Positron Annihilation Lifetime Spectra (PALS) for Defect Extraction C. H. Chen and Prof. M. H. Liao National Taiwan University 2019/9/1

Outline-1 Motivations for PALS/Has any specialty ? (1 Pages) PALS Principle (2 Pages) PALS Set-up (2 Pages) Previous Work Review for PALS IMP related defect extraction (2 Pages) SiGe related defect extraction (2 Pages) Defect information extraction by PALS in theory: Energy/Defect Depth (2 Pages) Defect Size (2 Pages) 2019/9/1

Outline-2 Defect information extraction by PALS in theory: Defect Concentration (2-3 Pages) Defect Type (V and I) (4 Pages) The demo results for TSMC’s sample (6-10 Pages) PALS and PL/Raman Spectra Back-Up Material: JAP, 1999. (2 Pages) Defect studies (V) on SiGe by HR-TEM. Another tool demonstration for V extraction. 2019/9/1

Motivations for PALS Why do we need PALS ? Does PALS have any specialty ? A powerful technique/tool for the defect extraction in the semiconductor industry. Non-destructive extraction tool. With the high resolution/capability for the defect analysis (size, concentration, type.) among the z-axis (depth) direction. With the high energy radiative light source EUV tool for the photo-exposure with high resolution 2019/9/1

Motivations for PALS Applications in PALS for future CMOS topics/device. Ge FET: Defect (energy level near Ec), Low solid solubility for Rs reduction (Defect recovery), SiGe Epi-quality, Ge/HK(Phonon scattering), Ge mobility (Alloy scattering). III-V FET: Dit (Mid-gap Fermi- pinning), Low solid solubility for Rs reduction (Defect recovery), Epi-layer quality, HK Junction Design: Co-IMP (C/N), V/I TED (B, P). Tri-Gate (100), (110), (111) surface properties. 2019/9/1

Outline-1 Motivations for PALS (1-2 Pages) PALS Principle (2 Pages) PALS Set-up (2 Pages) Previous Work Review for PALS IMP related defect extraction (2 Pages) SiGe related defect extraction (2 Pages) Defect information extraction by PALS in theory: Energy/Defect Depth (2 Pages) Defect Size (2 Pages) 2019/9/1

Principle on PALS (Positron Annihilation Lifetime Spectra) The emission of positrons from a radioactive source (t1). (Fast-Positron/Low-Poitron) Positron will interact with the electrons in the semiconductors and annihilate very rapidly. If there have defect in the material, positrons will reside and annihilate less rapidly on the time scales up to ~1 ns. (local electron density/electron momentum distribution) The annihilation releases gamma rays (t2) that can be detected. The time between emission of positrons from a radioactive source and detection of gamma rays due to an annihilation can reflect the characteristics of the material.  1274 keV t - PALS  511 keV t1 t2 Radioactive source High energy sample termalization e+ annihilation e- diffusion~ 100 nm  511 keV Positron Annihilation Lifetime Spectroscopy 2019/9/1

Principle on PALS (1. What/Why is POSITRON?) Radioactive light source with high energy for the high resolution. (EUV tool for the photo-exposure with high resolution) We use Na22 as light source MRI for cancer extraction 2019/9/1

Principle on PALS (Positron Annihilation Lifetime Spectra) The emission of positrons from a radioactive source (t1). (Fast-Positron/Low-Poitron) Positron will interact with the electrons in the semiconductors and annihilate very rapidly. If there have defect in the material, positrons will reside and annihilate less rapidly on the time scales up to ~1 ns. (local electron density/electron momentum distribution) The annihilation releases gamma rays (t2) that can be detected. The time between emission of positrons from a radioactive source and detection of gamma rays due to an annihilation can reflect the characteristics of the material.  1274 keV t - PALS  511 keV t1 t2 22Na sample termalization e+ annihilation e- diffusion~ 100 nm  511 keV Two key index: <1> lifetime and <2> r-line shape Positron Annihilation Lifetime Spectroscopy 2019/9/1

