† Corresponding author. E-mail:
Recent advances in the ultrafast transmission electron microscope (UTEM), with combined spatial and temporal resolutions, have made it possible to directly visualize the atomic, electronic, and magnetic structural dynamics of materials. In this review, we highlight the recent progress of UTEM techniques and their applications to a variety of material systems. It is emphasized that numerous significant ultrafast dynamic issues in material science can be solved by the integration of the pump–probe approach with the well-developed conventional transmission electron microscopy (TEM) techniques. For instance, UTEM diffraction experiments can be performed to investigate photoinduced atomic-scale dynamics, including the chemical reactions, non-equilibrium phase transition/melting, and lattice phonon coupling. UTEM imaging methods are invaluable for studying, in real space, the elementary processes of structural and morphological changes, as well as magnetic-domain evolution in the Lorentz TEM mode, at a high magnification. UTEM electron energy-loss spectroscopic techniques allow the examination of the ultrafast valence states and electronic structure dynamics, while photoinduced near-field electron microscopy extends the capability of the UTEM to the regime of electromagnetic-field imaging with a high real space resolution.
Pump–probe-based time-resolved experiments attract considerable attention owing to their capability of providing time-domain information regarding the characteristics/strengths of the couplings between structural, electronic, and spin degrees of freedom by following the dynamic response of a system to pulsed excitation.[1,2] By employing diverse probe pulses, the evolution of different microscopic or macroscopic properties of the excited systems can be measured. Pump–probe-based time-resolved spectroscopic experiments using visible, infrared, and ultraviolet (UV) light can be used to map the evolution of electronic states up to a time resolution of several femtoseconds.[3,4] Ultrafast x-ray techniques, particularly the x-ray free-electron laser[1,2,5,6] and electron diffraction,[1,7] are fast-growing and can be used to track the lattice dynamics with femtosecond temporal resolution. UTEM, which combines femtosecond temporal resolution with the excellent imaging, diffraction, and spectroscopy capabilities of conventional transmission electron microscopy (TEM) owing to the well-developed state-of-the-art electron optics, promises to become one of the most powerful experimental tools for the investigation of ultrafast dynamics on the nanoscale.[7] Inspired by pioneering works at the Technical University Berlin,[8] California Institute of Technology (Caltech),[9] and Lawrence Livermore National Labs (LLNL),[10] time-resolved electron microscopy, both the stroboscopic[7] and single-shot methods,[10] has been currently extensively explored in a growing number of laboratories worldwide. We successfully developed a UTEM at the Institute of Physics, Chinese Academy of Sciences (IOP, CAS) in 2014.[11]
Over the past few years, the UTEM has been developed through the integration of different microscopic techniques, including imaging modes (bright-field (BF) and dark-field (DF) imaging, high-resolution TEM (HRTEM), scanning TEM (STEM), Lorentz TEM (LTEM), energy-filtered TEM (EFTEM), and tomography) and diffraction modes (selected-area electron diffraction with parallel electron beams, convergent-beam electron diffraction (CBED), electron energy-loss spectroscopy (EELS), and photoinduced near-field electron microscopy (PINEM)), which has introduced significant opportunities to solve fundamental problems in physics, materials science, chemistry, and biology. UTEMs have been used to study elementary structural changes during photoinduced chemical reactions, non-equilibrium structural phase transitions/melting, the crystallization of solids, and nanoscale elastic deformations with a combination of Ångstrom-scale spatial resolution and femtosecond/picosecond time resolution. These results provide unprecedented insight into the non-equilibrium transformation pathways. The development and applications of the UTEM in materials science and biology have been extensively reviewed.[1,7,9,10,12–31]
Herein, we first provide a conceptual design of the UTEM and illustrate the detailed design of the UTEM of our research group at the Institute of Physics. Then, a brief overview of the literature is presented, with a focus on the results obtained by our group. Specific cases are discussed to demonstrate the utility of the different UTEM techniques in materials science with unprecedented real-time and real-space capabilities.
