Dose Measurements of Ultra-Shallow Implanted As and B in Si by RBS and ERD P. Pelicon, a,d G.V. Ravi Prasad, a,e M. El Bouanani, b B. N. Guo, c D. Birt, a J. L. Duggan, a and F. D. McDaniel, a a Ion Beam Modification and Analysis Laboratory, Dept. of Physics, University of North Texas, Denton, TX 76203-1427, USA b Laboratory for Electronic Materials and Devices, Dept. of Materials Science, University of North Texas, Denton, TX 76203-5310, USA c Varian Semiconductor Equipment Associates, Inc., 35 Dory Road, Gloucester, MA 01923, USA d Institute “Jožef Stefan”, P.O.B. 3000, SI-1001 Ljubljana, Slovenia e Institute of Physics, Bhubaneswar-751005, Orissa, India Abstract. Continuous miniaturization of integrated circuits requires narrower dopant profile depth in the Si channel and consequently the use of ultra-shallow implants in the manufacturing process. Secondary Ion Mass Spectroscopy (SIMS) is routinely used to measure the boron depth concentration profiles. However, due to the altered nature of the near-surface sputtering process inherent to SIMS, it underestimates the B implanted doses for implantation energies below 2 keV. Alternate ion beam methods for absolute dose measurements of ultra-shallow implanted As and B in Si are presented in this study. The dopant implant energies ranged from 250 eV, to 5 keV for boron and from 500 eV to 5 keV for arsenic. Implanted doses for both B and As varied from 2 x 1013 to 1 x 1015 atoms/cm2. The arsenic implants were studied with Rutherford Backscattering Spectrometry (RBS) using 2 MeV carbon ions. The absolute arsenic implanted doses were measured to an accuracy of better than 5%. The 1 keV arsenic implants were extensively studied for radiation damage with a 12C beam. No appreciable arsenic dose loss was observed during C irradiation for an integrated charge of ≤ 80 µC, which was the maximum used for these studies. For the B implants, Elastic Recoil Detection (ERD) was used with 14 MeV F4+ ions. A 9.4 µm Mylar foil was found to adequately stop the scattered 19F ions and give good energy separation for the 11B recoiled ions. The absolute dose measurements are ∼ 5% for the 5 keV 11B implants. Significant radiation damage was observed for the ultra shallow implants and the measured B dose has been obtained by extrapolation to the zero integrated charge of the beam. The absolute boron dose measurements of the ultra shallow (250 eV) implants were determined with an accuracy better than 10%. Spectroscopy (XPS), Auger Electron Spectroscopy (AES), Glow Discharge Optical Spectroscopy (GDOS), and the electron microprobe. In some cases, it is possible to measure concentrations down to 1 x 1015 atoms/cm2. For arsenic implanted into silicon, RBS with heavier ions is superior to standard 4He scattering since the Rutherford scattering crosssection increases quadratically with incident ion atomic number. Doyle et al. demonstrated that sensitivities better than 1011 atoms/cm2 may be obtained for heavy trace element contamination in Si using a 400 keV carbon beam for standard RBS and a 100 keV carbon beam for ToF-RBS 2, 3, 4. From an analytical point of view, the dose determination of shallow-implanted boron is much more challenging.5 The majority of the absolute dose measurements for 11B have been determined from the 11 B (p, α) 8Be nuclear reaction. Semiconductor INTRODUCTION Ion implantation is the preferred method for the formation of shallow junctions since it can be easily integrated into the production process. In order to produce narrow junctions, it is often necessary to use implantation energies below 1 keV. At these low implant energies, specific processes start to influence the implanted dose. Some of these are: selfsputtering of the dopant atoms, dopant loss into the surface oxide and dopant self-diffusion during the annealing process.1 For arsenic, several analytical techniques have been used to study shallow implants. These include Rutherford Backscattering Spectrometry (RBS), Particle Induced X-Ray Emission (PIXE), X-Ray Fluorescence (XRF), Secondary Ion Mass Spectrometry (SIMS), X-Ray Photoelectron CP680, Application of Accelerators in Research and Industry: 17th Int'l. Conference, edited by J. L. Duggan and I. L. Morgan © 2003 American Institute of Physics 0-7354-0149-7/03/$20.00 482 manufacturers usually rely on SIMS characterization for their 11B-implanted wafers. There are reproducibility problems with SIMS 6,7, particularly for ultra-shallow implants. In 1994, Simons wrote a rather extensive article,8 which outlines the short-and long-term reproducibility of SIMS measurements. For dose determination of ultra-shallow implants (∼ 250 eV), SIMS has serious limitations that are believed to be due to the non-equilibrium and altered nature of the near-surface sputtering process inherent to