Principle on PALS (2. Positron Annihilation in Matter) Two key index: <1> lifetime and <2> r-line shape Defect Type Size Materials Atomic Vacancies 0.1 nm Metals Dislocations 1 nm - 10 mm Voids 0.1 nm - 1 mm Holes 0.1 nm - 10 mm Polymers t1 t2 r line shape Lifetime Local electron density and the electron momentum distribution The time between emission of positrons from a radioactive source and detection of gamma rays due to an annihilation can reflect the characteristics of the material. 2019/9/1

Outline-1 Motivations for PALS (1-2 Pages) PALS Principle (2 Pages) PALS Set-up (2 Pages) Previous Work Review for PALS IMP related defect extraction (2 Pages) SiGe related defect extraction (2 Pages) Defect information extraction by PALS in theory: Energy Depth (2 Pages) Defect Size (2 Pages) 2019/9/1

Instrument/Set-Up (PALS) a. Tsukuba (AIST), b. NTU, and c. Harvard/MIT NTU: Slow-positron-beam technique for surface extraction. PM—photomultiplier, SCA—single-channel analyzer, TAC—time-to-amplitude Converter, MCA—multi-channel analyzer . 2019/9/1

Instrument/Set-Up (PALS): Positron Beam of World Map 2019/9/1

Outline-1 Motivations for PALS (1-2 Pages) PALS Principle (2 Pages) PALS Set-up (2 Pages) Previous Work Review for PALS IMP related defect extraction (2 Pages) SiGe related defect extraction (2 Pages) Defect information extraction by PALS in theory: Energy/Defect Depth (2 Pages) Defect Size (2 Pages) 2019/9/1

Previous PALS studies on SiGe Many previous works show that PALS can be used to extract the information of defect in SiGe. Our structure is similar to others. (a) (b) w/ P+ 288 Kev, 6E13 cm-2 Strained Si 28.5 nm w/ F+ 185 Kev, 9E14 or 1.85E15 cm-2 Si0.75Ge0.25 1 mm Si 130 nm Graded SiGe 2 mm Si0.94Ge0.06 50 nm (B 5E18 cm-3) Si Si J. Appl. Phys (2005) Vacancy-type defects in strained-Si layers deposited on SiGe∕Si structures probed by using monoenergetic positron beams J. Appl. Phys. (2007) Fluorine-vacancy complexes in Si-SiGe-Si structures (c) Relaxed SiGe (10%-30%) (d) w/ P+ 288 Kev, 6E13 cm-2 w/ F+ 185 Kev, 2.3E15 or 1E16 cm-2 1 mm Graded SiGe (10%-30%) Si 130 nm 1-3 mm Si0.89Ge0.11 40 nm(B 1.2E19 cm-3) p-Si (100) Si 400 nm Anneal: 800oC 20s J. Appl. Phys. (2012) Positron annihilation studies of fluorine-vacancy complexes in Si and SiGe J. Appl. Phys. (2007) Fluorine-vacancy complexes in Si-SiGe-Si structures 2019/9/1 9/1/2019 15

PALS measurement results on SiGe-1 “(a) S” parameter versus E (depth) and (b) Lifetimes of positrons in SiGe can indicate the information of defect in SiGe. (a) (b) S Defect Lifetime Defect J. Appl. Phys (2005) Vacancy-type defects in strained-Si layers deposited on SiGe∕Si structures probed by using monoenergetic positron beams Anneal Depth S parameter for Si/Si0.75Ge0.25/graded-SiGe/Si before and after annealing at 1050 °C (30 or 60 min). The S-E curves are fittings of the diffusion equation for positrons to the experimental data (Ref. 21). The inset shows a close up of the S-E curves in the range of E=0–5 keV. Lifetimes of positrons and their intensities for Si/ Si0.75Ge0.25/graded-SiGe/Si before and after annealing. The lifetime spectra of the positrons were measured at E=1 keV and decomposed into two components: T1 and I1 are the lifetime of the ith component and its intensity, respectively (i=1 and 2; I1+I2=1). 2019/9/1 16