The UTEM is reformed on the basis of a commercial transmission electron microscope, and the time-resolution function is realized by introducing an ultrafast laser via the pump–probe technique.[7,10,30] As shown in Fig.
In view of the different needs of the experiment, there are two experimental models: the stroboscopic mode and the single-shot mode.[7,10,30] As shown in Fig.
After the pioneering work of Zewail of Caltech and the research work of LLNL, there are some groups in the worldwide established thermal-emission TEM-based UTEM setup, such as Carbone et al. at EPFL,[32,33] Flannigan et al. in Minnesota,[31] Banhart et al. in France,[34] Kwon et al. at UNIST of Korea,[35] and other research groups at the KTH Royal Institute of Technology, Sweden,[36] and Russia.[37] These UTEMs are based on a thermionic gun (LaB6 cathode) with a Wehnelt electrode and can be operated in the nanosecond, single-shot, femtosecond, and single/few-electron modes. The Wehnelt bias and the cathode–Wehnelt distance significantly influence the temporal and energy spreads of the picosecond electron pulses, which are determined by the space-charge and Boersch effects, according to the number of electrons in a pulse.[36,38–41]
To obtain the dynamic structural features with high time and spatial resolutions, we developed a UTEM at the Institute of Physics, Chinese Academy of Sciences (IOP, CAS) in 2014,[11] as shown in Fig.
Before performing UTEM pump–probe measurements, we adjust our laser/TEM system and calibrate the synchronization between the pumping femtosecond laser and the electron-pulse probe with sub-picosecond precision. The electron-shadow image of surface plasma is used to determine the time zero. The Cu TEM grid is excited by the pump laser at the center to generate a plasma cloud, which pushes the probe electron pulses away from the Cu grid, as shown in Fig.
Temperature accumulation is a common phenomenon in high-repetition rate pump–probe experiments owing to the laser-induced heating effect. For the stroboscopic mode, it is important to check the signal difference between the laser-on (negative delay) and laser-off states to determine whether there is obvious temperature accumulation. The time constant of the recovery process can be determined by reducing the repetition rate in the pump–probe experiments, by replacing the probe laser with a nanosecond UV laser, and using a digital delay generator to tune the time delay between the pump laser and the probe electron beam at the timescale of nanoseconds to micro- or milliseconds. Usually, the time constant of the thermal-dominated recovery process can vary from microseconds to milliseconds depending on the condition of thermal dissipation. To assist the recovery process and avoid heat accumulation, a supporting grid with a large mesh number and high thermal conductivity (e.g., coated with C nanotubes), along with a pump laser having a small beam size, is often used.
UTEM techniques have been successfully used to study the ultrafast evolutions of atomic and electronic rearrangements under non-equilibrium conditions in organic, inorganic, and biological materials.[7,10,19,22,27,29] In this section, we provide a short review of the application of the UTEM in materials science, separately addressing the information obtained in the reciprocal (diffraction), real (imaging), and energy (spectroscopy) spaces.
Diffraction techniques (neutrons, x-rays, and electrons) are classic methods for determining the atomic structure. For the concept of pump–probe experiments, although pulsed x-ray sources with high brightness are realized in large-scale facilities, such as free-electron lasers or synchrotrons, laboratory-scale x-ray experiments suffer from comparably low photon numbers.[1,7] Owing to the large scattering cross section of electrons, it is possible to perform time-resolved electron-diffraction experiments using very compact laboratory-scale setups.[1] Via TEM diffraction, Bragg spots, Debye–Scherrer rings, and CBED diffraction patterns can be obtained for different types of samples with parallel or convergent electron beams. By analyzing the position, intensity, or width of the spots/ring/lines, information regarding the structural dynamics can be obtained. For example, thermal heating, no thermal effects, phase transitions, coherent lattice vibrations, and lattice phonon coupling can be analyzed according to the intensity/scattering vector change of particular diffraction peaks; the variation of the scattering vector for the observed Bragg reflections can be used for the study of the nanoscale energy transport, electron–phonon coupling, and lattice relaxation at the picosecond timescale.[7,10,31] When convergent-beam illumination is adopted, the transverse (shear) dynamics of certain lattice planes can be directly probed.[42] Spatially focused and timed electron packets of UTEM CBED allow the determination of structures in three dimensions with highly precise localization, showing potential for investigating dynamic processes of single particles and structures of heterogeneous media.
