Wet-chemical preparation of silicon by oxidation followed by hydrofluoric-acid (HF) etch leaves a surface covered by covalent Si–H bonds with a surface-recombination velocity (SRV) as low as 0.25 cm/s on Si(111) and about 2 cm/s on ultrapure germanium, values that after four decades still rank with the best chemical passivation ever reported. The microscopic identity of the small residual population of recombination-active defects that remains on such nominally ideal, oxide-free surfaces, the quantitative mechanism by which acidic and related solutions further suppress recombination, and the physical ultimate limit of the SRV remain unestablished. This problem asks for a microscopic, quantitative account of residual recombination at chemically (hydrogen-)passivated Si and Ge surfaces: identification of the residual recombination centers and their electronic levels, elucidation of the chemical passivation channel (the 1986 acid–base equilibrium model is phenomenological), and determination of the ultimate achievable SRV and of the process that sets that floor, together with its stability and transferability across orientations, dopings, and Ge surfaces.

Background

When excess carriers are injected into a semiconductor wafer, they recombine not only in the bulk but also at the surfaces, where the abrupt termination of the crystal creates broken ('dangling') bonds. These act as electronic gap states that capture carriers and mediate recombination; the strength of this loss channel is summarized by the surface-recombination velocity SS (in cm/s), the proportionality constant between the flux of carriers disappearing at the surface and the excess carrier density, so that in a wafer of thickness LL the surface adds 2S/L2S/L to the bulk decay rate of the excess carrier density. Rendering surfaces electrically inactive ('passivation') is essential for silicon solar cells, whose efficiency ceiling is set by recombination losses, and for measuring the intrinsic bulk carrier lifetime of high-quality material.

Two conceptually distinct passivation mechanisms exist. Chemical passivation removes the recombination centers themselves, e.g. by terminating dangling bonds with hydrogen or other monovalent species. Field-effect passivation does not remove defects but suppresses recombination electrostatically: fixed charge near the surface bends the bands and repels one carrier type from the defective region. In 1986, Yablonovitch, Allara, Chang, Gmitter and Bright showed that an ordinary preparation — thermal oxidation of Si(111) followed by dissolution of the oxide in HF — yields S=0.25S = 0.25 cm/s, by far the lowest value reported for any semiconductor at the time. Multiple-internal-reflection infrared spectroscopy (an IR technique in which light is bounced many times along the surface to amplify monolayer-level absorption) showed the surface to be covered by about a monolayer of covalent Si–H bonds, with only trace fluorine, so that essentially no dangling bonds remain to act as recombination centers. Interpreting S=σvthNtS = \sigma v_{\mathrm{th}} N_t (capture cross section σ≈10−15\sigma \approx 10^{-15} cm2^2 times thermal velocity vth≈107v_{\mathrm{th}} \approx 10^7 cm/s times areal defect density NtN_t) bounds the residual center density at fewer than about one defect per 4×1074\times 10^7 surface atoms. The authors reported that the low SS was insensitive to carrier density and doping, indicating that surface band bending was not the cause. Immersing the surface in concentrated acids reversibly lowered SS further, which they modeled with a phenomenological acid–base equilibrium x/(1−x)=Keq/Mx/(1-x) = K_{\mathrm{eq}}/M (with MM the acid molarity, xx the unreacted fraction, and S∝xS \propto x) in which acid coordination neutralizes a small finite population, of order 10910^9 cm−2^{-2}, of strained, weak surface bonds hypothesized to be the residual recombination centers. Applying similar chemistry to ultrapure germanium gave SS as low as 2 cm/s. A by-product of the long bulk lifetimes (up to 35 ms in float-zone Si, i.e. silicon grown by the crucible-free float-zone method, which yields exceptionally pure crystals) made accessible by this passivation was a low-carrier-density Auger recombination coefficient (Auger recombination: a three-carrier process whose rate ∝Cn2p\propto C n^2 p or Cnp2Cnp^2; the paper found CAuger≈2×10−30C_{\mathrm{Auger}} \approx 2\times 10^{-30} cm6^6/s at low density, well above the then-accepted value extrapolated from high-density measurements).

Later work substantially clarified the Auger part of this picture: band-to-band Auger recombination in silicon is Coulomb enhanced (strengthened by electron–hole correlation effects) at low carrier densities, as predicted by Hangleiter and Häcker and incorporated into the standard modern parameterizations of Kerr and Cuevas and of Richter et al. The surface part, by contrast, is still handled phenomenologically: the successor literature on temporary chemical passivation (iodine–ethanol, superacids, i.e. acids stronger than pure sulfuric acid, aprotic and Lewis-acid solutions) fits measured effective lifetimes with a combination of a chemical channel and a field-effect channel from an adsorbed charged layer, without identifying the residual centers or the chemistry of the chemical channel. The unresolved questions are a spectroscopic and chemical identification of the recombination-active sites that remain on an oxide-free, H-terminated Si (or Ge) surface, a quantitative account of the mechanism by which acids and related species suppress them, and a determination of the ultimate SRV floor and the physical process that sets it.