PALS measurement results on SiGe-2 “S” parameter versus E (depth) also can investigate the implant induced defect in SiGe.-1 S parameter versus incident positron energy for samples implanted with 185 keV F+ and annealed at 950 °C for 30 s. The data for an as-grown sample are also shown. The solid lines are fits obtained using VEPFIT. Depth J. Appl. Phys. (2007) Fluorine-vacancy complexes in Si-SiGe-Si structures 9/1/2019 2019/9/1 17

PALS measurement results on SiGe-3 “S” parameter versus E (depth) can investigate both (a) structures defect and (b) implant induced defect in different SiGe structures. Normalized S(E) plot for as-implanted relaxed and multi-layer samples of 10% and 30%Ge. All depth dependent data are shown for the annealed relaxed sample of 10%Ge (a) and the multi-layer sample of 30%Ge (b). The left-hand axis corresponds to the normalized S(E) plot and the right-hand axis corresponds to the SIMS F concentration data. Other plots include the two VEPFIT regions of F complexes and vacancy-rich defects and the initial Si and Ge vacancy and F ion profiles. Normalized S(E) plot for annealed relaxed and multi-layer samples of 10%–30%Ge. J. Appl. Phys. (2012) Positron annihilation studies of fluorine-vacancy complexes in Si and SiGe 2019/9/1 18

Outline-1 Motivations for PALS (1-2 Pages) PALS Principle (2 Pages) PALS Set-up (2 Pages) Previous Work Review for PALS IMP related defect extraction (2 Pages) SiGe related defect extraction (2 Pages) Defect information extraction by PALS in theory: Energy/Defect Depth (2 Pages) Defect Size (2 Pages) 2019/9/1

Defect information extraction by PALS in theory: Depth (4 Pages) How to calculate the depth by the positron intensity? It strongly depends on the density and the positron diffusion coeff. of the tested sample (Si, Ge, SiGe). Positron diffusion coeff. depends on the optical/acoustic phonon scattering in the materials.     Mean Depth Positron diffusion Coe. Density Temperature 2019/9/1

Outline-1 Motivations for PALS (1-2 Pages) PALS Principle (2 Pages) PALS Set-up (2 Pages) Previous Work Review for PALS IMP related defect extraction (2 Pages) SiGe related defect extraction (2 Pages) Defect information extraction by PALS in theory: Energy/Defect Depth (2 Pages) Defect Size (2 Pages) 2019/9/1

Defect information extraction by PALS in theory: Defect Size (4 Pages) 22Na t - PALS e+  1274 keV 511 keV sample e- termalization diffusion~ 100 nm annihilation How to extract the defect size ?   -1       Lifetime Size 2019/9/1

Defect information extraction by PALS in theory: Defect Size (4 Pages) These shortcomings of the prevalent model could in principle be overcome by introducing length parameters describing the diffusivity of the wall, its roughness and corrections due to finite Ps size. However, this would be too elaborate for the present purpose and we shall introduce these effects through a lumped parameter Δwith the dimension of length We take a more realistic density profile which smoothly interpolates between the value zero at the Centre of the cavity (r = 0) and the bulk density ρ0 as r becomes large: with Δ→ 0, ρ(r ) → ρ(0) (r − R) so that for small Δ/R the density profile is basically a spherical cavity with a little diffusivity in the boundary. Again we adopt the self-trapping potential to be of the same shape as ρ(r ), namely.( being the Heaviside step function which is zero for r < R and unity for r > R), Electron density Potential step function which is the Woods–Saxon (WS) potential widely used by nuclear physicists. Since Δ/R is expected (and in fact found) to be rather small we may treat the diffuseness as a perturbation on the SW(spherical well). The shift in energy from the SW value is then given by 2019/9/1

