2026 Volume 39 Issue 5
Article Contents

Jingyu Huang, Pengyun Yu, Donghai Li. Threshold Behavior of Quasi-Ballistic Transport in Ultrafast Depletion-Field Dynamics at GaAs Subsurface[J]. Chinese Journal of Chemical Physics, 2026, 39(5): 668-676. doi: 10.1063/1674-0068/cjcp2601010
Citation: Jingyu Huang, Pengyun Yu, Donghai Li. Threshold Behavior of Quasi-Ballistic Transport in Ultrafast Depletion-Field Dynamics at GaAs Subsurface[J]. Chinese Journal of Chemical Physics, 2026, 39(5): 668-676. doi: 10.1063/1674-0068/cjcp2601010

Threshold Behavior of Quasi-Ballistic Transport in Ultrafast Depletion-Field Dynamics at GaAs Subsurface

通讯作者: 陈斌, bchen63@163.com
  • 1. 

    沈阳化工大学材料科学与工程学院 沈阳 110142

  1. 本站搜索
  2. 百度学术搜索
  3. 万方数据库搜索
  4. CNKI搜索

Figures(5)

Article Metrics

Article views(27) PDF downloads(4) Cited by(0)

Access History

Threshold Behavior of Quasi-Ballistic Transport in Ultrafast Depletion-Field Dynamics at GaAs Subsurface

Abstract: Ultrafast dynamics of subsurface depletion fields play a central role in carrier separation and transport at semiconductor surfaces, yet the transport mechanisms governing their temporal evolution remain insufficiently constrained experimentally. Here, we identify a threshold behavior of quasi-ballistic transport in the ultrafast depletion-field dynamics at a GaAs subsurface, revealed by time-resolved electric-field-induced sum-frequency generation spectroscopy. By continuously tuning the pump wavelength, we observe a pronounced, step-like increase in the field recovery time once the excess energy of photoexcited carriers exceeds the Γ–L valley separation. This behavior defines a transport threshold at which intervalley scattering becomes competitive with, and ultimately faster than, field-driven carrier sweep-out across the depletion region, thereby giving rise to a scattering-limited space-charge reorganization regime. These results establish a direct and experimentally accessible criterion for identifying transport-regime transitions in ultrafast subsurface dynamics and highlight the capability of electric-field-sensitive nonlinear spectroscopy to resolve mechanism-level changes in carrier transport.

    • At a semiconductor surface, Fermi-level pinning by surface states induces pronounced band bending, forming a subsurface depletion region with a strong electric field oriented normal to the surface [1]. This depletion field is commonly treated as a static background potential that passively accelerates carriers away from the surface [2]. However, under ultrafast optical excitation, the depletion field directly couples to nonequilibrium carriers and can be dynamically reshaped by their motion and redistribution. As a result, the earliest stages of carrier transport beneath the surface are governed not only by the pre-existing band profile but also by a self-consistent, ultrafast feedback between carrier transport and depletion-field dynamics [3−5]. This dynamic tunability of subsurface electric fields provides a potential handle for controlling ultrafast photocurrent generation [6−8] and surface-sensitive electronic responses [9, 10].

      Achieving precise ultrafast control over subsurface depletion fields requires a quantitative understanding of how their temporal evolution is governed by the transport of photoexcited carriers. In most discussions, carrier motion within depletion regions is implicitly treated within a quasi-ballistic framework, where carriers traverse the built-in electric field with limited momentum randomization and the associated field screening and recovery are expected to evolve smoothly with carrier energy and excitation density [11, 12]. Within this picture, the characteristic recovery timescale is commonly assumed to reflect a continuous modulation of carrier sweep-out efficiency. Despite its central importance, however, this assumption has not been directly tested experimentally. Previous studies have largely inferred depletion-field dynamics indirectly from carrier lifetimes, photocurrent responses, or terahertz emission, where field evolution is intrinsically entangled with recombination, trapping, and population decay [13−15]. As a result, the experimental access to and mechanistic understanding of quasi-ballistic transport dynamics within semiconductor subsurface depletion layers remain limited.

