38 Therefore, the clustering of Sb atoms during furnace annealing, which leads to the deactivation of dopants, may account for the increase in Rs.
Figure 9 shows the transmission electron micrograph of the 70 keV-implanted Si after furnace annealing at 900°C for 30 min.
The density of EOR defects after furnace annealing is
1.86 defects/m, which is comparable to 1.90 defects/m after spike annealing at 1095°C. It was also observed that the size of EOR defects after furnace annealing共艋56 nm兲 is larger than that after spike annealing 共艋32 nm兲. Spike annealing thus results in smaller EOR defects than furnace annealing does. Defects with a large size will probably overlap the depletion region and serve as an additional leakage current path.39As a result, spike annealing is effective for reducing the size of EOR defects, which may have an important consequence for the reduction of the junction leakage.
Conclusions
In this work, we report on the characteristics of radiation damage induced by Sb implantation with various energies. The radiation damage and its recovery by spike annealing were examined with TEM, TW, and Raman spectroscopy. Cross-sectional TEM micro-graphs reveal that 20 keV Sb implantation is sufficient to induce an
Figure 6.共a兲 Comparison of TW values for Sb-implanted samples in
depen-dence on the implant energy, before and after spike annealing at 1095°C.共b兲 TW values of the 20 and 70 keV implanted samples after spike annealing at various temperatures.
Figure 7. Normalized peak height of the c-Si peak from Raman
spectros-copy for 20 and 70 keV Sb-implanted samples before/after spike annealing at various temperatures.
Figure 8. Sheet resistance of as-implanted 20 and 70 keV implanted Si
amorphous-like layer in Si共100兲. After spike annealing, visible de-fects were observed in the samples implanted at energies higher than 50 keV. Two discrete layers of defects were found after postimplan-tation annealing for the 70 keV implanpostimplan-tation. The near-surface de-fects dissolve after spike annealing at and above 1050°C. On the other hand, the density of dislocation loops in the EOR region de-creases as the annealing temperature inde-creases, and their size coars-ens as well. It was also found that Sb dopants are activated to a lesser extent in the short-duration spike annealing than in the iso-thermal furnace annealing. On the other hand, the spike annealing is more effective for reducing the size of EOR defects than the furnace annealing.
Acknowledgments
The authors express their thanks to Taiwan Semiconductor Manufacturing Company and the National Science Council of Tai-wan, R.O.C. 共grant no. 93–2216–E-006–015兲 for supporting this work.
National Cheng Kung University assisted in meeting the publication costs of this article.
References
1. C. Jones and E. Ishida, Mater. Sci. Eng., R., 24, 1共1998兲.
2. See, The International Technology Roadmap for Semiconductors, Semiconductor Industry Association, San Jose, CA共2003兲.
3. K. K. Bourdelle, A. T. Fiory, H.-J. L. Gossmann, and S. P. McCoy, Mater. Res. Soc. Symp. Proc., 669, J81共2001兲.
4. M. Y. L. Jung, R. Gunawan, R. D. Braatz, and E. G. Seebauer, J. Electrochem. Soc., 150, G838共2003兲.
5. H. Rücker, B. Heinemann, R. Barth, D. Bolze, V. Melnik, D. Krüger, and R. Kurps, Appl. Phys. Lett., 82, 826共2003兲.
6. A. Claverie, B. Colombeau, B. de Mauduit, C. Bonafos, X. Hebras, G. B. Assayag,
and F. Cristiano, Appl. Phys. A, 76, 1025共2003兲.
7. J. Gyulai, in Handbook of Ion Implantation Technology, J. F. Ziegler, Editor, pp. 69-117, Elsevier Science, Amsterdam共1992兲.
8. K. S. Jones and G. A. Rozgonyi, Rapid Thermal Processing-Science and Technol-ogy, R. B. Fair, Editor, pp. 123-168, Academic Press, San Diego共1993兲. 9. T. Motooka and O. W. Holland, Appl. Phys. Lett., 61, 3005共1992兲.
10. M. H. Clark, K. S. Jones, T. E. Haynes, C. J. Barbour, K. G. Minor, and E. Andideh, Appl. Phys. Lett., 80, 4163共2002兲.
11. J. Wong-Leung, S. Fatima, C. Jagadish, J. D. Fitz Gerald, C. T. Chou, J. Zou, and D. J. H. Cockayne, J. Appl. Phys., 88, 1312共2000兲.