Here, we highlight two of our recent works to demonstrate the capability of the ultrafast electron diffraction (UED) mode of the UTEM.
High spatial resolution of the UTEM is ideal for dynamical studies of nanostructured materials with an inherently small volume or mass. Previous results demonstrate that UTEM can be used for the study of nanoscale energy transport, phase transitions, chemical reaction, electron–phonon coupling, and lattice relaxation at the picosecond timescale.[7,10,11,31,43–45]
Our previous work on the ultrafast structural dynamics of multi-walled C nanotubes clearly revealed the strong anisotropic nature of the lattice dynamics, showing that distinguishable lattice relaxations appear in intra-tubular sheets at an ultrafast timescale of a few picoseconds, followed by lattice expansion along the radial direction.[11] This anisotropic feature is fundamentally correlated with the specific tubular structure and chemical bonding behaviors, i.e., the strong covalent bonds in the tubular sheet and the weak van der Waals inter-layer bonds along the radial direction.
We recently investigated the ultrafast structural dynamics of multi-walled boron nitride nanotubes (BNNTs) under femtosecond optical excitation using the UED mode of the UTEM.[46] Analysis of the time-resolved (100) and (002) diffraction profiles reveals the highly anisotropic lattice dynamics of the BNNTs, which are attributed to the distinct nature of the chemical bonds in the tubular structure. Moreover, the changes in the (002) diffraction positions and intensities suggest that the lattice response of BNNTs to the femtosecond laser excitation involves fast and slow dynamic lattice processes. The fast process, with a time constant of approximately 8 ps, can be understood as a result of electron–phonon coupling, while the slow process, with a time constant of approximately 100 to 300 ps depending on the pumplaser fluence, is ascribed to an Auger recombination effect. In Fig.
Recently, studies employing ultrafast spectroscopy and diffraction (UED, ultrafast x-ray diffraction) have revealed a variety of notable dynamic features in charge density wave (CDW) compounds, including coherently driven collective modes and photoinduced phase transitions. In transition-metal dichalcogenides, experimental observations reveal that photoinduced dynamics often involve multiple transient steps at different timescales.[47,48] The periodic lattice generally undergoes an initial fast decay in a few hundred femtoseconds after optically induced redistribution of the electron density, and then a rapid electron–phonon energy transfer occurs, followed by the recovery of the CDW when the excess energy is redistributed by thermalization.
We recently investigated the optically induced CDW transition of 1T-TaSe2 from the commensurate phase (C phase, qC = 0.225a* + 0.07b*) to the incommensurate phase (IC phase, qIC = 0.278a*) with a visible angular rotation of the CDW wave vector via UTEM investigation, as shown in Figs.
In ultrafast-diffraction studies, the diffraction-intensity change of the specific Bragg spot often exhibits a notable feature arising from the Debye–Waller effect and coherent oscillations.[7,49] It is believed that a certain anomalous intensity change originates from the zone-axis tilting caused by surface deformation or anisotropic lattice expansion. The dynamic phenomena in metals (Al, Au, and Ag), semimetals (Bi), and semiconductor films (Si) have been extensively reported. Moreover, layered-structure materials often have a far larger thermal expansion coefficient of the c-axis than of the a–b plane, and this anisotropic feature commonly yields a large amplitude-breathing standing wave and anisotropic periodic oscillation in the lattice parameters.[7] We recently reported a series of data on the ultrafast dynamic evolutions observed in 1T-TaSeTe, including the suppression of the satellite spots corresponding to the CDW modulation, the related Debye–Waller effect associated with lattice thermalization, and oscillatory evolutions of the diffraction intensity.[49] According to thin-crystal dynamic diffraction theory, the essential nature of temporal diffraction-intensity alterations is well-explained by anisotropic lattice expansion combined with the Debye–Waller effect.
