Carrier Illumination as a tool to probe implant dose and electrical activation. W.Vandervorsta,1, T.Claryssea, B.Brijsa, R.Looa, Y.Peytiera, B.J. Pawlakb, E.Budiartoc and P.Bordenc a IMEC, Kapeldreef 75, B-3001 Leuven, Belgium also : KULeuen, INSYS, Kasteelpark Arenberg, Leuven, Belgium b Philips Research Leuven, Kapeldreef 75,B- 3001 Leuven, Belgium c Boxer Cross Inc., Menlo Park, California 1 Abstract. The Carrier Illumination™ (CI) method is an optical technique for non-destructive in-line monitoring of postanneal junction depth, pre-anneal pre-amorphisation implant (PAI) depth, and dose. This work describes the sensitivity of the CI-signal to the as-implant dose and demonstrates that a universal response function can be derived for doses below the amorphisation limit. For the implants where the elements/doses cause amorphisation, the CI-signal reflects directly the thickness of the amorphous depth. In the case of annealed structures, it is shown that CI provides important information on the electrical activation of the dopant. This is illustrated by the analysis of CVD-layers subsequently annealed and of junction profiles produced by laser annealing. In both cases nearly identical dopant profiles are observed with secondary ion mass spectrometry while the electrical activation as derived from sheet resistance measurements is very different. This very different activation level is clearly reflected in the CI-signal. This indicates that the CI-signal is not solely related to the junction depth and the profile abruptness but also to the electrical activation of the dopants. INTRODUCTION EXPERIMENTAL The Carrier Illumination™ (CI) method is an optical technique for non-destructive in-line monitoring of post-anneal junction depth, pre-anneal pre-amorphisation implant (PAI) depth, and dose [1]. Originally proposed as an evaluation method for junction depths of annealed profiles, it has already been shown that the technique is also sensitive to the active dopant concentration and the profile abruptness [2]. Moreover recently it was shown that the technique can also measure the thickness of the amorphous depth after a pre-amorphizing implant [3]. This work further extends these studies and shows that the CI-method is also very sensitive for monitoring as-implanted low dose implants. A universal response function (independent of implant species/dose/energy) can be found relating the CIsignal to the implant dose. For high doses / heavy elements, this function is no longer valid and the CIsignal reflects the thickness of the amorphous depth. With respect to the sensitivity for annealed profiles, we have studied CVD and laser annealed profiles. It is shown that although no differences in the dopant distribution can be observed, different results for the CI-signal are obtained reflecting differences in electrical activation. Wafers were implanted with B, BF2, As and P at Imec and Sematech. Low energy Sb-implants were made at Applied Materials. These implants used 8 inch n-type Si (100) wafers at an angle of 7° tilt and 27° twist. All implants were done on the Applied Materials XR80 LEAP. CI-measurements were made using a Boxer Cross BX-10 located at Sematech on wafers implanted at Sematech and at Imec on the wafers implanted at Imec and AMAT. All signals reported relate to actual measurement values and no scaling or calibration between the different systems was performed. The amorphous depth of the Imec samples was calculated from the number of displaced Si-atoms as extracted from a 2MeV channeling RBSmeasurement. CVD-layers were grown at 700°C and subsequently annealed at 700 (5 to 10 min) and 800 °C (1 and 2 min). RESULTS CI-signal response for low dose implants CI-measurements were done on wafers with unannealed B, As and P-implants with a wide range of energies and low to high doses. The results are CP683, Characterization and Metrology for ULSI Technology: 2003 International Conference, edited by D. G. Seiler, A. C. Diebold, T. J. Shaffner, R. McDonald, S. Zollner, R. P. Khosla, and E. M. Secula © 2003 American Institute of Physics 0-7354-0152-7/03/$20.00 758 shown in Fig.1 for Boron (top), Phosporus (middle) and arsenic/antimony (bottom). 