Ultrafast generation and decay of a surface metal

The study investigates whether metallic conductivity can be induced and controlled on ultrafast timescales at a semiconductor surface, specifically ZnO(10-10). The motivation stems from the ability of band bending and doping at semiconductor surfaces to create metallic states with unique properties (high mobility, magnetism, superconductivity) not present in the bulk, and the appeal of optically generating such states for high-speed electronics and optoelectronics. The authors hypothesize that photoexcitation can photodope the ZnO surface by depopulating deep donor-type defects, leading to positive surface charging, downward surface band bending, and a transient metallic band at the surface. This would mimic the effect of chemical doping or a gate in a field-effect device and potentially enable phototransistor-like control of ultrafast currents, leveraging ZnO’s wide band gap, intrinsic n-doping, ease of nanostructuring, and visible transparency.

Prior work shows persistent photoconductivity in semiconductors due to metastable photodoping via long-lived defect excitons. ZnO is a wide-band-gap (3.4 eV), intrinsically n-doped semiconductor with numerous native deep donor defects and strong defect-related photoluminescence. Ultrafast semiconductor-to-metal transitions (SMTs) have been observed previously in bulk inorganic semiconductors but at much higher photon fluxes. Theoretical frameworks (Mott-Hubbard) predict critical behavior of effective mass near SMTs. Alternatives like band-gap renormalization (BGR) due to carrier-carrier screening can shift bands but would symmetrically shift CB down and VB up, unlike surface photodoping-induced band bending which shifts both bands downward near the surface. Surface photovoltage (SPV) effects are known to track photoinduced changes in surface charge on ultrafast timescales.

• Technique: Time- and angle-resolved photoelectron spectroscopy (trARPES) in a pump-probe scheme on a ZnO(10-10) single-crystal surface prepared by annealing in ultrahigh vacuum.
• Probes and pumps: For states near EF, probe photon energy hνprobe = 6.3 eV; for valence band (VB) tracking, hνprobe = 4.25 eV (two-photon photoemission, as work function ≈ 4.4 eV). Pump photon energy primarily 3.43 eV (near band gap), with additional tests at 3.2 eV and 3.0 eV to address in-gap states below the fundamental gap.
• Repetition rate variation: Laser repetition rate tuned from 200 kHz (5 µs separation) down to 5 kHz (200 µs separation) to test for long-lived photostationary defect excitons contributing to steady-state shallow-donor features.
• Spectral analyses: Angle-resolved maps to determine band dispersion and curvature; fits of dispersive features with a Gaussian peak multiplied by a Fermi–Dirac distribution to extract the dispersion and effective mass meff. Angle-integrated spectra around EF fitted with Fermi–Dirac distributions to extract electronic temperatures and thermalization dynamics.
• Dynamics and fitting: Time-dependent photoelectron intensity integrated below EF to quantify surface electron density; simultaneous tracking of band curvature (1/meff) and VB peak shifts. Global fits yielded characteristic times T1, T2, T3 for rise and decay dynamics. Instrument response considered.
• Fluence series: Pump fluence scanned from a few to tens of µJ/cm² to identify a threshold fluence Fc for metallization and to study critical behavior. meff vs fluence fitted with meff^-1 = A(F−Fc)^α + m0 to test Mott-like criticality.
• SPV diagnostics: Observation of negative-delay energy shifts in probe spectra to confirm transient positive surface charging (electrostatic field in near-surface vacuum).
• Sample conditioning: Comparative measurements after different annealing temperatures (750 K vs 950 K) to assess defect creation and its impact on the photoinduced metal phase.
• Controls for mechanism: VB energy shift direction and magnitude vs delay and pump photon energy used to distinguish surface photodoping-induced band bending from bulk BGR. Pump at 3.2 eV vs 3.0 eV tested defect-level energetics (onset of photodepletion).