      Time-resolved electric-field-induced sum-frequency generation (EF-SFG) spectroscopy provides direct sensitivity to the evolution of subsurface electric fields at semiconductor surfaces [16−18]. In GaAs, the measured SFG signal arises from the coherent superposition of the bulk nonlinear polarization and an additional electric-field-induced contribution associated with the depletion region beneath the surface [19]. Pump-induced variations in the depletion electric field therefore manifest as changes in the total SFG response, enabling the field dynamics to be tracked in real time. Crucially, because EF-SFG directly probes the electric field itself rather than carrier populations [16, 17], the extracted timescales reflect carrier transport and redistribution within the depletion region, largely disentangled from recombination or trapping processes. When implemented in a pump–probe geometry, EF-SFG thus offers a uniquely direct experimental window into ultrafast carrier transport beneath semiconductor surfaces on femtosecond to picosecond timescales.

      In this work, we use time-resolved EF-SFG spectroscopy to systematically resolve the ultrafast dynamics of the depletion-layer electric field at a GaAs subsurface. By tuning the pump wavelength and excitation density, we directly correlate the characteristic recovery timescale of the depletion field with the excess energy, momentum distribution, and density of photoexcited carriers. We observe a pronounced wavelength-dependent transition in the field recovery dynamics, revealing a departure from the simple quasi-ballistic picture and the emergence of scattering-limited regimes governed by intervalley processes. This behavior establishes a wavelength–timescale regime map for subsurface depletion-field dynamics and demonstrates that ultrafast electric-field control in semiconductors is fundamentally constrained by carrier transport pathways rather than carrier lifetimes. Our results provide a transport-based framework for understanding and manipulating ultrafast electric fields beneath semiconductor surfaces.

    II.   METHODS
    • Time-resolved EF-SFG measurements were performed using a femtosecond laser system operating at 50 kHz. The output was used to drive two optical parametric amplifiers that generated a tunable near-infrared (NIR) pulse and a broadband short-wave infrared (SWIR) pulse. The NIR beam was divided into a pump pulse and a probe pulse, while the SWIR pulse served as the second probe. The two probe pulses were spatially and temporally overlapped at the GaAs surface to generate the sum-frequency signal, whereas the pump pulse was incident on the same spot with a controllable time delay, as illustrated in FIG. 1. The pump–probe delay was adjusted using a motorized translation stage in the pump beam path, while the relative delay between the two probe pulses was fixed to ensure efficient SFG generation. The reflected SFG signal from the interface was spectrally resolved using a grating spectrometer. Wavelength-dependent measurements were conducted by tuning the NIR probe wavelength, with the time zero recalibrated after each adjustment. For power-dependent measurements, the pump fluence varied while keeping the wavelength fixed. All experiments were carried out under the PPP polarization configuration.

      In GaAs, the detected EF-SFG signal originates from the coherent superposition of a bulk second-order nonlinear polarization, governed by χ(2), and an electric-field-induced contribution proportional to χ(3)·E associated with the subsurface depletion field. Photoexcitation induces an ultrafast screening of the built-in electric field, thereby modifying the amplitude of the χ(3)·E term and its interference with the bulk χ(2) response. To isolate this pump-induced field modulation from static optical factors such as Fresnel coefficients, phase-matching conditions, and frequency-dependent nonlinear susceptibilities, the observable is plotted as the normalized EF-SFG intensity ratio, defined as the SFG intensity with pump excitation divided by that without pump excitation. Under these conditions, deviations of the normalized ratio from unity directly reflect changes in the depletion electric field (ΔE).

      The time-resolved EF-SFG signal was detected using a spectrometer within the phase-matching bandwidth and spectrally integrated over the sum-frequency window, rather than evaluated at a single SF frequency. Over the near-infrared spectral range and ultrafast timescales explored here, the relative phase between the bulk χ(2) and field-induced χ(3)·E contributions is assumed to remain constant, such that the normalized EF-SFG intensity ratio faithfully tracks the instantaneous magnitude of the depletion electric field. Consequently, the apparent field “recovery” time constant observed as a fast decay of the normalized ratio corresponds to the restoration of the depletion field following its initial ultrafast screening, rather than to carrier recombination or population decay. This normalization and interference framework ensures that the extracted recovery times directly reflect subsurface electric-field dynamics.

