Imaging mechanisms

Dynamical imaging of S-UEM to date has been achieved by obtaining transient contrast changes resulting from ultrafast photoexcitation (the photon‒electron probing regime) or electron impact (the electron‒photon probing regime),^{Reference Mohammed, Yang, Pal and Zewail22,Reference Sun, Adhikari, Shaheen, Yang and Mohammed29} with intensity referencing to unilluminated or nonabsorbing (thus nondynamical) regions or with image processing algorithms for background removal.^{Reference Najafi, Liao, Scarborough and Zewail30} A positively biased Everhart‒Thornley detector^{Reference Everhart and Thornley31} is often used to acquire secondary electrons (SEs) emitted mostly from the first few nanometers of the sample surface region. For a given dwell time at each pixel, a fixed number of pump-probe events with the same adjustable delay time (Δt) take place at each pixel, depending on the laser repetition rate used (Figure 1c). A scan image at that delay time is then constructed through rastering the pulsed electron beam and averaging multiple frames (Figure 2a); a series of scan images are acquired by varying Δt through changing the optical path of the laser beam with respect to the electron-generating beam using a computer-controlled delay stage. It should be noted that the present tens-of-micrometers laser footprint can be further reduced to a near-diffraction-limited size or a subdiffraction level with new optical designs.

Bright and dark image contrasts referenced to a nondynamical region may correlate with time-dependent energies (hence scattering cross sections) and populations of nonequilibrium charge carriers in materials,^{Reference Mohammed, Yang, Pal and Zewail22,Reference Sun, Melnikov, Khan and Mohammed23,Reference Shaheen, Sun, Yang and Mohammed26,Reference Sun, Adhikari, Shaheen, Yang and Mohammed29} as well as potential structural^{Reference Yang, Mohammed and Zewail27} or morphological^{Reference Najafi, Liao, Scarborough and Zewail30} changes caused by electronic excitation and relaxation. In a semiconductor, the presence of energetic electrons, promoted by photons or by the impact of PEs (1‒30 keV) to the conduction bands, may lead to a higher probability and yield of SE emission, causing bright contrast. Such processes can be regarded as energy-gain mechanisms (Figure 2a) and have been observed mostly in silicon- and cadmium-based semiconductors in the positive-time photon‒electron regime^{Reference Sun, Adhikari, Shaheen, Yang and Mohammed29} as well as in thin CdSe films in the negative-time electron‒photon regime.^{Reference Shaheen, Sun, Yang and Mohammed26} However, electrons carrying different energies in various bands may also exhibit sufficiently different scattering cross sections with primary and secondary electrons, which can result in a suppression, instead of an enhancement, of SEs escaping materials surfaces and hence a reduction in the SE yield. Consequently, a dark contrast is seen via such energy-loss scattering mechanisms in the presence of photocarriers, as observed in a number of materials, including GaAs^{Reference Cho, Hwang and Zewail32} and InGaN nanowires.^{Reference Sun, Adhikari, Shaheen, Yang and Mohammed29} Furthermore, displacement of energetic electrons and holes may lead to the observation of bright and dark contrasts in adjacent regions, signifying charge separations in different directions and speeds.^{Reference Najafi, Scarborough, Tang and Zewail33,Reference Liao, Najafi, Li, Minnich and Zewail34} Therefore, by using S-UEM, the rich carrier dynamics of photoactive materials may be imaged and visualized in real space and time.

It is worth mentioning that the unique opportunity in S-UEM to study electron-impact dynamics in the electron−photon dynamical probing regime at “negative” times, where the PE pulse arrives at the specimen earlier than the optical pulse and nonnegligible dynamics may still be found (Figure 2b). The electron impact of pulsed high-energy PEs in the specimen causes a cascade of scattering events and generates nonequilibrium electron‒hole pairs with relatively high energies on average. Time-dependent signals in SE emission may then be observed through clocking by late-arriving photon pulses.^{Reference Shaheen, Sun, Yang and Mohammed26} Such dynamical information cannot be accessed by time-resolved all-optical spectroscopy or by 4D imaging in the transmission mode, which may be crucial and relevant to the interpretation of results from electron-beam-induced current (EBIC) analysis used as an analytical method to evaluate the performance of optoelectronic materials and devices, because of the high initial energies of EBIC carriers.^{Reference Shaheen, Sun, Yang and Mohammed26}