In electron diffraction, a high time resolution, up to femtoseconds, is possible, but the data are recorded in the reciprocal space. Ultrafast imaging in the UTEM provides opportunities for studying the elementary processes of structural and morphological changes in real space. Benefiting from state-of-the-art electron optics and the combination of well-developed conventional TEM techniques (BF, DF, HRTEM, LTEM, EFTEM), the UTEM provides the capability of investigating the atomic motion, valence states, magnetic domain, structural domain, structural defects, and strain dynamics in real space with high time and spatial resolutions.[7,26,31,50–59]
The pioneering work regarding the application of ultrafast imaging techniques of the UTEM was performed at Caltech. Zewail et al. investigated the structural and morphological changes in single-crystal Au and graphite films using a second-generation ultrafast electron microscope (UEM-2).[13] A variety of dynamic responses to laser pulse excitation were observed, at timescales ranging from picoseconds to microseconds.
Recently, Flannigan et al. reported the direct, real-space imaging of the emergence and evolution of acoustic phonons at individual defects in crystalline WSe2 and Ge via BF imaging with a UTEM. The sub-picosecond nucleation, the launch of wavefronts at step edges, and the resolved dispersion behaviors during propagation and scattering were imaged.[51,52,55] These observations provide unprecedented insight into the roles of the individual atomic and nanoscale features on the acoustic–phonon dynamics and introduce a new route for the ultraprecise manipulation and control of coherent energy propagation at the atomic scale.
Here, we highlight our recent work to demonstrate the capability of the imaging mode of the UTEM.
We studied the shape-memory alloy Mn50Ni40Sn10 using our home-built UTEM modified from a JEOL-2000EX microscope and reported a picosecond view of a martensitic (MT) transition and nucleation.[59] The experimental results demonstrate that the MT transition and reverse transition in this Heusler alloy contain a variety of structural dynamic features at the picosecond timescale. Imaging and diffraction observations clearly show that the MT transition and MT domain nucleation, which are related to cooperative atomic motions, occur between 10 and 20 ps. Figures
LTEM is a powerful technique for performing quantitative analysis of magnetic-domain structures at the sub-50-nm length scale. In a transmission electron microscope, relativistic electrons traversing a thin magnetic sample under an electrostatic field and a magnetic field experience the Lorentz force generated by the sample. If the Lorentz forces acts normal to the travel direction of the electrons, a deflection occurs. Only the in-plane magnetic induction contributes to the deflection. The deflection angle is proportional to the in-plane magnetization and the thickness of the sample. There are two modes in Lorentz microscopy: the Fresnel mode, in which domain walls and magnetization ripples are observed, and the Foucault mode, in which domains are imaged. The LTEM technique evolved from a qualitative magnetic-domain observation technique into a quantitative technique for the determination of the magnetization state of a sample.[60] It has been used to image and analyze complex magnetic textures such as magnetic skyrmions[61] with 2-nm spatial resolution. LTEM has also been applied very successfully to the visualization of flux vortex lattices in Type II superconductors.[62] Recently, by combining ultrafast laser techniques with LTEM, UTEM LTEM has been successfully applied to study ultrafast magnetization dynamics with a high temporal resolution ranging from nanoseconds to picoseconds.[57,58] Zewail et al. reported the imaging of the magnetic domain-wall dynamics in a partially oxidized Fe film via the out-of-focus Fresnel method with a nanosecond time resolution.[57] The success of this study demonstrates the promise of the Lorentz UTEM for the real-space imaging of magnetization switching, ferromagnetic resonance, and laser-induced demagnetization in ferromagnetic nanostructures.