      The EF-SFG measurements were carried out on a commercially obtained n-type GaAs single crystal with a nominal donor concentration of ~1×1018 cm−3, for which Fermi-level pinning by native surface states produces a well-defined near-surface depletion region with a width of several tens of nanometers and a built-in electric field oriented normally to the surface (FIG. 2). This relatively high doping level ensures a strong depletion field while maintaining a short carrier transit time across the space-charge region, allowing the measured depletion-field recovery dynamics to be directly correlated with ultrafast carrier transport. The sample was oriented along the (001) crystallographic plane, where bulk GaAs (zinc-blende structure) possesses a single nonzero bulk second-order susceptibility, χ(2), such that the detected SFG signal arises from coherent interference between the bulk response and an electric-field–induced contribution associated with the subsurface depletion field, χ(3)·E, providing direct sensitivity to changes in the built-in electric field. Prior to optical measurements, the surface was cleaned by sequential acetone and isopropanol rinses followed by mild thermal annealing under ambient conditions, leading to stable native surface states and robust Fermi-level pinning.

    III.   RESULTS AND DISCUSSION
    • FIG. 3 presents the pump–probe dynamics measured at different pump wavelengths spanning from 700 to 880 nm. In GaAs, the observed SFG signal arises from the coherent superposition of the bulk second-order nonlinear response and an additional component induced by the depletion-layer electric field, with the two contributions interfering destructively [19]. Consequently, the pump-induced transient increase and subsequent decay of the signal reflect the ultrafast screening and recovery of the depletion electric field following optical excitation. At short wavelengths (≤ 800 nm), the transient response is dominated by a pronounced fast decay component occurring within the first few picoseconds, followed by a weak or nearly flat long-time tail. As the pump wavelength increases, the relative weight of this fast component gradually diminishes. Around 850 nm, the initial fast decay becomes barely discernible, and for wavelengths ≥ 850 nm the dynamics is instead governed by a slow recovery process. To resolve this slow process, we plot in FIG. 3(g–i) a long-range temporal evolution of the dynamics from 0 to 500 ps for excitation wavelengths of 820, 850, and 880 nm. Fitting the decay yields characteristic time constants of 1.1, 2.6, and 2.4 ns, respectively.

      Importantly, this slow component remains substantial even under excitation conditions where interband absorption and free-carrier generation are strongly suppressed, ruling out an origin based on the trapping of mobile carriers. The slow dynamics is instead attributed to direct optical modulation of surface or near-surface defect states located at or near the band edge [20]. At a fixed pump power of 8 mW, the slow decay time increases from ~1.10 ns at 820 nm to ~2.69 ns at 850 nm and ~2.41 ns at 880 nm, before decreasing again to ~0.99 ns at 920 nm. This non-monotonic wavelength dependence is consistent with a defect-mediated mechanism that is most effective for excitation near the GaAs band edge, where photocarriers are generated with low excess energy and longer residence times in the near-surface region, enhancing their interaction with surface and near-surface defect states. Photo-induced charging and passivation of these defects result in a persistent modulation of the surface band bending, giving rise to a long-lived contribution to the depletion-field recovery. For excitation further above the band edge or very close to the absorption edge, more efficient carrier sweep-out or reduced carrier generation suppresses defect involvement, leading to a shorter slow decay component. The excitation-density dependence further supports the assignment of photo-induced modulation of surface and near-surface defect states. For 820 nm excitation, increasing the pump power from 8 mW to 15 mW prolongs the slow decay time from ~1.10 ns to ~2.00 ns, indicating enhanced defect charging at higher carrier densities, whereas a further increase to 20 mW shortens the decay to ~1.60 ns, suggesting partial saturation of defect states and the onset of competing relaxation pathways.

      By contrast, the fast component observed at shorter wavelengths is a sub-picosecond to few-picosecond process that reflects the transport of mobile carriers out of the depletion region. For excitation well above the band edge, carriers are generated with substantial excess energy and are rapidly driven by the built-in electric field, enabling efficient diffusion and field-assisted sweep-out from the near-surface region [21, 22]. The corresponding recovery of the electric field is therefore limited by carrier transport.

