Abstract
As machine vision technology generates large amounts of data from sensors, it requires efficient computational systems for visual cognitive processing. Recently, in-sensor computing systems ...have emerged as a potential solution for reducing unnecessary data transfer and realizing fast and energy-efficient visual cognitive processing. However, they still lack the capability to process stored images directly within the sensor. Here, we demonstrate a heterogeneously integrated 1-photodiode and 1 memristor (1P-1R) crossbar for in-sensor visual cognitive processing, emulating a mammalian image encoding process to extract features from the input images. Unlike other neuromorphic vision processes, the trained weight values are applied as an input voltage to the image-saved crossbar array instead of storing the weight value in the memristors, realizing the in-sensor computing paradigm. We believe the heterogeneously integrated in-sensor computing platform provides an advanced architecture for real-time and data-intensive machine-vision applications via bio-stimulus domain reduction.
Abstract
The optical Stark effect is a coherent light–matter interaction describing the modification of quantum states by non-resonant light illumination in atoms, solids and nanostructures. ...Researchers have strived to utilize this effect to control exciton states, aiming to realize ultra-high-speed optical switches and modulators. However, most studies have focused on the optical Stark effect of only the lowest exciton state due to lack of energy selectivity, resulting in low degree-of-freedom devices. Here, by applying a linearly polarized laser pulse to few-layer ReS
2
, where reduced symmetry leads to strong in-plane anisotropy of excitons, we control the optical Stark shift of two energetically separated exciton states. Especially, we selectively tune the Stark effect of an individual state with varying light polarization. This is possible because each state has a completely distinct dependence on light polarization due to different excitonic transition dipole moments. Our finding provides a methodology for energy-selective control of exciton states.
Coherent light–matter interaction can transiently modulate the quantum states of matter under nonresonant laser excitation. This phenomenon, called the optical Stark effect, is one of the promising ...candidates for realizing ultrafast optical switches. However, the ultrafast modulations induced by the coherent light–matter interactions usually involve unwanted incoherent responses, significantly reducing the overall operation speed. Here, by using ultrafast pump–probe spectroscopy, we suppress the incoherent response and modulate the coherent-to-incoherent ratio in the two-dimensional semiconductor ReS2. We selectively convert the coherent and incoherent responses of an anisotropic exciton state by solely using photon polarizations, improving the control ratio by 3 orders of magnitude. The efficient modulation was enabled by transient superpositions of differential spectra from two nondegenerate exciton states due to the light polarization dependencies. This work provides a valuable contribution toward realizing ideal ultrafast optical switches.
Strongly bound excitons are a characteristic hallmark of 2D semiconductors, enabling unique light–matter interactions and novel optical applications. Platinum diselenide (PtSe2) is an emerging 2D ...material with outstanding optical and electrical properties and excellent air stability. Bulk PtSe2 is a semimetal, but its atomically thin form shows a semiconducting phase with the appearance of a band‐gap, making one expect strongly bound 2D excitons. However, the excitons in PtSe2 have been barely studied, either experimentally or theoretically. Here, the authors directly observe and theoretically confirm excitons and their ultrafast dynamics in mono‐, bi‐, and tri‐layer PtSe2 single crystals. Steady‐state optical microscopy reveals exciton absorption resonances and their thickness dependence, confirmed by first‐principles calculations. Ultrafast transient absorption microscopy finds that the exciton dominates the transient broadband response, resulting from strong exciton bleaching and renormalized band‐gap‐induced exciton shifting. The overall transient spectrum redshifts with increasing thickness as the shrinking band‐gap redshifts the exciton resonance. This study provides novel insights into exciton photophysics in platinum dichalcogenides.
This work for the first time provides direct observation of ultrafast exciton dynamic in mono‐, bi‐, and tri‐layer platinum diselenide single crystals, a rising 2D material star. It is found that photoinduced modulation of excitons and their thickness dependence dominate overall ultrafast transient spectra in the broadband from visible to the near‐infrared edge, confirmed by theoretical calculations.
Quantum beats, periodic oscillations arising from coherent superposition states, have enabled exploration of novel coherent phenomena. Originating from strong Coulomb interactions and reduced ...dielectric screening, two-dimensional transition metal dichalcogenides exhibit strongly bound excitons either in a single structure or hetero-counterpart; however, quantum coherence between excitons is barely known to date. Here we observe exciton quantum beats in atomically thin ReS2 and further modulate the intensity of the quantum beats signal. Surprisingly, linearly polarized excitons behave like a coherently coupled three-level system exhibiting quantum beats, even though they exhibit anisotropic exciton orientations and optical selection rules. Theoretical studies are also provided to clarify that the observed quantum beats originate from pure quantum coherence, not from classical interference. Furthermore, we modulate on/off quantum beats only by laser polarization. This work provides an ideal laboratory toward polarization-controlled exciton quantum beats in two-dimensional materials.