12. D. Skarlatos, C. Tsamis, and D. Tsoukalas, J. Appl. Phys., 93, 1832共2003兲. 13. I. W. Wu and L. J. Chen, J. Appl. Phys., 58, 3032共1985兲.
14. T. E. Haynes and O. W. Holland, Appl. Phys. Lett., 58, 62共1991兲.
15. L. S. Robertson, A. Lilak, M. E. Law, K. S. Jones, P. S. Kringhoj, L. M. Rubin, J. Jackson, D. S. Simons, and P. Chi, Appl. Phys. Lett., 71, 3105共1997兲. 16. J. Wong-Leung, C. Jagadish, M. J. Conway, and J. D. Fitz Gerald, J. Appl. Phys.,
89, 2556共2001兲.
17. K. S. Jones, S. Prussin, and E. R. Weber, J. Appl. Phys., 62, 4114共1987兲. 18. S. N. Hong, G. A. Ruggles, J. J. Wortman, E. R. Myers, and J. J. Hren, IEEE Trans.
Electron Devices, 38, 28共1991兲.
19. D. Skarlatos, D. Tsoukalas, L. F. Giles, and A. Claverie, J. Appl. Phys., 87, 1103 共2000兲.
20. S. B. Herner, V. Krishnamoorthy, K. S. Jones, T. K. Mogi, M. O. Thompson, and H.-J. Gossmann, J. Appl. Phys., 81, 7175共1997兲.
21. S. Takeda and T. Kamino, Phys. Rev. B, 51, 2148共1995兲.
22. J. Dąbrowski, H. J. Müssig, G. Wolff, and S. Hinrich, Surf. Sci., 411, 54共1998兲. 23. B. C. Forget and D. Fournier, Progress in Photothermal and Photoacoustic Science and Technology-IV. Semiconductors and Electronic Materials, A. Mandelis, Editor, pp. 199-225, SPIE Press, Bellingham, WA共2000兲.
24. M. L. Shendeleva, Phys. Rev. B, 65, 134共2002兲.
25. P. Prussin, D. I. Margolese, and R. N. Tauber, J. Appl. Phys., 57, 180共1985兲. 26. T. E. Haynes and O. W. Holland, Appl. Phys. Lett., 58, 62共1991兲. 27. W. P. Maszara and G. A. Rozgonyi, J. Appl. Phys., 60, 2310共1986兲.
28. L. Pelaz, L. A. Marqués, M. Aboy, J. Barbolla, and G. H. Gilmer, Appl. Phys. Lett., 82, 2038共2003兲.
29. J. F. Ziegler and J. P. Biersack, Calculations performed with the SRIM 2003 pro-gram. Available at http://www.srim.org
30. R. B. Fair, M. L. Manda, and J. J. Wortman, J. Mater. Res., 1, 705共1986兲. 31. M. K. El-Ghor, O. W. Holland, C. W. White, and S. J. Pennycook, J. Mater. Res.,
5, 352共1990兲.
32. R. B. Fair and Y. Kim, Defects in Silicon II, W. M. Bullis, U. Gosele, and F. Shimura, Editors, PV 91-9, p. 423, The Electrochemical Society Proceedings Se-ries, Pennington, NJ共1991兲.
33. L. J. Chen and I. W. Wu, J. Appl. Phys., 52, 3310共1981兲.
34. J. M. Lifshiftz and V. V. Slyozov, J. Phys. Chem. Solids, 19, 35共1961兲. 35. P. A. Stolk, H.-J. Gossmann, D. J. Eaglesham, D. C. Jacobson, C. S. Rafferty, G. H.
Gilmer, M. Jaraíz, J. M. Poate, H. S. Luftman, and T. E. Haynes, J. Appl. Phys., 81, 6031共1997兲.
36. H. H. Lin, S. L. Cheng, L. J. Chen, C. Chen, and K. N. Tu, Appl. Phys. Lett., 79, 3971共2001兲.
37. F. Garrido and A. Salazar, J. Appl. Phys., 95, 140共2004兲.
38. Y. Takamura, A. Vailionis, A. F. Marshall, P. B. Griffin, and J. D. Plummer, J. Appl. Phys., 92, 5503共2002兲.
39. H. L. Liu, S. S. Gearhart, J. H. Booske, and W. Wang, J. Vac. Sci. Technol. B, 16, 415共1998兲.
Figure 9. TEM image of the 70 keV implanted Si sample after furnace
annealing at 900°C for 30 min. The defect density was 1.86 defects/m.
G516 Journal of The Electrochemical Society, 152共7兲 G511-G516 共2005兲