      To visualize the wavelength dependence of the characteristic timescales and to disentangle the underlying transport mechanisms, we summarize the extracted dynamics in the wavelength–timescale diagram shown in FIG. 4. Three distinct regimes can be identified. At long wavelengths near and above the GaAs band edge (blue region), the dynamics exhibits a slow process associated with modulation of the depletion field by photo-passivation of defects [23, 24]. When the wavelength is below band edge (pink region), a fast component governs the recovery of the electric field, reflecting the transport of mobile carriers out of the depletion region. This fast-recovery area is further divided into two regimes as discussed in the following.

      In the wavelength range from 850 nm down to 730 nm, the fast recovery time constant $ {\tau }_{\text{fast}} $ decreases smoothly from approximately 1.6 ps to about 0.4 ps. This systematic reduction reflects a transport regime in which carrier motion remains quasi-ballistic and becomes progressively more efficient as the excitation photon energy increases. In this regime, $ {\tau }_{\text{fast}} $ is primarily governed by the transit time of carriers across the depletion region and is only weakly influenced by recombination or trapping processes.

      The continuous decrease of $ {\tau }_{\text{fast}} $ within this wavelength interval can be understood as a gradual enhancement of field-driven carrier transport. With increasing excess energy, photoexcited carriers are injected with higher initial velocities and undergo more efficient acceleration by the depletion electric field, leading to a reduced sweep-out time. Importantly, the smooth and monotonic nature of this trend indicates that the underlying transport mechanism remains unchanged throughout this range. The observed behavior therefore defines a quasi-ballistic sweep-out regime and establishes the minimal recovery time achievable for depletion-field dynamics under these excitation conditions.

      Characteristic timescales extracted from pump–probe EF-SFG measurements are plotted as a function of pump wavelength. Distinct dynamical regimes are identified, including a scattering-limited reorganization regime at short wavelengths, a quasi-ballistic sweep-out-limited regime at intermediate wavelengths, and a long-lived defect-mediated field modulation regime at longer wavelengths. The onset of Γ–L intervalley scattering and the GaAs band-edge region are indicated, highlighting the mechanism transition boundaries.

      By contrast, when the pump wavelength is tuned below 730 nm, $ {\tau }_{\text{fast}} $ exhibits an abrupt increase, signaling a mechanism transition of quasi-ballistic transport. In GaAs, the energy separation between the Γ and L valleys in the conduction band is approximately 0.29 eV [25]. Using a room-temperature direct bandgap of Eg≈1.42 eV for GaAs, the carrier excess energy is given by Eexcess=hνpump−Eg. For representative pump wavelengths used in this work, the corresponding excess energies are approximately 0.35 eV (700 nm), 0.28 eV (730 nm), 0.21 eV (760 nm), and ≤0.10 eV (820–880 nm). Notably, the excess energy at ~730 nm closely matches the Γ–L separation, providing a natural energetic boundary for the onset of efficient intervalley scattering. This quantitative mapping explains why the step-like increase in the fast recovery time emerges around the ~730 nm threshold in FIG. 4.

      Below this boundary, carriers remain predominantly in the Γ valley, whereas above it, energetically allowed Γ–L intervalley scattering rapidly randomizes carrier momentum and fundamentally alters the depletion-field recovery mechanism. At these higher excess energies, the Γ–L energy separation is exceeded and efficient intervalley scattering channels become energetically accessible [26−28]. Once the intervalley scattering time becomes comparable to or shorter than the field-driven sweep-out time, carrier momentum is rapidly randomized and the larger effective mass in the satellite valleys suppresses directed transport along the depletion field. As a result, carriers reside within the depletion region for a longer duration, sustaining electric-field screening and producing a pronounced increase in the recovery time constant. The sharp, threshold-like increase of $ {\tau }_{\text{fast}} $ therefore reflects the threshold behavior of quasi-ballistic transport rather than a gradual slowing of carrier motion.

      This interpretation is further supported by the distinct excitation-power dependence observed on either side of the threshold wavelength (shown in FIG. 5). Above the threshold, where intervalley scattering channels are already energetically accessible, increasing the pump power leads to a systematic reduction of the fast recovery time constant. By contrast, below the threshold wavelength, where quasi-ballistic transport remains operative, $ {\tau }_{\text{fast}} $ shows an opposite dependence on excitation power, consistent with a regime in which carrier transit is already efficient and only marginally influenced by density-dependent effects.