Flannigan et al. employed a UTEM for picosecond Fresnel imaging of the photoinduced magnetic-domain wall dynamics in FePt thin films.[58] Under operation with a low instrument repetition rate (5 kHz) and without objective-lens excitation, the picosecond demagnetization of an FePt film was directly imaged via in situ femtosecond laser excitation. The dynamics were quantified and monitored according to the time-dependent change in the degree of electron coherence within the magnetic domain walls. Figure
EELS is a well-established technique for retrieving fundamental information regarding the chemical bonding, electronic structure, and charge distribution at length scales as small as a few nanometers. The low-loss (0–50 eV) region of the spectrum contains electronic information in the form of valence intra- and interband transitions and plasmon excitations. The core-loss (>100 eV) region of the spectrum contains information regarding the chemical state, local geometric structure, and the nature of chemical bonding centered around the absorbing atom. Particular energy ranges of interest can be chosen by using slits to form energy-filtered images, which is known as EFTEM. The typical energy resolution for TEMs with thermionic field-emission sources is approximately 0.8 eV, and the energy resolution of cold field-emission guns is a few hundred millielectronvolts. A monochromated field-emission electron source can increase the energy resolution to approximately 100 meV by reducing the effects of partial temporal coherence, while the state-of-the-art unique design of the monochromator results in an ultrahigh energy resolution of 10 meV or better (7 meV Nion HERMES 200), which is comparable to phonon energies and other low-energy excitations.[63]
Several UTEM EELS studies have been conducted. Ultrafast core-loss spectroscopy has been demonstrated to probe the electronic and structural dynamics after laser excitation of a graphite nanometer-thin film and C nanotubes, to investigate the coupling between the electronic and lattice motions.[32,64–66] The ability to record at a high resolution (in energy, space, and time) makes it possible to obtain both in situ nanoscale elemental and oxidation-state mappings via element-specific imaging.[33,67,68] The value of this methodology is demonstrated by studying the photocatalyst hematite, i.e., α-Fe2O3. By employing EFTEM using UTEM equipment, it has been determined that Fe4+ ions have a lifetime of a few picoseconds, and the associated photoinduced electronic transitions and charge-transfer processes have been observed.[68]
Here, we demonstrate the capability of UTEM EELS by considering the example of the photoinduced electronic structural dynamics in layered manganite.
Bi-layered PrSr0.2Ca1.8Mn2O7 has a peculiar checkerboard pattern made of ordered orbital stripes originating from the 1:1 ratio of Mn3+and Mn4+ ions in the ground state. Carbone et al. demonstrated coherent structural motions of a PrSr0.2Ca1.8Mn2O7 single crystal induced by the photoinduced temperature jump, which were detected as periodic modulations of the Bragg peak intensities.[33] Using the same instrument, the resultant modulation of the electron energy-loss function was recorded over a broad energy range (2 to 70 eV), indicating the response of the different electronic states to specific atomic motions.[33] Such an interplay was understood via density functional theory (DFT) calculations of the EEL spectra, in which equilibrated electronic structures for different lattice parameters were obtained, as shown in Fig.
PINEM was first reported by García De Abajo et al. as “electron energy-gain spectroscopy”,[69] and was realized by Zewail et al. in 2009 using a UTEM.[17] Having a femtosecond time resolution and nanometer spatial resolution, PINEM mainly exploits the absorption and emission of integer photon quanta during the scattering of swift electrons by the field near the boundary or interface of the sample. As an example, figures
A high spatial resolution up to 6 nm in PINEM was achieved by Aycan Yurtsever et al. in their experimental study involving Ag nanoparticles.[70] In Figs.