• Ultrafast surface-confined SMT: Upon photoexcitation, a dispersive band crossing the equilibrium EF appears near the ZnO(10-10) surface, exhibiting metallic characteristics: density of states at EF, a partially filled dispersive band, and a hot electron distribution described by a Fermi–Dirac function that equilibrates with the lattice within ~200 fs.
• Effective mass: Early-time dispersion fits yield meff ≈ 1.2(1) me at low positive delays; at highest fluence, meff ≈ 0.7 me. Extrapolation with critical fit yields m0 ≈ 0.2(6) me, consistent with the bulk CB effective mass (~0.25 me).
• Threshold fluence and criticality: Metallization occurs above a low threshold fluence Fc = 13.6 µJ/cm². The curvature 1/meff follows a critical form meff^-1 = A(F−Fc)^α + m0 expected for a Mott transition, indicating that photoexcitation acts like chemical doping on femtosecond timescales.
• Dynamics: The electron density in the metallic band (integrated PE intensity below EF), the band curvature (1/meff), and the VB peak position exhibit synchronized dynamics with characteristic times: • T1 = 20(20) fs (abrupt rise within resolution, electronic screening timescale), • T2 = 1.20(15) ps (further increase attributed to delayed hole trapping at surface defects), • T3 = 219(13) ps (decay due to electron–hole recombination and relocalization/localization increase).
• Electronic temperature: Hot electron temperatures start around ~1300 K and cool to the lattice within ~200 fs, similar to metals.
• Mechanism: VB shifts downward abruptly and for hundreds of ps upon pumping, evidencing downward surface band bending from positive surface charging due to photodepletion of deep donor defects. Negative-delay probe shifts corroborate transient positive surface charge (SPV effect). This behavior rules out BGR as the dominant cause (which would shift VB upward) and supports surface photodoping as the driving mechanism.
• Defect energetics: Pump at 3.2 eV still induces downward band bending; 3.0 eV does not. Thus, the relevant occupied deep donor levels lie within ≲0.4 eV above the VB maximum.
• Defect origin: Stronger metallization signal after 950 K annealing compared to 750 K (factor 2–3 increase) indicates defect creation during high-temperature annealing; literature and energetics implicate oxygen vacancies (neutral in equilibrium; predicted level ~0.4 eV above VB).
• Photostationary donors: A dispersionless feature ~0.1 eV below EF decreases when reducing repetition rate (200 kHz → 5 kHz), indicating long-lived defect excitons forming a photostationary shallow-donor population.
• Temperature window: The PIPT operates from 100 K to at least 256 K (near room temperature).
• Comparison to other SMTs: The decay back to semiconducting state is sub-nanosecond, faster than some known photoinduced SMTs (e.g., VO2).

The findings demonstrate that optical excitation can transiently dope the ZnO surface by depopulating deep donor defects, creating positive surface charge and inducing downward band bending that pulls the conduction band below the equilibrium EF at the surface. Above a critical carrier density (set by pump fluence), a Mott transition occurs, delocalizing carriers into a surface-confined metallic band exhibiting free-electron-like dispersion and hot-electron thermalization dynamics typical of metals. The synchronized evolution of metallic carrier density, band curvature, and VB downward shift across femtosecond to hundreds-of-picoseconds timescales directly ties the SMT to the photodoping-induced band bending mechanism, rather than bulk band-structure renormalization. This mechanism effectively mimics chemical doping on ultrafast timescales and should be broadly applicable to other semiconductors with sufficient densities of donor-type deep surface defects. The ability to generate and quench a surface metal with low fluence, on sub-ps to sub-ns timescales, and near room temperature highlights potential for ultrafast, transparent photoswitches and for studying transient 2DEGs with emergent functionalities.

The work establishes a simple, universal mechanism for ultrafast, low-fluence semiconductor-to-metal transitions confined to a surface: photodepletion of deep donor defects causes positive surface charging and downward band bending, yielding a transient metallic band below EF. The transition exhibits Mott-like critical behavior in effective mass versus fluence, ultrafast build-up (tens of fs), enhancement via hole trapping (~1.2 ps), and sub-ns decay (~219 ps). Defect energetics and annealing dependence point to oxygen vacancies as the deep donors. The phenomenon persists up to at least 256 K, suggesting practical relevance. Future research directions include: quantitative control of defect densities and band bending; extending the approach to other semiconductors and heterostructures; exploring emergent properties (e.g., magnetism, superconductivity) in transient 2DEGs; and integrating such photodoped surface metals into ultrafast optoelectronic devices.

• Exact publication date and some experimental specifics (e.g., absolute critical exponent value) are not specified in the provided text.
• The magnitude of maximum band bending at the topmost surface cannot be quantified due to depth averaging in photoemission; the VB shift is not entirely rigid and appears broadened.
• The chemical identification of defects is inferred (oxygen vacancies) from energetics and annealing dependence rather than directly measured spectroscopically here.
• The temperature range tested is up to at least 256 K, not explicitly at room temperature; broader environmental stability and device-level implications are not explored.
• The deep defect peak is not directly resolved in spectra (likely due to proximity to high DOS near the VB edge).