Real-space and real-time imaging of carrier dynamics in solids and 2D materials

A recent S-UEM study on hydrogenated amorphous silicon (a-Si:H) showed surface imaging of photocarrier dynamics,^{Reference Liao, Najafi, Li, Minnich and Zewail34} which is technologically relevant to thin-film solar-cell applications and possesses drastically different carrier-transport properties from those of crystalline silicon. With the capability of high spatial and temporal resolutions of S-UEM, two intriguing phenomena were observed. First, qualitatively different from the ambipolar diffusion observed in crystalline silicon,^{Reference Najafi, Ivanov, Zewail and Bernardi35} energetic electrons and holes with different mobilities are spatially separated in a-Si:H after photoexcitation, as observed in the respective bright and dark contrasts as a result of a population displacement mechanism (Figure 3a), due to a large dielectric relaxation time caused by the low electrical conductivity.^{Reference Liao, Najafi, Li, Minnich and Zewail34} This “relaxation semiconductor” behavior was theoretically predicted in the 1970s,^{Reference Van Roosbroeck and Casey36} but could only be directly visualized for the first time via S-UEM. Second, it was observed that immediately after photoexcitation, hot electrons and holes with excess energy in a-Si:H diffuse away from the laser footprint center with significantly higher diffusivities than the near-equilibrium values, despite the strong structural disorder; a similar phenomenon of fast diffusion of hot carriers was also seen in crystalline silicon.^{Reference Najafi, Ivanov, Zewail and Bernardi35} Together with charge separations directly imaged across the interface of a p–n junction,^{Reference Najafi, Scarborough, Tang and Zewail33} these findings have provided new details of the spatial-temporal dynamics of hot electrons and holes in semiconductors immediately after they are photoexcited.

Figure 3.

Photocarrier dynamics in hydrogenated amorphous silicon and black phosphorus imaged by scanning ultrafast electron microscopy. (a) Time-resolved secondary electron images of the surface of hydrogenated amorphous silicon. The bright and dark contrasts indicate electron- and hole-rich regions. A clear spatial separation of electrons and holes is observed at an 827-ps delay time.^{Reference Liao, Najafi, Li, Minnich and Zewail34} (b) Variance of the spatial distribution of photoexcited hot holes on the surface of black phosphorus projected onto different directions and normalized to the initial distribution at 40 ps. It is clearly seen that the hot-carrier diffusion occurs preferentially along the armchair direction, labeled as “x.”^{Reference Liao, Zhao, Najafi, Yan, Tian, Tice, Minnich, Wang and Zewail37}

In addition, S-UEM has been applied to directly image the strongly anisotropic in-plane transport of hot carriers in black phosphorous, an emerging material for next-generation electronic and optoelectronic devices.^{Reference Liao, Zhao, Najafi, Yan, Tian, Tice, Minnich, Wang and Zewail37} Surprisingly, it was found that the anisotropy ratio of hole diffusivity along the “fast” (armchair) direction to that along the “slow” (zigzag) direction is about 15, much higher than the value (<3) observed under near-equilibrium conditions (Figure 3b). Such an observation demonstrates drastically different dynamics of hot carriers and near-equilibrium carriers, which is also consistent with the observed dynamics of energetic carriers generated by electron impact in the electron‒photon probing regime.^{Reference Shaheen, Sun, Yang and Mohammed26} In addition, high-energy electrons from a deeper region may experience scattering by lower-energy carriers generated through photoexcitation, causing a dark image contrast via the energy-loss mechanism. Thus, by using different parameters of external stimuli, carrier dynamics as a function of space, time, and excess energy may be tracked using S-UEM imaging.

Imaging of carrier dynamics of nanostructured materials

Visualization of charged-carrier dynamics in space and time has also been achieved on the surface of copper indium gallium selenide nanocrystal (CIGSe NC) and indium gallium nitride nanowire (InGaN NW) arrays before and after surface passivation and also after silicon doping.^{Reference Bose, Bera, Parida, Adhikari, Shaheen, Alarousu, Sun, Wu, Bakr and Mohammed38–Reference Bose, Adhikari, Burlakov, Liu, Haque, Priante, Hedhili, Wehbe, Zhao, Yang, Ng, Goriely, Bakr, Wu, Ooi and Mohammed41} Figure 4a–b shows time-resolved SE snapshots and intensity changes of InGaN NWs before and after passivation with an octadecylthiol (ODT) treatment, respectively, upon exposure to 515-nm photoexcitation. At a large negative delay time (‒400 ps), no change is seen in the image contrast, indicating full recovery of the InGaN NWs to the initial state before the next excitation pulse arrives at the specimen surface. A dark contrast observed in the laser-illuminated area both at negative tens of ps (electron-impact dynamics) and at positive times (photo-induced dynamics) signifies the energy-loss imaging mechanisms. Such observations imply that the effective scattering cross section of SEs with low-energy electrons in the conduction band is sufficiently higher than that of conduction electrons with PEs. At longer times, the intensity (negative change) of the recorded SE signal gradually recovers, which indicates surface charge-carrier recombination.

Figure 4.