Quantum optoelectronic devices capable of isolating a target degree of freedom (DoF) from other DoFs have allowed for new applications in modern information technology. Many works on solid-state ...spintronics have focused on methods to disentangle the spin DoF from the charge DoF1, yet many related issues remain unresolved. Although the recent advent of atomically thin transition metal dichalcogenides (TMDs) has enabled the use of valley pseudospin as an alternative DoF2,3, it is nontrivial to separate the spin DoF from the valley DoF since the time-reversal valley DoF is intrinsically locked with the spin DoF4. Here, we demonstrate lateral TMD–graphene–topological insulator hetero-devices with the possibility of such a DoF-selective measurement. We generate the valley-locked spin DoF via a circular photogalvanic effect in an electric-double-layer WSe2 transistor. The valley-locked spin photocarriers then diffuse in a submicrometre-long graphene layer, and the spin DoF is measured separately in the topological insulator via non-local electrical detection using the characteristic spin–momentum locking. Operating at room temperature, our integrated devices exhibit a non-local spin polarization degree of higher than 0.5, providing the potential for coupled opto-spin–valleytronic applications that independently exploit the valley and spin DoFs.
Abstract
Artificial neural networks (ANNs) are widely used in numerous artificial intelligence‐based applications. However, the significant amount of data transferred between computing units and ...storage has limited the widespread deployment of ANN for the artificial intelligence of things (AIoT) and power‐constrained device applications. Therefore, among various ANN algorithms, quantized neural networks (QNNs) have garnered considerable attention because they require fewer computational resources with minimal energy consumption. Herein, an oxide‐based ternary charge‐trap transistor (CTT) that provides three discrete states and non‐volatile memory characteristics are introduced, which are desirable for QNN computing. By employing a differential pair of ternary CTTs, an artificial synaptic segregation with multilevel quantized values for QNNs is demostrated. The approach establishes a platform that combines the advantages of multiple states and robustness to noise for in‐memory computing to achieve reliable QNN performance in hardware, thereby facilitating the development of energy‐efficient AIoT.
Defect identification has been a significant task in various fields to prevent the potential problems caused by imperfection. There is great attention for developing technology to accurately extract ...defect information from the image using a computing system without human error. However, image analysis using conventional computing technology based on Von Neumann structure is facing bottlenecks to efficiently process the huge volume of input data at low power and high speed. Herein efficient defect identification is demonstrated via a morphological image process with minimal power consumption using an oxide transistor and a memristor‐based crossbar array that can be applied to neuromorphic computing. Using a hardware and software codesigned neuromorphic system combined with a dynamic Gaussian blur kernel operation, an enhanced defect detection performance is successfully demonstrated with about 104 times more power‐efficient computation compared to the conventional complementary metal‐oxide semiconductor (CMOS)‐based digital implementation. It is believed the back end of line (BEOL)‐compatible all‐oxide‐based memristive crossbar array provides the unique potential toward universal artificial intelligence of things (AIoT) applications where conventional hardware can hardly be used.
Efficient defect identification is demonstrated with oxide memristive crossbar arrays. The hardware consisting of an oxide transistor and a memristor operates as a neuromorphic analogue computing system that can imitate the human brain's operation. Based on the hardware and software codesigned image processing, defect information is extracted with 104 times more power‐efficient computation compared to the conventional digital implementation.
Strong interactions between excitons are a characteristic feature of two-dimensional (2D) semiconductors, determining important excitonic properties, such as exciton lifetime, coherence, and ...photon-emission efficiency. Rhenium disulfide (ReS2), a member of the 2D transition-metal dichalcogenide (TMD) family, has recently attracted great attention due to its unique excitons that exhibit excellent polarization selectivity and coherence features. However, an in-depth understanding of exciton-exciton interactions in ReS2 is still lacking. Here we used ultrafast pump-probe spectroscopy to study exciton-exciton interactions in monolayer (1L), bilayer (2L), and triple layer ReS2. We directly measure the rate of exciton-exciton annihilation, a representative Auger-type interaction between excitons. It decreases with increasing layer number, as observed in other 2D TMDs. However, while other TMDs exhibit a sharp weakening of exciton-exciton annihilation between 1L and 2L, such behavior was not observed in ReS2. We attribute this distinct feature in ReS2 to the relatively weak interlayer coupling, which prohibits a substantial change in the electronic structure when the thickness varies. This work not only highlights the unique excitonic properties of ReS2 but also provides novel insight into the thickness dependence of exciton-exciton interactions in 2D systems.
3D sensing is a primitive function that allows imaging with depth information generally achieved via the time‐of‐flight (ToF) principle. However, time‐to‐digital converters (TDCs) in conventional ToF ...sensors are usually bulky, complex, and exhibit large delay and power loss. To overcome these issues, a resistive time‐of‐flight (R‐ToF) sensor that can measure the depth information in an analog domain by mimicking the biological process of spike‐timing‐dependent plasticity (STDP) is proposed herein. The R‐ToF sensors based on integrated avalanche photodiodes (APDs) with memristive intelligent matters achieve a scan depth of up to 55 cm (≈89% accuracy and 2.93 cm standard deviation) and low power consumption (0.5 nJ/step) without TDCs. The in‐depth computing is realized via R‐ToF 3D imaging and memristive classification. This R‐ToF system opens a new pathway for miniaturized and energy‐efficient neuromorphic vision engineering that can be harnessed in light‐detection and ranging (LiDAR), automotive vehicles, biomedical in vivo imaging, and augmented/virtual reality.
Depth information is achieved by a neuron‐like behavior of memristive intelligent matter. The HfO2 memristors exhibit in‐memory depth sensing capability that can replace the conventional complex memory and time‐to‐digital converter (TDC) logics. By using the proposed resistive time‐of‐flight principle, the 3D imaging capability as well as crossbar‐based image reconstruction and classification are demonstrated.