      Together, these contrasting power dependences reinforce the conclusion that the sharp change in $ {\tau }_{\text{fast}} $ reflects a transport-regime threshold rather than a gradual, density-controlled slowing of carrier motion. Below the threshold, carrier transport across the depletion region remains quasi-ballistic and is primarily governed by the field-driven transit time. Increasing excitation density slightly shortens the fast recovery time due to enhanced carrier–carrier interactions and weak carrier-induced screening that modestly reduce the effective depletion width and accelerate space-charge neutralization, without qualitatively altering the transport mechanism. Above the threshold, once intervalley scattering becomes efficient, the power dependence reverses: higher pump fluence leads to a shorter apparent recovery time. This indicates a transition from single-carrier, field-driven sweep-out to collective space-charge reorganization, where stronger carrier-induced screening and rapid energy relaxation reduce the effective screening length and facilitate faster macroscopic field recovery despite momentum randomization from intervalley transfer.

      We note that the power-dependent measurements are sampled at a limited number of excitation fluences. Accordingly, the purpose of these data is not to establish a detailed functional dependence of the recovery time on excitation density, but rather to qualitatively corroborate the existence of two distinct charge-transport regimes on either side of the threshold. The consistency of the observed trends across multiple pump wavelengths supports this interpretation, despite the restricted number of power points.

      Although, in the present experimental configuration, varying the pump wavelength indeed leads to a concomitant change in the near-infrared probe wavelength, which in principle could influence the EF-SFG response through the frequency dependence of the nonlinear susceptibilities χ(2) and χ(3), this wavelength dependence cannot account for the threshold-like behavior observed in our experiments for the following reasons. In the near-infrared spectral window explored here, the optical frequencies lie close to the GaAs direct band edge, where χ(2) and χ(3) are enhanced but vary smoothly with frequency due to the broad continuum of interband transitions rather than sharp excitonic resonances. As a result, the nonlinear susceptibilities exhibit gradual spectral dispersion without introducing abrupt changes in temporal response. Consequently, changing the probe wavelength primarily affects (i) the absolute EF-SFG signal amplitude and (ii) the relative weighting and interference contrast between the bulk χ(2) contribution and the field-induced χ(3)·E term. These effects rescale the signal magnitude but do not introduce additional temporal response functions.

      In time-resolved EF-SFG, the χ(3) tensor acts as a frequency-dependent yet temporally static transfer coefficient that linearly maps the instantaneous subsurface depletion field E(t) onto the detected nonlinear signal. Therefore, while probe-wavelength variation can modify the signal strength and contrast, it cannot create, suppress, or shift intrinsic recovery timescales, which are governed by the carrier-driven evolution of E(t). Importantly, the observed step-like change in the extracted recovery time occurs abruptly at a specific excitation-energy threshold, whereas the χ(2)/χ(3) frequency dependence is continuous across the same spectral range. This clear disparity rules out a probe-wavelength-dependent optical artifact as the origin of the threshold behavior.

      In addition, the optical penetration depth in GaAs varies with excitation wavelength, leading to wavelength-dependent spatial averaging of the subsurface depletion electric field probed by EF-SFG. At shorter wavelengths, stronger absorption confines photocarrier generation closer to the surface, whereas near the band edge the increased penetration depth distributes carriers over a larger depth range. However, for n-type GaAs with donor concentrations on the order of 1017–1018 cm−3, the depletion width is limited to several tens of nanometers due to the high carrier density and typical built-in potentials. In contrast, the optical penetration depth δ=1/α in the near-infrared (~700–900 nm), varies smoothly from several tens to a few hundred nanometers as the photon energy approaches the band edge. This hierarchy of length scales implies that the EF-SFG response effectively averages over the entire depletion field rather than being limited by optical attenuation. Importantly, because the penetration depth varies smoothly and monotonically with wavelength, it cannot account for the abrupt, threshold-like change in the fast recovery time observed near the critical excitation wavelength.

