My works are uniquely intriguing blends of materials science, physics, and mechanical engineering
My works are uniquely intriguing blends of materials science, physics, and mechanical engineering
In this project I manufacture Ti-TiB nanocomposites via vacuum sintering method at 1200C assisted by argon. The samples then undergo metallographic polish followed by mechanical characterization from nano-scale to micro and macro scale. I use Bruker Hysitron TI980 Nanoindenter to characterize mechanical properties at nano and microscale. I employed nano-indentations on both the Ti and TiB phases to map the mechanical properties like Young's modulus and hardness at nanoscale. Then I switched to a high-load transducer to perform indents at higher load to induce cracks. Since these materials are more ductile, no cracks were induced. Then I employed scratch testing to chracterize fracture toughness. I performed scratches of 5 different loads on each type of samples to calculate the fracture toughness. For macroscale characterization I used dynamic mechanical analyzer (DMA), and fatigue tester from TestResources. I also used a furnace enclosure to raise the temperature of the sample while doing testing. The furnace temperature was raised to as high as 350 C to characterize the Young's modulus of the larger samples. Furthermore, I assess the metrology and microstructure of the samples using metrology techniques such as XRD, SEM, EDS, X-ray micro tomography, profilometry, scanning probe microscopy (SPM).
The size scale is very crucial for any memristor because of their application demand. Previously we reported geopolymer memristors fabricated at milimeter scale. These devices also had limitations in retaining properties due to loss of moisture. We have been working to address these critical by significantly improving the geopolymer memristors. We are now able to fabricate these memristors at 200 micron size. Also, by treating these memristors with ionic liquid, their properties are now retained by 200% longer. Also, we are investigating the effect of silver oxide formation during voltage sweeps, to the memory properties of these memristors. Advanced microscopy techniques like SEM, EDS, Scanning TEM, and other materials analysis tools like XRD, FTIR, etc. were utilized for these investigations. This is a significant milestone in geopolymer memristor research.
Related Publications:
Ahmadipour, M., Shakib, M. A., Gao, Z., Sarles, S., Lamuta, C., Montazami, R., Scaled-Down Ionic Liquid-Functionalized Geopolymer Memristors. Materials Horizons. Article Link.
Features in the Cover of Materials Horizons (designed by me)!
Reservoir computing (RC) is a computing paradigm that involves projection of features generated by physical reservoirs with short-term memory utilizing temporal data.
The RC systems are more advantageous to the traditional neural networks in that it requires significantly less weight training that cuts down the computational cost. This is accompanied by replacing the hidden layers by a readout layer which is a linear output layer. Schematics of a memristor-based RC system consisting of 2 parts is shown in figure. The first part is the reservoir which contains the memristors. This part is connected to the input, which is generated by computer-generated or hand-written digit, or any other physical or virtual input. The reservoir has nodes, which evolve dynamically over time as temporal data is fed into the reservoirs. The temporal state of each memristor in the reservoir is called memristors state. A collection of all memristor states is called reservoir state. The reservoir performs a non-linear transformation of the temporal input data and maps them into a new space. The reservoir state is further trained by the outing random weights on them in the readout layer, which is the second part of the RC system. By using machine learning algorithms like linear or logistic regression, the readout layer can perform classification task.
Related Publications:
Shakib, M. A., Maraj, J. J., Gao, Z., Ahmadipour, M., Montazami, R., Sarles, S. A., Lamuta, C., Geopolymer Memristor-based Physical Reservoir Computing for Pattern Recognition (in review).
Memristors, also known as artificial synapses, are devices able to mimic the memory functions of biological synapses. To emulate synaptic functions, memristors need to exhibit plasticity which is a pivotal phenomenon in their biological counterparts. In a previous work we demonstrated that geopolymers present memristive properties. In this work, we study different types of synaptic plasticity properties of geopolymer memristors. We demonstrate short-term and long-term memory resulting from potentiation-depression, Hebbian learning inspired Spike-Timing-Dependent Plasticity (STDP), Spike-Rate-Dependent Plasticity (SRDP), History-Dependent Plasticity, Paired-Pulse Facilitation (PPF) and Depression (PPD), and Post-Tetanic Potentiation (PTP). These synaptic properties can be ascribed to the electro-osmosis-induced movement of ions in the capillaries and pores of the geopolymer memristors. These properties are extremely promising for the use of geopolymers in neuromorphic computing applications.
Related Publications:
Shakib, M. A., Gao, Z., Lamuta, C., Synaptic Properties of Geopolymer Memristors: Synaptic Plasticity, Spike-Rate-Dependent Plasticity, and Spike-Timing-Dependent Plasticity. ACS Appld. Elect. Mater. 2023. Article Link.
Memristors are electric components that emulate the memory and computational properties of biological synapses by remembering the current that flows through them. In this research we have demonstrated, for the first time, the memristive properties of geopolymers. Geopolymers are inexpensive ceramic materials manufactured at room temperature from alkaline activation of amorphous aluminosilicate precursors. We have proposed a physics-based model, which demonstrates that electroosmosis in the bulk geopolymer pores induces ion channels that foster change in the overall conductance of the bulk material, contributing to the observed memristive behavior. This model opens the door to a new category of porous electroosmosis-based bulk memristors. We have also demonstrated synaptic functions such as short-term plasticity and long-term plasticity, as well as endurance and retention capabilities. The reported findings pave the way to the use of geopolymers for low-cost applications in neuromorphic computing.
Related Publications:
Shakib, M. A., Gao, Z., Candamano, S., Lamuta, C., Ion Channels and Electroosmosis in Porous Geopolymers: A Novel Category of Low-Cost Memristors. Adv. Funct. Mater. 2023, 2306535.Article link
Related Publication:
Utku Uzun, Parth Kotak, Mahmudul Alam Shakib, Rabiu Onorouiza Mamman, Sawsan Daws, Chia-Nung Kuo, Chin Shan Lue, Antonio Politano, Caterina Lamuta. Ion Nanomechanical properties and wear resistance of Palladium diselenide (PdSe2) for flexible electronics. Mat. Sci. Engg B, 2022, Article link
In this project I looked at the microstructure of geopolymers at sub-micron scale to assess the pore geometry and defects. The pore geometry play an essential role in the change of conductance of the geopolymer memristors. X-ray microtomography is a non-destructive method to image the microstructure of the materials in 3D. The samples were scanned using ZEISS Xradia 520 Versa that at a resolution of 370 nanometers and each sample took 10 hours to scan at 3D. The 3D images were analyzed using ORS Dragonfly software.
In this project I reinforced geopolymer mortars with graphene nanoplatelets and carbon black to enhance their piezoelectric properties. The use of graphene nanoplatelets has improved the direct piezoelectric charge coefficient by 200% and the use of 1% carbon black has improved the coefficient by 8000%.
Conference Talk:
ASME SMASIS 2024 (Atlanta, Georgia).