Semiconducting nanocrystals with a size of less than 10 nm are referred to as quantum dots (QDs). Their unique electrical, optical, and chemical properties make them attractive in various scientific and commercial applications including quantum and neuromorphic computing.1
Image Credit: ArtemisDiana/Shutterstock.com
Quantum computing relies on intrinsic quantum properties of superposition and entanglement to create algorithms faster than the classical ones. The separable and entangled classification of quantum states can be achieved through QDs.
Alternatively, neuromorphic computing drives inspiration from the human brain's neural structures and processing methods to simultaneously store data and perform computing.2
Quantum Dots: An Overview
QDs are commonly called zero-dimensional nanomaterials or artificial atoms. The incredibly small size of quantum dots gives them unique properties, driven by the quantum confinement effect, which occurs when an electron cloud is physically restricted. Initially, they were used to control the colors of silicate glasses as their optical properties change with size. However, recent advances in materials science allow for precise size control of QDs, opening new avenues for their application like display screens, health care, wastewater management, solar energy, biosensors, and quantum and neuromorphic computing.1
Quantum dots can emit any color of light based solely on their size, as size variations alter the bandgap. Additionally, the magnetic memory (coercive force) required for reversing a QD’s internal magnetic field is size-dependent. Single-component (selenides or sulphides) QDs are called core-type and their properties are tunable by size change. Core-shell type QDs have a coating of another material on the core (for example, CdSe core and ZnS shell) and exhibit higher quantum yield, efficiency, and brightness. Novel tunable properties can be achieved in alloyed QDs with varying compositions irrespective of their size. Such unique and tunable size, shape, and composition make them suitable for various applications.1
QDs are generally synthesized as colloids suspended in solutions or epitaxial structures grown on solid crystalline substrates. The former method results in QDs with distinguishable optical properties while the latter allows for better control of size and shape through regular patterns. Advanced techniques like focused ion beam, wet chemical etching, reactive ion etching, and vapor phase methods are also used for QD synthesis as they offer higher efficiency and control over quantum confinement properties. The choice of the method depends on the desired properties and target application.1
Role of Quantum Dots in Neuromorphic Computing
In traditional computers, the physically separate central processor and data storage system experience the von Neumann bottleneck, where data transfers between the processor and memory create significant congestion. The need for the construction of non-von Neumann configurations for scalable networks led researchers to learn from the human brain.2 The human brain performs data processing and memory tasks parallelly with extremely low power consumption and in a very compact space using approximately 1011 neurons interconnected by more than 1014 synapses.2,3 The neural networks get reconfigured by learning experiences and alter the strength of synaptic connections similar to enabling memory and learning capacities in humans.2
The brain-like neuromorphic computing systems have lately become attractive for simultaneous data storage and computing on a compact chip. These can be employed for brain-inspired applications like self-learning, logic computing, pattern identification, and speech recognition. Several materials have been explored to fabricate electronic chips comprising artificial neurons and learning synapses like inorganic and organic oxides, carbon nanomaterials, 2D materials, ferroelectrics, perovskites, and semiconducting QDs.2
QD-based memristors exhibit high performance and energy efficiency in neuromorphic computing devices. Memristors are crucial for constructing simulated neural networks as they have reconfigurable history-dependent resistance-switching and can imitate biological synaptic behavior. Furthermore, QD-based memories have a simple sandwiched structure and can operate at a high speed (fast erasing process) with low fabrication cost and power consumption.3 QDs can be integrated with neuromorphic systems in the form of resistive random-access memory (RRAM), flash memory, and photonic synaptic devices.4
Advantages of Quantum Dots in Computing
Conventional semiconducting materials generally depict a trade-off between functionality and substrate size. For instance, high-quality crystalline Si substrates with high career mobility can be produced only with limited area, while large-area amorphous Si thin films have low switching speeds. Organic semiconductors offer large-area integration but exhibit low stability and relatively poor electrical functionality.2
Semiconducting QDs allow the development of scalable neuromorphic devices with efficient optoelectronic performance, stability, and size-tunability at lower costs. Surface modification of QDs with various ligands like octadecyl phosphonic acid and polystyrene enhances charge transfer between them, thereby enabling dynamic switching and fast erasing time in QD-based memories. The distinctive optical properties of quantum dots can be leveraged to enhance synaptic functions, such as plasticity, in artificial neuromorphic computing systems through photonic potentiation.4
Electro-photoactive self-assembled semiconductor QD films can be incorporated in flash memories as traps, floating gates, or channels and in memristors as resistive switches. The high surface-to-volume ratio of semiconductor QDs offer novel physical mechanisms in RRAMs to ensure fast switching, long-state retention, high reproducibility, and low-energy operation.2 The facile colloidal synthesis and robust switching properties impart QDs the potential to revolutionize computing through enhanced learning capabilities and efficiency.