Time-resolved secondary-electron (SE) images showing contrast evolution in InGaN nanowires (NWs) at an excitation of 515 nm at selected times for (a) an as-grown sample and (b) an octadecylthiol (ODT)-passivated one. The dashed ellipse indicates the laser spot on the NWs, which was approximately 40 µm in size. Directly comparing both samples indicates that the as-grown NWs exhibit more rapid recovery. (c) Extracted kinetics of the time-resolved SE intensities of the InGaN NWs from the as-grown and ODT-treated samples during a time window of 6000 ps. The faster recovery of the as-grown sample is evident. The structures of the two samples are shown in the inset micrographs.^{Reference Khan, Adhikari, Sun, Priante, Bose, Shaheen, Ng, Zhao, Bakr, Ooi and Mohammed40} (d‒f) Time-resolved SE difference images at indicated delay times for undoped and Si-doped NWs at 515-nm excitation. The dashed ellipse indicates the laser footprint on the sample. The scale bar is 50 µm. (g) Temporal evolution of the SE intensity at the center of the laser footprint on the 200-ps time scale. The inset shows the dynamics on a 6-ns time window. The reverse trends of time constants (e.g., faster intensity drop and slower recovery for the red versus slower intensity drop and faster recovery for the green) suggest different roles of silicon dopants on the photocarrier dynamics in different temporal ranges.^{Reference Bose, Adhikari, Burlakov, Liu, Haque, Priante, Hedhili, Wehbe, Zhao, Yang, Ng, Goriely, Bakr, Wu, Ooi and Mohammed41}

Interestingly, within a 5-ns time window under the same experimental conditions, the SE intensity change of the as-grown sample is significantly reduced compared to that of the ODT-passivated one (Figure 4c), which provides clear indication of a faster charged-carrier recombination for the as-grown sample. Thus, longer carrier lifetimes of InGaN NWs can be achieved by minimization of surface defects after the ODT treatment; this is also corroborated by the NW elemental mapping results of surface oxygen removal and formation of a carbon-rich amorphous layer.^{Reference Khan, Adhikari, Sun, Priante, Bose, Shaheen, Ng, Zhao, Bakr, Ooi and Mohammed40} Such performance improvement through surface passivation has also been examined in a recent study for CIGSe NCs with a large-gap zinc sulfide layer. Time-resolved SE images of specimen surfaces clearly demonstrate that the density of trap states is significantly reduced after ZnS capping The removal of trap states and increase in charge-carrier lifetimes were also experimentally confirmed by the increased photocurrent of self-biased photodetectors fabricated using the shelled NCs.^{Reference Bose, Bera, Parida, Adhikari, Shaheen, Alarousu, Sun, Wu, Bakr and Mohammed38}

Figure 4d‒f shows S-UEM images acquired on the surface of InGaN NWs with and without silicon dopants of varying concentration processed at different temperatures.^{Reference Bose, Adhikari, Burlakov, Liu, Haque, Priante, Hedhili, Wehbe, Zhao, Yang, Ng, Goriely, Bakr, Wu, Ooi and Mohammed41} A difference in the time-dependent dark contrast in the laser-illuminated region is evident. Within a time window of 6 ns, a greater silicon dopant concentration is shown to cause a faster recovery of the dark image contrast, which is indicative of a higher rate of charged-carrier recombination (Figure 4g, inset). However, the SE kinetics at hundreds of picoseconds exhibit a surprising opposite trend to that observed on the longer time scale, hinting at the possibility of a different mechanism operative within this time window (Figure 4g). The increase of silicon dopants appears to add trap states that effectively induce a lag for the early-time escape of the emitted SEs, thus resulting in deferred growth of the dark contrast in the case of Si-doped NW samples. Hence, the increase in silicon dopant concentration in InGaN NW arrays manifests itself in the trapping dynamics at early times for SEs and at longer times for photocarriers.

Concluding remarks and outlook

The selective study of ultrafast carrier dynamics of materials, surfaces, and interfaces with high spatial and temporal control is one of the major challenges impeding the optimization of photoactive materials-based devices. Four-dimensional S-UEM provides a versatile tool to tackle the problem. With this methodology, one can now have access to dynamical phenomena at surfaces with adequate spatiotemporal resolution, including surface diffusion, charge separation, transport, recombination, and anisotropic motions, a territory that has been beyond reach by static electron imaging or conventional time-resolved laser spectroscopy until now. A few studies such as hydrogenated amorphous silicon, black phosphorus, indium gallium nitride nanowires, and copper indium gallium selenide nanocrystals were presented in this article.

The state of the art of 4D S-UEM can be further advanced by new optical designs to reach a diffraction-limited or subdiffraction laser footprint, the use of nanostructured specimens and devices with patterns of photoactive and inactive materials for nanoscopy-level imaging, implementation of a high-brightness source using a cold nano-sized emitter in the field-emission regime, integration of a lock-in amplification mechanism for high-sensitivity detection,^{Reference Zani, Sala, Irde, Pietralunga, Manzoni, Cerullo, Lanzani and Tagliaferri42} and incorporation of various imaging modes in conventional scanning electron microscopy with electrons, photons, or induced currents. We anticipate 4D S-UEM to become more broadly adopted in the near future for direct visualization of carrier and surface dynamics in space and time, taking advantage of the versatile use of scanning electron microscopy in various fields of scientific and technological research.