Problem Statement

Provide a microscopic, quantitative account of residual surface recombination at chemically hydrogen-passivated silicon and germanium surfaces, i.e. surfaces prepared by oxidation followed by HF etch (or equivalent wet chemistry) that are covered by a near-monolayer of covalent Si–H (or Ge–H) bonds and are essentially free of native oxide. Specifically:

  1. Identify the microscopic structure, areal density, and electronic energy levels (positions within the band gap, capture cross sections) of the residual recombination-active sites on the oxide-free, H-terminated Si surface — the finite population that the 1986 work hypothesized to be of order 10910^{9} cm−2^{-2} strained, weak surface bonds (a population this large is compatible with the measured SS only if the centers' effective capture cross sections are far below the 10−1510^{-15} cm2^{2} assumed in the naive estimate S=σvthNtS = \sigma v_{\mathrm{th}} N_t, which caps a uniformly active population at roughly 2.5×1072.5\times 10^{7} cm−2^{-2}) — and establish their chemical origin (step edges, microfacets, dihydride/trihydride species, sub-surface strain, or contaminants).

  2. Determine the quantitative mechanism by which contact with concentrated acids (and the modern successors: superacids, bis(trifluoromethanesulfonyl) solutions, Lewis acids) suppresses recombination on such surfaces: which species bind or adsorb, with what geometry and thermodynamics/kinetics, how the chemical channel is to be separated from the field-effect channel of the charged adsorbed layer, and whether the 1986 acid–base equilibrium x/(1−x)=Keq/Mx/(1-x) = K_{\mathrm{eq}}/M corresponds to a real chemical equilibrium (and if so, of what reaction).

  3. Establish the ultimate limit of the surface-recombination velocity achievable at chemically passivated Si and Ge surfaces — how far below the 1986 values (0.250.25 cm/s on Si(111), about 22 cm/s on Ge) it is physically possible to go, and which residual process (unpassivable defect tail, the chemical channel's saturation, or an electrostatic/transport floor) sets that limit — and characterize how this limit depends on ambient and processing exposures, crystallographic orientation, doping type and level, and transferability from Si to Ge.