Defect information extraction by PALS in theory: Defect Size (4 Pages) To the same order of accuracy the perturbed wavefunction has the form of the unperturbed solution (equation (4)) except that k0 and K0 are replaced by k0 and K0 which solve the modified eigenvalue condition Eigen-value for momentum conservation Where and he energy E0 being given by the eigenvalue condition The introduction of diffusivity in the bubble surface enables us to incorporate in a natural manner the influence of the radius of curvature on the surface energy leading to the notion of an effective surface tension (σeff ) in place of the bulk value (σ ). via a formula put forward by: 2019/9/1

Defect information extraction by PALS in theory: Defect Size (4 Pages) and accordingly the surface energy of the bubble, instead of being 4πR2ρwill now be In the present version, however, the total energy depends (see equation (3)) both on R and Δand as a consequence it must be minimized with respect to each, and hence the single condition is now replaced by two (in the case of liquids), namely BC: Therefore, even though a new parameter has been introduced, at least for liquids, we havean additional minimizing condition so that we have no more free parameters than the primitive model which we have corrected 2019/9/1

Defect information extraction by PALS in theory: Defect Size (4 Pages) However such a totally confined Ps does not penetrate the walls of the cavity and thus is unable to indulge in pick-off annihilation. To amend this unrealistic situation a virtual electron layer of thickness R(= R − R0) was added to the inside wall of the well and the overlap of the Ps wavefunction with this electron layer of density ρ0 was taken to parametrize the pick-off decay of the trapped o-Ps. The resulting formula becomes 2019/9/1

Outline-2 Defect information extraction by PALS in theory: Defect Concentration (2-3 Pages) Defect Type (V and I) (4 Pages) The demo results for TSMC’s sample (6-10 Pages) PALS and PL/Raman Spectra Back-Up Material: JAP, 1999. (2 Pages) Defect studies (V) on SiGe by HR-TEM. Another tool demonstration for V extraction. 2019/9/1

Defect information extraction by PALS in theory: Defect Concentration (4 Pages) 22Na t - PALS e+  1274 keV 511 keV sample e- termalization diffusion~ 100 nm annihilation How to extract the defect concentration ?     C = defect concentration   Trapping Time Trapping Time Vacancy concentration Vacancy concentration 2019/9/1

Annihilation radiation Defect information extraction by PALS in theory: Defect Concentration (4 Pages) Concentration base on Trapping in Independent Defect Type All three defects are considered to be non-interacting. The corresponding different equation are Defect d1 Defect d2 Defect d3 Defect-free bulk positron Annihilation radiation The Lifetimes τ1 to τ4 are obtained as : 2019/9/1

Annihilation radiation Defect information extraction by PALS in theory: Defect Concentration (4 Pages) Concentration base on Trapping in Independent Defect Type Defect d1 Defect d2 Defect d3 Defect-free bulk positron Annihilation radiation The intensities I1 to I4 are obtained as : 2019/9/1

Defect information extraction by PALS in theory: Defect Concentration (4 Pages) Concentration base on Trapping in Independent Defect Type :μ=Trapping coefficients Material Defect μ(10-8)cm3s-1 T(K) Reference method Si <0.2 300 Hall effect 20 GaP 473 GaAs:Te GaAs:Si 2019/9/1

Outline-2 Defect information extraction by PALS in theory: Defect Concentration (2-3 Pages) Defect Type (V and I) (4 Pages) The demo results for TSMC’s sample (6-10 Pages) PALS and PL/Raman Spectra Back-Up Material: JAP, 1999. (2 Pages) Defect studies (V) on SiGe by HR-TEM. Another tool demonstration for V extraction. 2019/9/1

Defect information extraction by PALS in theory: Defect Type (V and I) (4 Pages) Two key index: <1> lifetime and <2> r-line shape 22Na t - PALS e+  1274 keV 511 keV sample e- termalization diffusion~ 100 nm annihilation How to extract the defect type (V/I)? 1. Lifetime: Atomic Vacancies (a) Interstitials Vacancies Intensity (a. u.) Defect type Potential Energy Electron density Lifetime (ps) (b) Vacancies defect in semiconductor (c) Different kinds of defect in Ge. 2019/9/1