PINEM extends the capabilities of the UTEM to the regime of EM-field imaging, and retains the resolution advantage of the UTEM. It is a technology with wide applications in the imaging of photoinduced processes in complex materials, nanostructures, and macromolecules, providing both a direct physical picture of the EM-field distribution and energetic degree-of-freedom information via EELS.[70–77]
Considered one of the most exciting frontiers in electron microscopy, the field of ultrafast TEM is presently at a pivotal moment in its development. From a technical viewpoint, the greatest challenges that the UTEM faces are to obtain intense high-quality electron pulses and to increase the spatiotemporal resolution. Solutions that have been proposed and are now actively pursued include the development of new emission sources and new schemes for pulse compression, such as a radiofrequency (RF) cavity synchronized to the laser system. To obtain a high brightness of photoelectrons, Ropers et al. from the University of Göttingen developed the first UTEM driven by localized photoemission from a field-emitter cathode, which was based on a JEOL JEM-2100 F transmission electron microscope (thermal field emission). Using this UTEM, they achieved record pulse properties in ultrafast electron microscopy with a 9-Å focused beam diameter, a 200-fs pulse duration, and 0.6-eV energy widths, and conducted ultrafast imaging, diffraction, holography, and spectroscopy.[78,79] G. Caruso et al. proposed an architecture that allows the introduction of an ultrafast laser beam inside the cold field-emission source of a commercial TEM, aligning and focusing the laser spot on the apex of the nano-emitter.[80] Another method was proposed for realizing ultrafast electron microscopy: RF electron gun-based ultrafast relativistic electron microscopy (Yang Jinfeng; SJ; UCLA).[81] Here, high-energy electrons eliminate the space charge-dominated pulse broadening with sub-100 fs temporal resolution, but this technical scheme is still under development. Recent findings suggest that ultrafast electron microscopy with attosecond-structured wavefunctions may be feasible. Claus Ropers et al. introduced a framework for the preparation, coherent manipulation, and characterization of free-electron quantum states, experimentally demonstrating attosecond electron pulse trains.[82] Peter Baum et al. recently demonstrated that the time-resolved diffraction from crystalline Si reveals a <10-as delay of the Bragg emission, demonstrating the possibility of analytic attosecond–ångström diffraction.[83] The combination of attosecond science with electron microscopy can greatly improve the temporal resolution of the UTEM. Furthermore, a direct electron-detection camera has been integrated in the UTEM, which allows for imaging at low repetition rates.[35] As part of the Synergetic Extreme Condition User Facility project of CAS, China, a new UTEM based on the Cs-corrected TEM will be installed in five years in Huairou, Beijing. Meanwhile, a new UTEM based on a cryo-electron microscopy is being installed at the Institute of Physics, CAS.
The UTEM promises to become one of the most powerful experimental tools for the investigation of ultrafast dynamics at the nanoscale, combining a femtosecond temporal resolution with vast capabilities in imaging, diffraction, and spectroscopy. The next-generation UTEM with a field-emission gun and a Cs-corrector can provide a high brightness of the electron source to improve the spatial coherence or increase the spatial resolution for applications such as holography and spatially resolved spectroscopy. From a scientific viewpoint, although various phenomena have been demonstrated using the UTEM—including atomic motions during structural dynamics, transient near-field visualization, phase transitions, carrier transport, biological imaging, cryo-electron microscopy, and biomechanics—systematic research must be performed, and the potential of the UTEM must be explored for various aspects.
[1] | |
[2] | |
[3] | |
[4] | |
[5] | |
[6] | |
[7] | |
[8] | |
[9] | |
[10] | |
[11] | |
[12] | |
[13] | |
[14] | |
[15] | |
[16] | |
[17] | |
[18] | |
[19] | |
[20] | |
[21] | |
[22] | |
[23] | |
[24] | |
[25] | |
[26] | |
[27] | |
[28] | |
[29] | |
[30] | |
[31] | |
[32] | |
[33] | |
[34] | |
[35] | |
[36] | |
[37] | |
[38] | |
[39] | |
[40] | |
[41] | |
[42] | |
[43] | |
[44] | |
[45] | |
[46] | |
[47] | |
[48] | |
[49] | |
[50] | |
[51] | |
[52] | |
[53] | |
[54] | |
[55] | |
[56] | |
[57] | |
[58] | |
[59] | |
[60] | |
[61] | |
[62] | |
[63] | |
[64] | |
[65] | |
[66] | |
[67] | |
[68] | |
[69] | |
[70] | |
[71] | |
[72] | |
[73] | |
[74] | |
[75] | |
[76] | |
[77] | |
[78] | |
[79] | |
[80] | |
[81] | |
[82] | |
[83] |