      Based on the nominal doping concentration of the GaAs substrate, we provide an order-of-magnitude estimate of the depletion width and the associated built-in electric field to connect the measured ultrafast field-recovery times to physically plausible carrier transport timescales. For an n-type GaAs sample with an electron concentration on the order of ND≈1×1018 cm−3, Fermi-level pinning at the surface leads to upward band bending and the formation of a depletion region beneath the surface. Using a typical surface band-bending potential of Vbi≈0.6–0.8 eV for oxidized GaAs surfaces, the depletion width can be estimated from the standard depletion approximation,

      which yields a depletion width on the order of W≈10–30 nm.

      The corresponding built-in electric field within the depletion region is therefore estimated to be E ≈Vbi/W ≈ (2–8)×107 V·m−1. Under such a strong internal field, photoexcited carriers experience rapid field-driven acceleration, and the characteristic transit time across the depletion region can be approximated as τtr≈W/vdrift, where the drift velocity approaches the high-field saturation velocity of GaAs (vsat ≈107 cm·s−1). This yields transit times on the order of sub-picosecond to a few picoseconds, in good agreement with the experimentally observed fast recovery time constants. These estimates support the interpretation that the fast component of the EF-SFG dynamics reflects carrier sweep-out and space-charge redistribution within the depletion region, rather than recombination- or trapping-limited processes.

      The identification of distinct transport regimes governing the ultrafast recovery of subsurface depletion fields has important implications for optical control of electric fields in semiconductors. Our results demonstrate that the response time of depletion-layer electric fields is not a continuously tunable parameter set solely by carrier lifetime or excitation density. Instead, it is fundamentally constrained by carrier transport pathways and their associated scattering thresholds. In particular, increasing carrier excess energy beyond a critical value does not accelerate field recovery but can instead induce a qualitative slowdown.

      These findings place intrinsic limits on strategies that seek to achieve ever-faster electric-field modulation by simply increasing excitation photon energy or fluence. The wavelength–timescale diagram established in this work provides a practical framework for identifying regimes in which ultrafast field control is feasible and for avoiding parameter spaces where transport-induced bottlenecks dominate the dynamics. More broadly, the present work highlights that subsurface electric-field dynamics encodes detailed information about nonequilibrium carrier transport, offering a route to interrogate transport regimes that are otherwise difficult to access by conventional electrical or optical probes.

      Finally, while the present study focuses on GaAs as a model semiconductor, the underlying physical picture—namely, the competition between field-driven transport and energy-dependent scattering—is expected to be broadly applicable to other semiconductor systems with surface band bending. Extending this approach to materials with different band structures, scattering landscapes, or surface-state densities may enable systematic exploration of ultrafast electric-field dynamics across a wide range of technologically relevant interfaces.

    IV.   CONCLUSION
    • In summary, we have uncovered a distinct transition in the ultrafast dynamics of the GaAs subsurface depletion electric field by directly resolving its time-domain evolution. Our results show that the recovery of the depletion field does not vary monotonically with excitation energy but instead exhibits a pronounced threshold-like change, reflecting a qualitative modification of carrier transport under nonequilibrium conditions.

      The observed increase in the recovery time constant at a well-defined excitation energy provides direct evidence that the efficiency of field-driven transport can be abruptly reduced when additional scattering channels become accessible. This behavior delineates two transport regimes—one characterized by efficient, quasi-ballistic sweep-out and the other dominated by scattering-induced momentum randomization—demonstrating that ultrafast depletion-field dynamics is governed by more than a simple continuous scaling with carrier excess energy or density.

      Beyond the specific case of GaAs, this work establishes EF-SFG as a powerful approach for probing ultrafast subsurface electric-field dynamics with femtosecond resolution. By accessing the electric field itself rather than carrier populations, this method offers a complementary perspective on nonequilibrium transport processes at semiconductor surfaces and interfaces. More broadly, our results highlight the importance of resolving electric-field dynamics to uncover transport regimes and transition behaviors that may remain hidden in conventional carrier-based measurements, providing a framework for future studies of ultrafast field control and transport in semiconductor systems.

    Figure (5)  Reference (28)

Catalog

    /

    DownLoad:  Full-Size Img  PowerPoint
    Return
    Return