3.E+04 B, 0.5 keV B, 1 keV B, 2 keV B 3 keV B, 5 keV B 2.E+04 CI signal [uV] 1.E+04 0.E+00 very smooth and follows a well-behaved, predictable pattern, suggesting a common underlying mechanism. Therefore, it seems likely that there should be a single response curve that contains all the physics. As the CI-signal is governed by the shape of the excess carrier profile induced by the pump laser, recombination of these excess carriers on the defects produced by the implant seems to be a likely mechanism to explain the signal dependence on species, energy, and dose. In fact one could speculate on a proportionality between the CI-signal and the total numbers of defects produced during the implants. Within this frame a proper scaling factor correlating different species and energies, would then be the number of defects produced per incident ion by the implantation process. Disregarding any in-situ annealing or recombination, the number of vacancies produced during the implant can be calculated using TRIM (Transport of Ions in Matter) [4]. Fig. 2 shows the predicted number of vacancies per incident ion as a function of implant energy for the species Boron, Arsenic, Antimony, Phosphorus and BF2. For the latter case we calculated the damage as the sum of the vacancies produced by one B and two F atoms with an energy scaled to their mass relative to the total mass. B 10 B 31 B 50 B 200 B 500 -1.E+04 -2.E+04 -3.E+04 -4.E+04 -5.E+04 -6.E+04 1.E+11 1.E+12 1.E+13 1.E+14 1.E+15 1.E+16 B dose (cm-2) Phosphorus 0 CI signal (uV) -20000 -40000 Energy (keV) 10 50 200 500 1000 2000 -60000 -80000 -100000 -120000 1.E+11 1.E+12 1.E+13 1.E+14 -2 10000 Dose (cm ) Vacancies per ion As/Sb 0.0E+00 -2.0E+04 CI- signal (µ V) -4.0E+04 -6.0E+04 -8.0E+04 -1.0E+05 -1.2E+05 -1.4E+05 -1.6E+05 -1.8E+05 -2.0E+05 1E+11 As 5 KeV As 10 KeV As 10 KeV, BX As 30 KeV As 50 KeV,BX 1000 100 Boron Arsenic Phosphor BF2 10 As 200 KeV,BX Sb Sb 2 KeV Sb 5 KeV 1E+12 1E+13 1E+14 1E+15 1E+16 1 1E+17 0.1 Dose (cm -2) 1 10 100 1000 10000 Energy (keV) FIGURE 2. The number of vacancies produced per FIGURE 1. CI-signal vs. implant dose and energy for as- implanted ion in silicon as a function of implant energy. Data are generated by SRIM-2000 [4]. implanted Boron, Arsenic, Sb and Phosphorus-doped samples. The As-measurements done at Sematech are labeled as BX. Not too surprisingly, similar energies lead to the same number of vacancies regardless of the implant species, as most of the energy is dissipated in nuclear collisions absorbing the same amount of energy per displaced atom. Only at high energies and for light ions, electronic energy loss becomes important as well and the curves start to depend on the implant species. Although the response is non-linear, the data in Fig.1 show a very high sensitivity. Even in the region with the weakest response (dose 1012 – 1013 cm-2), the sensitivity is as high as 500-1000 µV/dec. Relative to the noise in the present systems this suggests a resolution in the 1010 ions/cm2 range. For all dopant species the signal response to dose and energy is 759 becomes amorphous. When the continuous amorphous film has been formed, a thermal mechanism determines the CI-signal instead of the electronic interactions with the defects [5]. Under these conditions a direct correlation with the thickness of the amorphous layer can be found [3]. The data in Fig.5 combine CI-results on Si amorphized by As, BF2 and Ge-implants [3] with different doses and energies. A sensitivity of ~2.5x103 µV/nm can be seen, implying an ultimate resolution better than 0.02 nm! Note that in [3] the CI-signals are plotted without sign and shown as positive whereas they are negative. For the comparison, in Fig.5 we have inverted them and multiplied by 3. The latter factor probably reflects differences in the two measurement set ups and, in particular, the pump laser intensity. It is clear that there is no relation to the element species and the signal solely depends on the thickness of the amorphous layer. Intrinsically there is of course a dependence on the implant dose/energy/species as for each condition the efficiency of the amorphization process and the amorphous film growth are different. When the CI-signal is plotted as a function of the vacancy “dose”, as given by the product of implant dose and the number of vacancies per ion, the result is a single universal response curve for the preannealed sample, as shown in Fig. 3. The convergence is most evident in the low and medium vacancy dose range. Energy (keV) 20000 B 3 keV B 10 B 50 B 200 B 500 As 10 As 50 As 200 P 10 P 50 P 200 P 1000 10000 CI signal (uV) 0 -10000 -20000 -30000 -40000 -50000 -60000 -70000 -80000 1.E+13 1.E+14 1.E+15 1.E+16 1.E+17 1.E+18 1.E+19 Dose x Vacancies per ion (cm-2) FIGURE 3. Universal response curve of the CI-signal as a function of the product of dose and number of vacancies per ion, demonstrating a common underlying mechanism for the as-implanted samples, regardless of implant species, energy. The curve is valid for low doses that do not cause any overlap between atomic displacements and amorphisation.. 0.E+00 CI-signal (µ V) As, 5e15 cm-2 CI response for amorphizing implants The situation becomes different when heavy elements and/or high doses are used. As shown in Fig.4, the response curves for As, Sb and BF2 deviate significantly from this universal curve for higher doses (cfr the top axis for the 5 keV-As-implant). In this case the simple calculation of the number of vacancies produced, based on the results in Fig.2, no longer holds as they start to overlap and the Si CI signal (uV) 1.E+13 1.E+14 10000 -10000 -30000 -50000 -70000 -90000 Non overlapping -110000 vacancies -130000 -150000 1.E+14 1.E+15 1.E+16 1.E+15 1.E+16 -1.E+05 BF2, 1e15 cm-2 BF2 5e15 cm-2 IIT-ref.3 -2.E+05 -3.E+05 -4.E+05 0 50 100 Amorphous depth (nm) 150 FIGURE 5. Correlation between CI-signal and amorphous depth for As (5, 10, 30 keV), BF2 (2,5,10 keV) and Ge-implants (ref.). The absolute CI-signals from ref [3] have been changed of sign and multiplied by 3. 5keV As-dose (5 kev)(at/cm2) 1.E+12 As, 1e15 cm-2 10 keV 1.E+17 10 30 keV CI as a probe for electrical activation 50 200 Originally conceived as a method to probe junction depths of annealed samples [1], considerable effort has been devoted to the correlation between SIMS)junction depths and the CI-signal [6]. For an ideal box-like profile, it can theoretically be shown that the signal response (CI versus junction depth) follows a cosine behavior. Of course a deviation for very thin profiles is observed (Fig.6), as all the curves need to converge to the same bulk value for zero thickness. For more gradual profiles it has been shown that the technique is not measuring the actual junction depth (i.e. the depth at which a p/n or Sb 5 keV Sb 2 KeV amorphization 1.E+17 1.E+18 Vacancy Dose (cm -2 ) BF2, 2 keV BF2 5 keV 1.E+19 BF2 10 keV FIGURE 4. CI-response plotted as a function of the number of vacancies produced during the implant for different doses and energies. All As-implants except otherwise indicated. The corresponding dose for 5 keV As is indicated at the top. The solid dashed line refers to the “universal” response curve as can be deduced from in fig.3. 