SIMS. SIMS measurements do not allow the extraction of reliable quantitative depth profiles from ultra-shallow implants and ultra-thin films due to uncertainties in sputtering probabilities and chargeexchange processes, radiation-enhanced diffusion, ion mixing effect near the surface, and preferential sputtering. Recently, it was pointed out by Suzuki et al. 9 that for ultra-shallow implants, SIMS analysis is unsuitable for accurate quantification. Their data shows that 11B concentrations in Si are only 80% those found by the 11B (p, α) 8Be reaction, for 200 eV implants. It is widely accepted, in the Ion Beam Analysis community that NRA can determine 11B doses with + 5% relative accuracy. In fact, Liu et al.11 have recently measured the reaction cross section at the 660 keV resonance. They claim their cross section values are accurate to 3.3%. Another technique, which gives excellent results for 10B analysis, is Neutron Depth Profiling with NRA using the reaction 10B(n, α) 7Li. This method was developed in the early 1970’s by Ziegler et al. 10 The method is non destructive and the depth distribution can easily be determined by measuring the resultant recoil energy of the alpha particles from the reaction. By this technique depth resolutions of 10 nm are possible for 100 keV 10B implants with concentrations of 2 x 1014 atoms/cm2. However, Neutron Depth Profiling measurements require a reactor and rather specialized equipment. For the study presented in this paper, Elastic Recoiled Detection (ERD) was used to determine absolute ultra-shallow 11B implanted dose in silicon. F4+ ions of 14 MeV energy were used to impinge on the sample at a glancing angle of 10 degrees. The recoil boron ions were measured at 25 degrees. It will be shown that this method can be used to accurately determine 5 keV boron implant doses. Good results can also be obtained for the ultrashallow 250 eV implants, using the zero incident beam integrated charge extrapolation method. arsenic using a high-current ion implanter. The implanted doses were in the range between 1x 1015 atoms/cm2 and 2 x 1013 atoms/cm2 with intermediate values of 3 x 1014 atoms/cm2, 1 x 1014 atoms/cm2, and 5 x 1013 atoms/cm2. The implanted energies of arsenic were 500 eV, 1000 eV, 2000 eV and 5000 eV. For the boron-implanted samples, the implantation energies used were 250 eV, 500 eV, 1000 eV and 5000 eV. DOSE MEASUREMENT OF SHALLOW-IMPLANTED ARSENIC IN SILICON BY RBS Experimental Arrangement The Arsenic measurements were made using the RBS beam line of the 3 MV tandem accelerator in the Ion Beam Modification and Analysis Laboratory at the University of North Texas. A 2 MeV 12C+ beam O 10000 As 1000 Yield Si 100 10 1 50 100 150 200 250 300 350 400 Channel FIGURE 1. Measured RBS spectrum for an implanted As dose of 1 x 1015 atoms/cm2 and an implantation energy of 500 eV. impinged on the arsenic implanted silicon samples, which were tilted at 10 degrees to avoid channeling. The backscattered 12C ions were detected at 130 degrees with a silicon surface barrier detector, which had a solid angle of 3.5 msr. To normalize the measurements, the number of impact particles times the solid angle of the detector was determined with an accuracy better than 3% by means of a mesh charge collector 12 and RBS reference standards. Our RBS reference standards were checked against a thin Rhodium film on vitreous carbon (thickness accuracy 2 %) supplied by a European Union precision standards laboratory (IRMM) in GEEL, Belgium.13 SAMPLE PREPARATION Silicon wafers with the crystal orientation <100> and a diameter of 8” were implanted with boron and 483 Results for the Arsenic Measurement even at arsenic concentrations of 2 x 1013 atoms/cm2. For concentrations below 2 x 1013 atoms/cm2, the background uncertainty starts to contribute significantly to the overall uncertainty. A typical measured spectrum for 1x 1015 arsenic at an implant energy of 500 eV is shown in Figure 1. In order to check for 12C beam damage, consecutive measurements were made at the same sample position for different time-integrated current values. Figure 2 shows that for the 1 keV arsenic implants, there is no measurable difference in the integrated charge. Sample AF03852: 1e15 As at/cm Experimental Arrangement 2 after 17 34 51 68 85 400 Counts 300 200 As peak area does not change with accumulated beam charge 100 0 280 300 DOSE MEASUREMENT OF SHALLOW BORON IMPLANTS BY HEAVY ION ERD µC µC µC µC µC The ERD method was first proposed in 1976 by L’Ecuyer et al. and reported in the Handbook of Modern Ion Beam Materials Analysis. 14 The first practical experiments were done in 1979 by Doyle, et al. 15 The experimental arrangement used in this study is similar to that outlined in the Handbook. 