References

  1. E. Yablonovitch, D. L. Allara, C. C. Chang, T. Gmitter, T. B. Bright, 'Unusually Low Surface-Recombination Velocity on Silicon and Germanium Surfaces', Phys. Rev. Lett. 57, 249–252 (1986), DOI: 10.1103/PhysRevLett.57.249, https://doi.org/10.1103/PhysRevLett.57.249
  2. A. Hangleiter, R. Häcker, 'Enhancement of band-to-band Auger recombination by electron-hole correlations', Phys. Rev. Lett. 65, 215–218 (1990), DOI: 10.1103/PhysRevLett.65.215, https://doi.org/10.1103/PhysRevLett.65.215
  3. R. A. Sinton, A. Cuevas, M. Stuckings, 'Quasi-steady-state photoconductance, a new method for solar cell material and device characterization', Proc. 25th IEEE Photovoltaic Specialists Conference, 457–460 (1996), DOI: 10.1109/pvsc.1996.564042, https://doi.org/10.1109/pvsc.1996.564042
  4. A. W. Stephens, M. A. Green, 'Effectiveness of 0.08 molar iodine in ethanol solution as a means of chemical surface passivation for photoconductance decay measurements', Sol. Energy Mater. Sol. Cells 45, 255–265 (1997), DOI: 10.1016/S0927-0248(96)00061-X, https://doi.org/10.1016/S0927-0248(96)00061-X
  5. K. Kurita, T. Shingyouji, 'Low Surface Recombination Velocity on Silicon Wafer Surfaces due to Iodine-Ethanol Treatment', Jpn. J. Appl. Phys. 38, 5710 (1999), DOI: 10.1143/jjap.38.5710, https://doi.org/10.1143/jjap.38.5710
  6. Armin G. Aberle, 'Surface passivation of crystalline silicon solar cells: a review', Prog. Photovolt. Res. Appl. 8, 473–487 (2000), DOI: 10.1002/1099-159x(200009/10)8:5<473::aid-pip337>3.0.co;2-d, https://doi.org/10.1002/1099-159x(200009/10)8:5%3C473%3A%3Aaid-pip337%3E3.0.co;2-d
  7. J. Schmidt, M. Kerr, P. P. Altermatt, 'Coulomb-enhanced Auger recombination in crystalline silicon at intermediate and high injection densities', J. Appl. Phys. 88, 1494 (2000), DOI: 10.1063/1.373878, https://doi.org/10.1063/1.373878
  8. M. J. Kerr, A. Cuevas, 'General parameterization of Auger recombination in crystalline silicon', J. Appl. Phys. 91, 2473 (2002), DOI: 10.1063/1.1432476, https://doi.org/10.1063/1.1432476
  9. D. J. Michalak, F. Gstrein, N. S. Lewis, 'The Role of Band Bending in Affecting the Surface Recombination Velocities for Si(111) in Contact with Aqueous Acidic Electrolytes', J. Phys. Chem. C 112, 5911 (2008), DOI: 10.1021/jp075354s, https://doi.org/10.1021/jp075354s
  10. B. Hoex, J. Schmidt, P. Pohl, M. C. M. van de Sanden, W. M. M. Kessels, 'Silicon surface passivation by atomic layer deposited Al2O3', J. Appl. Phys. 104, 044903 (2008), DOI: 10.1063/1.2963707, https://doi.org/10.1063/1.2963707
  11. K. Koyama, K. Ohdaira, H. Matsumura, 'Extremely low surface recombination velocities on crystalline silicon wafers realized by catalytic chemical vapor deposited SiNx/a-Si stacked passivation layers', Appl. Phys. Lett. 97, 082108 (2010), DOI: 10.1063/1.3483853, https://doi.org/10.1063/1.3483853
  12. H. Matsumura, M. Miyamoto, K. Koyama, K. Ohdaira, 'Drastic reduction in surface recombination velocity of crystalline silicon by surface treatment using catalytically-generated radicals', Sol. Energy Mater. Sol. Cells 95, 797 (2011), DOI: 10.1016/j.solmat.2010.08.034, https://doi.org/10.1016/j.solmat.2010.08.034
  13. B. P. Swain, H. Takato, I. Sakata, 'Passivation of Germanium Surfaces by a Quinhydrone–Methanol Solution Treatment', Jpn. J. Appl. Phys. 50, 071302 (2011), DOI: 10.7567/jjap.50.071302, https://doi.org/10.7567/jjap.50.071302
  14. A. Richter, S. W. Glunz, F. Werner, J. Schmidt, A. Cuevas, 'Improved quantitative description of Auger recombination in crystalline silicon', Phys. Rev. B 86, 165202 (2012), DOI: 10.1103/PhysRevB.86.165202, https://doi.org/10.1103/PhysRevB.86.165202
  15. J. Bullock, D. Kiriya, N. Grant, A. Azcatl, M. Hettick, T. Kho, P. Phang, H. C. Sio, D. Yan, D. Macdonald, M. A. Quevedo-Lopez, R. M. Wallace, A. Cuevas, A. Javey, 'Superacid Passivation of Crystalline Silicon Surfaces', ACS Appl. Mater. Interfaces 8, 24205 (2016), DOI: 10.1021/acsami.6b07822, https://doi.org/10.1021/acsami.6b07822
  16. N. E. Grant, T. Niewelt, N. R. Wilson, E. C. Wheeler-Jones, J. Bullock, M. Al-Amin, M. C. Schubert, A. C. van Veen, A. Javey, J. D. Murphy, 'Superacid-Treated Silicon Surfaces: Extending the Limit of Carrier Lifetime for Photovoltaic Applications', IEEE J. Photovolt. 7, 1574 (2017), DOI: 10.1109/jphotov.2017.2751511, https://doi.org/10.1109/jphotov.2017.2751511
  17. N. E. Grant, J. D. Murphy, 'Temporary Surface Passivation for Characterisation of Bulk Defects in Silicon: A Review', Phys. Status Solidi RRL 11, 1700243 (2017), DOI: 10.1002/pssr.201700243, https://doi.org/10.1002/pssr.201700243
  18. A. I. Pointon, N. E. Grant, R. S. Bonilla, E. C. Wheeler-Jones, M. Walker, P. R. Wilshaw, C. E. J. Dancer, J. D. Murphy, 'Exceptional Surface Passivation Arising from Bis(trifluoromethanesulfonyl)-Based Solutions', ACS Appl. Electron. Mater. 1, 1322 (2019), DOI: 10.1021/acsaelm.9b00251, https://doi.org/10.1021/acsaelm.9b00251
  19. W. Ji, Y. Zhao, H. M. Fahad, J. Bullock, T. Allen, D.-H. Lien, S. De Wolf, A. Javey, 'Dip Coating Passivation of Crystalline Silicon by Lewis Acids', ACS Nano 13, 3723 (2019), DOI: 10.1021/acsnano.9b01038, https://doi.org/10.1021/acsnano.9b01038
  20. A. I. Pointon, N. E. Grant, S. L. Pain, J. T. White, J. D. Murphy, 'Sub-2 cm/s passivation of silicon surfaces by aprotic solutions', Appl. Phys. Lett. 116, 121601 (2020), DOI: 10.1063/5.0003704, https://doi.org/10.1063/5.0003704
  21. N. Araki, H. Udono, 'Effects of dopant type and concentration on surface recombination velocity in hydrogen-terminated silicon', Jpn. J. Appl. Phys. 61, 095504 (2022), DOI: 10.35848/1347-4065/ac829a, https://doi.org/10.35848/1347-4065/ac829a