760 p+/p: 1.00E+19 p+/p: 1.00E+20 (4e19 actual) p+/p: 5.00E+19 p+.p: CVD1:/9e19 p+/p: CVD1 1e20 p+.p: CVD1:/1e20 p+.p: CVD1-2/1.2e20 p+.p: CVD1/1.5e20 p+/p: 3.00E+20 p+/n/p: CVD1 1e20 (refer) i/p+/n: 1e19 p+/n: 5e19 i/p+/n: 5e19 4.0E+04 3.0E+04 CI- signal uV no anneal CVD 2.0E+04 anneal 1.0E+04 LTA 1&2 0.0E+00 0 -1.0E+04 200 400 600 CVD LTA 800 1000 CVD2: level 5e19 -2.0E+04 CVD1: nom. 1e20 4e20 (?) -3.0E+04 nom 3e20 -4.0E+04 SIMS@1e19 Depth (A) FIGURE 6 : CI-response as a function of junction depth (SIMS) for box like profiles grown with different concentrations levels. Grown by low temperature CVD, the indicated concentration levels may not be entirely active (cfr Fig 8-9). Arrows indicate the measurement values for the samples discussed in Fig.8-10 together with their corresponding junction depths (dashed lines. n/p junction occurs) but rather the depth of a particular concentration level [2]. This level is determined by the amount of excess carrier generation induced by the generation laser. Probing the sample with different excitations levels thus provides a way to determine profile abruptness [2]. As pointed out in Fig.6 the CI-response depends not only on the junction depth (here taken as the 1e19 at/cm3 in the SIMS profiles) but also very strongly on the concentration. It should be noted that the calibration curves in Fig.6 are based on the SIMSconcentration levels which may not be entirely (7080%?) active as the CVD-layers were grown at low temperature (700°C). Hence some uncertainty in the correlation to dopant concentration may exist (see also the following section on annealed CVD-layers). However the data sufficiently support the observation (Fig.7) that for a fixed junction depth the concentration sensitivity is as high as 2.5 104 µV/dec for concentrations in excess of 1 1019 at/cm3 or a sensitivity in the order of 0.1 % is present. CI-signal (µ V) 4.E+04 3.E+04 2.E+04 1.E+04 0.E+00 1.E+18 1.E+19 1.E+20 1.E+21 3 Concentration (at/cm ) FIGURE 7. Concentration dependence of CI-signal for a box like profile (junction 12 nm). This makes CI an extremely sensitive tool to probe dopant activation. The sensitivity to electrical activation is exemplified in a study on a series of box-profiles (grown by CVD at 700°C) which after growth, were subjected to a further thermal treatment i.e. an anneal at 700°C (5-10 min) and 800°C (1-2 761 layers. However again the observed strong decrease in signal illustrates the excellent sensitivity of the CItechnique to the electrical properties of the samples (change in CI-signal by 50% for a 20% change in concentration). 1.E+21 2.5-700C-5min 2.6-700-6min 2.7-700-7min 2.8-700-9min 2.10-800-1min 2.11-800-2min 1.5- unannealed (p+/p) sheet res matched Concentration (at/cm 3) 1.E+20 In a similar study we have analysed an unannealed CVD-layers versus a box profile produced by laser thermal annealing. Whereas the SIMS-profiles are almost identical (Fig.10) the LTA-sample has a sheet resistance of 178 Ohm/sq and the CVD-layer of only 356 Ohm/sq, corresponding respectively to an active concentration level of 1 1020 and 4 1019 at/cm3. The CI-signal changes from –13000 (LTA) to –7100 µV (CVD). Again such a large difference in CI-signal for the same junction depth (40 nm) can only be explained as the result of a very different activation level (~1 1020 versus 3 1020 at/cm3, Fig.6). 1.E+19 apparent electrical profile before anneal (70% activation) 1.E+18 1.E+17 0 5 10 15 20 25 30 Depth (nm) FIGURE 8 : SIMS profiles of annealed B-doped CVDlayers. 1.E+21 LTA min). The corresponding SIMS profiles, shown in Fig.8, indicate the absence of any diffusion upon those treatments. On the other hand sheet resistance measurements on these wafers show a clear reduction from 1800 Ohm/sq to 1400 Ohm/sq implying an increased activation of the Boron in the annealed samples from 4x1019 at/cm3 to 5x1019 at/cm3. CI-measurements on these samples show a similar change (Fig.9). In this case the signal changes from ~28,000 to 16-17,000. Based on the data in Fig.6 the active dopant concentration appears to increase by a factor, which is larger than the observed decrease in sheet