14 The ERD detector was positioned at 25 degrees relative to the beam direction. The sample was tilted 10 degrees with respect to the beam. A 9.4 µm Mylar foil was placed in front of the ERD detector together with a rectangular aperture 3 mm wide and 10 mm high. The thickness of the stopping foil and the ERD recoil angle are optimized to stop the scattered 14 MeV F4+ incident ions and yield an isolated boron peak in the high-energy part of the detected spectra. Figure 3 shows a typical measured spectrum with the 11B peak clearly resolved from H, O, and C. An RBS detector at 150 degrees was used to calibrate the incident ion integrated charge during the measurements. The solid angles of the RBS and the ERD detectors were measured by a 2% NIST 241Am standard. The major problem with all heavy ion experiments is determining the number of incident ions. This was accomplished by using an “80 % transmission tungsten Mesh-beam integration monitor” in the beam line prior to the scattering chamber.12 This mesh is insulated from its surroundings (floating electrically) and enclosed in a cylindrical tube biased at -300 volts, so that secondary electrons do not escape the mesh. This device is used for all of our high-precision RBS measurements. In order to normalize the 14 MeV F4+ ions, RBS measurements were made on a 2% RBS reference standard. In this particular case, the 14 MeV F4+ ions were backscattered from a Au standard. Since the RBS cross section, solid angle of the detector, and thickness of the Au target are all known, the integrated charge from the mesh can be accurately scaled to the true current on the target. With this technique, the true charge which impinges on the target for the ERD and RBS measurements could be determined to better than 3%. Shift due to C deposition associated with the vacuum conditions 320 Channel No. FIGURE 2. Arsenic peak in the RBS spectrum measured in the intervals of accumulated charge of 17 µC for a 2 MeV 12C1+ beam at the same irradiated sample spot. The sample was implanted to 1 x 1015 atoms/cm2 at an energy of 1 keV. The observed energy shift of the peak is a result of beam-associated contamination build-up on the sample surface during the measurements. The area of the irradiation spot is ~ 1 mm2. The vacuum in the sample chamber during the measurements was better than 5 x 10-7 mbar. For the lowest arsenic implant dose (2 x 1013 atoms/cm2) this method is still applicable. It was found that for 70 µC, which was the integrated value similar to Figure 2, the arsenic peak showed 500 counts. At values below 2 x 1013 atoms/cm2, the solid angle of the detector would have to be increased and longer run times would be required to minimize the statistical error contribution and obtain reasonable accuracy. The relative accuracy of the As measurement described here is determined by the sum of the normalization uncertainty and the statistical uncertainty in the number of accumulated As counts. The normalization error is ~ 3% and consists of ~ 1% instability of the charge integration and ~ 2% in the absolute error in the thickness of the RBS reference sample. Since there is no appreciable radiation damage, the statistical uncertainty could be kept well below 2% by extending the data collection times 484 Results of the B Implantation Measurements peaks, which come from contaminants on the surface. This good separation was seen for all of the 11B implanted samples with implant energies up to 5 keV. For higher implant energies, the B peak would move to the lower channels and overlap with the surface hydrogen and carbon peaks. A major concern, for all ERD measurements with heavy ions, is the induced damage of the sample. In order to study this effect, a series of measurements were made on the same target spot. Figure 4 shows a plot of the boron fluence as a function of accumulated charge for different implant energies. It can be seen from Figure 4 that there is very little ion-induced damage for the 5 keV implants for an integrated beam charge up to 220 µC. For implant energies below 1 keV, considerable 11B loss was observed as can be seen in the figure. For heavy ion bombardment, the surface layers of the target are usually eroded away by sputtering and ion induced erosion (desorption). 16 This damage versus accumulated beam studies have usually been fitted by one or two exponentially decreasing functions. 