Defect information extraction by PALS in theory: Defect Type (V and I) (4 Pages) Two key index: <1> lifetime and <2> r-line shape How to extract the defect type (V/I)? 22Na t - PALS e+  1274 keV 511 keV sample e- termalization diffusion~ 100 nm annihilation 2. Emission Gamma Ray Profile S-parameter W-parameter S-parameter defect concentration (S/W ratio) J. Appl. Phys (2005) Vacancy-type defects in strained-Si layers deposited on SiGe∕Si structures probed by using monoenergetic positron beams 2019/9/1

Defect information extraction by PALS in theory: Defect Type (V and I) (4 Pages) Two key index: <1> lifetime and <2> r-line shape How to extract the defect type (V/I)? 22Na t - PALS e+  1274 keV 511 keV sample e- termalization diffusion~ 100 nm annihilation 2. Emission Gamma Ray Profile The (S,W) values denoted by Ge, Si, and V2 in Si are the characteristic values of (S,W) for positron annihilation in defect-free Ge, defect-free Si, and Si-implanted Si, respectively. I- Interstitial I0 I+ W-parameter V V2- Vacancy V20 V2++ S-parameter 2019/9/1

Defect information extraction by PALS in theory: Defect Type (V and I) (4 Pages) Two key index: <1> lifetime and <2> r-line shape 22Na t - PALS e+  1274 keV 511 keV sample e- termalization diffusion~ 100 nm annihilation How to extract the defect type ? Vacancy/Interstitial. “Charged” defect V+, V0, V-, V--. - + Lifetime for “Charged” defect. 2019/9/1

Outline-2 Defect information extraction by PALS in theory: Defect Concentration (2-3 Pages) Defect Type (V and I) (4 Pages) The demo results for TSMC’s sample (6-10 Pages) PALS and PL/Raman Spectra Back-Up Material: JAP, 1999. (2 Pages) Defect studies (V) on SiGe by HR-TEM. Another tool demonstration for V extraction. 2019/9/1

TSMC Sample Prior Test and Demonstration -1: Epi-Ge layer on Si NTJ039918.01#25-B has less defect than NTJ039918.01#24-A. S parameter (S defect ) Ge Si Positron Energy (keV) E Penetration depth 2019/9/1 2019/9/1 38

TSMC Sample Prior Test and Demonstration -1: Epi-Ge layer on Si NTJ039918.01#25-B has less defect than NTJ039918.01#24-A. 2019/9/1

TSMC Sample Prior Test and Demonstration -1: Epi-Ge layer on Si (V0, V-, V+) NTJ039918.01#25-B has less defect than NTJ039918.01#24-A. 2019/9/1

TSMC Sample Prior Test and Demonstration -1: Epi-Ge layer on Si Raman Spectra indicate: #25-B has better quality than #24-A. (consistent with PALS) Condition : Wavelength:488nm, Integration time:60s Ge(91nm) Si-Sub Sample’s Structure M1 M2 K 2019/9/1 2019/9/1 41

TSMC Sample Prior Test and Demonstration -1: Epi-Ge layer on Si PL Spectra indicates: #25-B has better quality than #24-A. (consistent with PALS) Ge(91nm) Si-Sub Sample’s Structure 2019/9/1

TSMC Sample Prior Test and Demonstration -2: Epi-SiGe layer on Si N53H68.J5#7 has less defect than N53H68.J7#11. (N53H68.J5#7 should have better device performance than N53H68.J7#11) S parameter (S defect ) Has special process treatment ? SiGe Si Positron Energy (keV) E Penetration depth 2019/9/1 2019/9/1 43

TSMC Sample Prior Test and Demonstration -2: Epi-SiGe layer on Si (Conti.) The defect size in SiGe sample is at ~ 0.2 nm level. The defect size in J5#7 is smaller than J7#11. Big Big Defect Size Small Defect Size Defect Size Small Depth 2019/9/1 2019/9/1 44