resistance. This discrepancy indicates that the calibration levels in Fig.6 are inaccurate (all based on non-annealed CVD-layers) and need to be remeasured on annealed CVD- Concentration (cm-3) 1.E+20 0 CI signal (µ V) 10 20 30 40 50 Depth (nm) 60 70 80 FIGURE 10. SIMS–profiles of unannealed CVD-layer and laser thermal annealed implant profile. In a final example we used the CI-technique to study different laser anneal cycles. Again the resulting SIMS profiles are very similar with a minor difference on the junction depth (13 and 15 nm) (Fig.11). Sheet resistance measurements show values as low as 187 and 246 Ohm/sq, which would correspond to an active concentration as high as 4 1020 at/cm3. As this is much above the solid solubility and at the same time extremely shallow, concerns on the correctness of the sheet resistance values (and thus the concentration levels) could be raised as probe penetration (potentially as high as 100 nm) would lead to a sampling of the substrate and an equally low sheet resistance value [7]. CImeasurements show a dramatic difference between both samples 14000 µV for the shallowest 246 Ohm/sq-sample versus -17000 µV for the deeper one. Again based on Fig.6 one must conclude that increased activation 24000 22000 20000 18000 16000 14000 12000 1500 1.E+18 1.E+16 28000 10000 1000 1.E+19 1.E+17 30000 26000 CVD 2000 sheet resistance (ohm/sq) FIGURE 9. CI-signal versus sheet resistance for annealed and unannealed CVD-layers. Two different batches of CVD-layers/anneals lead to the same increase in activation upon anneal. 762 such a drastic change in CI-signal for a 2 nm difference in junction depth, can only occur for extremely high doping levels (~ 4-5 1020 at/cm3 ) confirming the validity of the sheet resistance measurements. 4. 5. 6. concentration (at/cm3) 1E+22 1E+21 7. 187 ohm/sq 1E+20 1E+19 246 ohm/sq 1E+18 1E+17 0 10 20 30 40 depth (nm) FIGURE 11. SIMS profiles of laser annealed implants. Based on the indicated sheet resistance values active concentrations would be as high as 4 1020 at/cm3. CONCLUSIONS A detailed study on the response function of Carrier Illumination has been performed as a probe for unannealed and annealed profiles. For low dose implants where no amorphization occurs, a universal response function can be defined relating the CIsignal to the number of vacancies produced during the implants. In the case of amorphization, a one-to one relation between the CI-signal and the amorphous depth can be observed. Finally it is shown that the CI-technique is extremely sensitive to the degree of activation and strong signal differences can be found for samples with nearly identical dopant profiles. ACKNOWLEDGMENTS The authors are grateful to E.Collart (AMAT) for the Sb-implants. REFERENCES 1. P.Borden, AIP Conf. Proc 550, 175 (2001) 2. Clarysse T., Vandervorst W., Lindsay R., Borden P., Budiarto E., Madsen J. and Nijmeijer R., “Junction and profile analysis using Carrier Illumination” in Silicon front-end junction formation technologies, edited by D.F.Downey, M.E.Law, A.Claverie and M.J.Rendon, MRS Symposium Proceedings Vol 717, Warrendale, Pennsylvania, 2002, pp.285-290. 3. Borden P., Ferguson C., Sing D., Larson L., Bechtler L., Jones K. and Gable P.,”In-line characterization of preamorphous implants (PAI)”, Proceedings of the 2000 International Conference on Ion Implantation 763 technology IIT-2000, 17-22 Sept. 2000, Alpbacb, Austria, pp. 635-638. Version SRIM-2000.40, J.F.Ziegler and J.P. Biersack P.Borden, E.Budiarto, to be published Clarysse T., Lindsay R., Vandervorst W., Budiarto E. and Borden P., “Carrier Illumination for the characterization of ultra shallow profiles”, proceedings USJ-2003, april 27-may 1, 2003, ed. P.Ronsheim (to be published in Journ. Vac. Sci. Techn.) Clarysse, T.; Vanhaeren, D. and Vandervorst, W., Journal. Vac. Sci. Techn. B., Vol. 20 (1) 459-446 (2002)
© Copyright 2026 Paperzz