17,18 The curves shown in Figure 4 are seen to fit the experiment quite well. These theoretical curves were generated using the exponential equations outlined in reference 16 for ion-induced damage. In Figure 4, for the case of the 250 eV and 500 eV implants, the absolute implant dose may be determined by extrapolating the accumulated beam charge to zero using the exponential functions outlined by Behrisch, et al. 16 As mentioned earlier, the accepted value is in good agreement with the 5 keV implant data. The error with the accepted values for the 5 keV implant measurements is + 5%. The extrapolated value for the 500 eV data is within 3% of the 5 keV implant measurements. The worse case extrapolation is for the 250 eV data where the implanted boron is mostly on the surface. However, even for these very shallow implants, the extrapolation is within 10% of the accepted value. Figure 3 shows a typical ERD spectrum for 11B. This spectrum was made for 1 x 1015 boron atoms/cm2 implanted in Si(100) at an energy of 250 eV. It can be seen from the spectrum that the B is clearly separated from the hydrogen, carbon and oxygen 19 4+ Eimpl = 250 eV, ERDA 14.2 MeV F , θ=25 deg, 9.4 µm mylar O Beam charge per measurement 46 µC, cumulative: after 46 µC after 184 µC H C 100 Yield B 10 1 50 100 150 200 250 300 Channel FIGURE 3. ERD spectra measured for an implanted B dose of 1 x 1015 atoms/cm2 and implantation energy of 250 eV. The first and fourth consecutive spectra were obtained at the same irradiation spot, which accumulated 46 µC of integrated beam charge. 10 2 B fluence [1e14 at/cm ] 9 8 7 6 5 CONCLUSIONS 250 eV 500 eV 5000 eV 4 For Arsenic implanted into silicon, RBS measurements were made with 2 MeV 12C ions. The RBS spectra showed a well-separated arsenic peak with no measurable radiation damage for up to 84 µC of charge on the same beam spot. The measurements were made with an accuracy of ± 5% for shallow implants as low as 2 x 1013 arsenic atoms/cm2. The data indicate that lower concentrations could be measured by increasing the solid angle of the RBS detector. 3 0 50 100 150 200 250 Accumulated beam charge [µC] FIGURE 4. The resulting boron fluence measured by the 14.2 MeV 19F4+ beam as a function of integrated charge on the measured spot. Extrapolation, using exponential functions justified in the work of Behrisch et al., 16 has been used to evaluate the implanted fluence. 485 15. Doyle, B.L. and Peercy, P.S., Appl. Phys. Lett. 34, 811 (1970). 16. Behrisch, R., vonder Linden, W., Von Toussaint, V., and Grambole, D., Nucl. Inst. Meth. B155, 440-446 (1999). 17. Scherzer, B.M.U., Bleuer, R.S., Behrisch, R., Schuls, R., Roth, J., Borders, J., and Langley, R., J. Nucl. Mater. 85, 102 (1979). 18. Roth, J., Scherzer, B.M.U., Bleuer, R.S., Brice, D.K., Picraux, S.T., and Wampler, W.R., J. Nucl. Mater. 93/94, 601(1980). For boron implanted into silicon, the ERD measurements were made with 14 MeV 19F4+ ions. The measured boron absolute dose is in excellent agreement (+5%) with the nominal value for the 5 keV implant data. Furthermore, there is no measurable ion induced damage for the 5 keV data. For the 250 eV and 500 eV implants, ion beam damage was observed. It was, as expected, seen to increase as the accumulated charge was increased on the same beam spot. Well established zero integrated charge methods such as those proposed by Behrisch et al.16 were used to fit experimental dose loss data versus accumulated incident ion integrated charge. The calculated boron dose agreed with the nominal dose to better than 10 %. ACKNOWLEDGEMENTS The work at UNT supported in part by the National Science Foundation, the Office of Naval Research, the Texas Advanced Technology Program, and the Robert A. Welch Foundation. REFERENCES 1. Scanlon, P.J., Ridgway, M.C., and Brongersma, H.H., Nucl Inst and Meth B45 615 (1990). 2. Doyle, B.L., Knapp, J.A. and Buller, D.L., Nucl. Inst. And Meth. In Physics Research B42 295-297 (1989). 3. Doyle, B.L., Knapp, J.A. and Buller, D.L., Nucl. Inst. And Meth. In Physics Research B42 143-146 (1990). 4. Banks, J.C., Doyle, B.L, Knapp, J.A., Werho, D., Gregory R.B., Anthony, M., Hurd, T.Q. and Diebold, A.C., Nucl. Inst. And Meth. In Physics Research B136-138 12231228 (1998). 5. Moncoffre, M., Nucl. Inst. And Meth B66 1226-138 (1992). 6. Magee, C.W., Jacobson, D., and Gossmann, H.J., J. Vac. Sci. Technol. B18 (1) 489 (2000). 7. Loseing, R., Guryanov, G.M., Hunter, J.L., and Griffis, D.P., J. Vac. Sci. Technol. B18(1) 509 (2000). 8. Simons, D.S., Secondary Ion Mass Spectrometry, SIMS IX, John Wiley and Sons, Chichester, 1994, pp 140-145. 9. Suzuzi, M., Tomita, M., and Murakoshi, A., Nucl. Inst. And Meth. In Physics Research B190 552-555 (2002). 10. Ziegler, J.F., Cole, G.W., and Baglin, J.E.E., J. Appl. Phys. 43 3809 (1972). 11. Liu, J., Lu, X., Wang, X., and Chu, W.K., Nucl. Inst. And Meth. In Physics Research B190 107-111 (2002). 12. El Bouanani, M., to be published. 13. These precision standards may be obtained from The European Commission, Joint Research Center, Institute for Reference Materials and Measurements (IRMM), email: waetjen@irmm.jrc.be 14. Tesmer, J.R. and Nastasi, M., “Handbook of Modern Ion Beam Materials Analysis,” Materials Research Society (1995) p 85-137. 486
© Copyright 2025 Paperzz