TSMC Sample Prior Test and Demonstration -2: Epi-SiGe layer on Si (Conti.) The defect concentration can be extracted by o-Ps Intensity. J5#7 has low defect concentration than J7#11, because it has lower o-Ps Intensity. High Higher Defects Concentration Defect Concentration Lower Defects Concentration Low Depth 2019/9/1 2019/9/1 45

TSMC Sample Prior Test and Demonstration -2: Epi-SiGe layer on Si (Conti.) Raman Spectra indicate: J5#7 has better quality than J7#11. (consistent with PALS) Condition : Wavelength:488nm, Integration time:60s SiGe (70nm ?) Si-Sub Sample’s Structure 2019/9/1 2019/9/1 46

TSMC Sample Prior Test and Demonstration -2: Epi-SiGe layer on Si (Conti.) PL Spectra indicates: J5#7 has better quality than J7#11. (consistent with PALS) SiGe (70 nm ?) Si-Sub Sample’s Structure 2019/9/1 2019/9/1 47

TSMC Sample Prior Test and Demonstration -3: Epi-SiGe layer on Si (Conti.) New Sample SiGe (15nm ) Si-Sub N53H68.J5#7(3) Defect SiGe (230nm ) Si-Sub N53H68.J7#11(4) Defect See TEM SiGe (230nm ) Si-Sub N53H70.J7#2(5) Defect 2019/9/1 2019/9/1 48

TSMC Sample Prior Test and Demonstration -3: Epi-SiGe layer on Si SiGe (15nm ) Si-Sub N53H68.J5#7(3) Defect S parameter (S defect ) SiGe (230nm ) Si-Sub N53H68.J7#11(4) Defect New Sample SiGe (230nm ) Si-Sub N53H70.J7#2(5) Defect Positron Energy (keV) E Penetration depth 2019/9/1

TSMC Sample Prior Test and Demonstration -3: Epi-SiGe layer on Si (Conti.) New Sample SiGe (15nm ) Si-Sub N53H68.J5#7(3) Defect SiGe (230nm ) Si-Sub N53H68.J7#11(4) Defect SiGe (230nm ) Si-Sub New sample(5) Defect 2019/9/1 2019/9/1 50

TSMC Sample Prior Test and Demonstration -3: Epi-SiGe layer on Si (Conti.) PALS, PL, Raman indicates: J5#7 has better quality than J7#11. We check the TEM image on J7#11 and indeed can find out many defects. The most of defects concentrated in the interface Interface Si SiGe Interface Defect Defect 2019/9/1 2019/9/1 51

TSMC Sample Prior Test and Demonstration -3: Epi-SiGe layer on Si (Conti.) PALS, PL, Raman indicates: J5#7 has better quality than J7#11. We check the TEM image on J7#11 and indeed can find out many defects. The defects most of the centralized in the interface SiGe Si Defect Interface Si SiGe 2019/9/1 2019/9/1 52

Outline-2 Defect information extraction by PALS in theory: Defect Concentration (2-3 Pages) Defect Type (V and I) (4 Pages) The demo results for TSMS’s sample (6-10 Pages) PALS and PL/Raman Spectra Back-Up Material: JAP, 1999. (2 Pages) Defect studies (V) on SiGe by HR-TEM. Another tool demonstration for V extraction. 2019/9/1

Previous Defect studies on SiGe by high-voltage electron microscope ~HR-TEM Split-interstitial atom b c [110] HREM image of a [113] defect in strained SiGe with 0.3% misfit. The small distortion of the [113] planes parallel to the defect plane is shown by white lines. (b) An atomic model of the [113] defect is superimposed on the image. (c) A simulated HREM image of a [113] defect based on the unit cell indicated in (b) by the white rectangle. J. Appl. Phys (1999) In situ HREM irradiation study of point-defect clustering in MBE-grown strained Si1-xGex / (001)Si structures 2019/9/1