Challenges and Research Frontiers
The major challenge associated with the application of QDs in any field is their toxicity. QD toxicity may be due to their own chemical and physical properties of size, shape, charge, functional groups, stability, concentration, and composition, as well as the environment. Thus, significant research efforts are being to assess and resolve the issue.1
The oxidative, mechanical, and photocatalytic stabilities of QDs make their integration in neuromorphic computing architectures challenging.1 A commercial-level development of QD-based neuromorphic computing devices is difficult due to flexibility and robustness issues. Novel material synthesis and thin-film fabrication methods are crucial to improving the uniformity and morphology of QD building blocks and reducing device variability.2
The scaling up of QD memristors requires integration with the complementary metal-oxide semiconductor (CMOS) technology as CMOS circuits are used for control and interface operations. An efficient CMOS-QD memristor is the key to achieving commercial-level data storage and processing capabilities of neuromorphic systems. Several efforts are being made to resolve such issues like the dynamical characterization of the switching process in QD-based synapses along with device modeling strategies.2
Future Perspectives and Conclusion
The transformative potential of QDs in neuromorphic computing has been successfully demonstrated by several researchers. Their application in neuromorphic computing holds broader implications for artificial intelligence, robotics, and energy-efficient computing. Substantial improvement in computing speed, energy efficiency, and fabrication costs can be achieved through device miniaturization only to a certain extent due to the approaching physical limit of Moore’s law. Replication of the biological brain by electronic devices can prove to be sustainable in achieving future computing and robotic standards.2
QD-based neuromorphic systems capable of event-driven spiking communications can significantly impact the commercialization of artificial intelligence and deep learning technologies. Future research and exploration to obtain a range of ideal QDs for various applications at a commercial scale should use an interdisciplinary approach based on materials science and computing.
References and Further Reading
- Mohamed, W. A. A., Abd El-Gawad, H., Mekkey, S., Galal, H., Handal, H., Mousa, H., & Labib, A. (2021). Quantum dots synthetization and future prospect applications. Nanotechnology Reviews, 10(1), 1926–1940. https://doi.org/10.1515/ntrev-2021-0118
- Ziyu Lv, Wang, Y., Chen, J., Wang, J., Zhou, Y., & Han, S.-T. (2020). Semiconductor Quantum Dots for Memories and Neuromorphic Computing Systems. Chemical Reviews, 120(9), 3941–4006. https://doi.org/10.1021/acs.chemrev.9b00730
- Wang, Z., Wang, W., Liu, P., Liu, G., Li, J., Zhao, J., Zhou, Z., Wang, J., Pei, Y., Zhao, Z., Li, J., Wang, L., Jian, Z., Wang, Y., Guo, J., & Yan, X. (2022). Superlow Power Consumption Artificial Synapses Based on WSe 2 Quantum Dots Memristor for Neuromorphic Computing. Research, 2022. https://doi.org/10.34133/2022/9754876
- Kim, J., Song, S., Kim, Y.-H., & Park, S. K. (2020). Recent Progress of Quantum Dot‐based Photonic Devices and Systems: A Comprehensive Review of Materials, Devices, and Applications. Small Structures. https://doi.org/10.1002/sstr.202000024
Disclaimer: The views expressed here are those of the author expressed in their private capacity and do not necessarily represent the views of AZoM.com Limited T/A AZoNetwork the owner and operator of this website. This disclaimer forms part of the Terms and conditions of use of this website.