As a promising solution to address energy dissipation in computing-intensive systems, Computing-in-Memory (CiM) architectures significantly reduce the large cost for data transition by integrating data computation in memories (such as SRAMs). However, when widely used error correction codes (such as Hamming codes) are employed in CiMs to guarantee a dependable operation against soft errors in memory, existing implementation schemes either require separate codec hardware, or a high energy dissipation design. This often offsets the benefits brought by CiMs. Therefore, this paper proposes the so-called Voltage-Translating SRAM-based CiM architecture (VTSCiM), to support an energy-efficient Hamming implementation as well as XOR operations. By employing a novel XOR computational principle based on a voltage-translating strategy, the implementation complexity of an XOR computation is reduced by half in most cases, and the supported number of operators is flexible. This enables a fast and low power operation in the proposed XOR unit design. Moreover, a pipeline architecture is proposed to support common Hamming codes (e.g., the (38, 32) codes). Hardware results obtained with post-layout simulation show that compared to existing schemes the proposed design achieves a reduction of over 29.2 % and 47.5% in terms of latency and energy dissipation, respectively. The benefits of energy efficiency, scalability and flexibility make the proposed design more attractive for CiMs in safety-critical applications.
This study describes a dielectric-modulated tunnel field-effect transistor (DM-TFET) biosensor with an H-shaped intrinsic Si channel, SiGe source pocket, and InGaAs source to improve electrical and sensing performance. The proposed design combines H-shaped channel engineering for effective biomolecule induced electrostatic modulation with SiGe pocket and InGaAs source engineering to increase source to channel junction side band-to-band tunneling (BTBT) performance. The Device was created and simulated with Silvaco ATLAS TCAD. Its electrical performance is measured in terms of Ion current, Ioff current, and Ion/Ioff ratio while the biomolecules induced sensing behaviour is investigated using drain current, energy band, electric field, and surface potential. The device is systematically optimized by varying the SiGe pocket thickness and composition, cavity Thickness and Length, source material, gate work function, and source/pocket doping concentrations. The optimized biosensor has sensitivities of 1.20 × 10⁷ for neutral biomolecules at K = 12, 1.18 × 10⁷ at Qf = +1 × 10¹⁰ cm⁻², and 3.84 × 10⁷ at Qf = -1 × 10¹² cm⁻² for positively and negatively charged biomolecules, respectively. Under actual sensing conditions, the impacts of cavity filling, non-uniform biomolecule distribution, and temperature impact are studied further. Finally, the biosensor's performance is evaluated in terms of selectivity and response time, highlighting its potential for extremely sensitive and label-free biomolecule detection.
Nanoporous evaporators offer a promising route to ultra-high heat flux thermal management because nanoscale confinement amplifies capillary pressure and enables precise regulation of interfacial transport. However, reducing pore size also increases liquid viscous resistance and drives vapor transport into slip, transitional, and free-molecular regimes, creating a fundamental trade-off that cannot be resolved through uniform cylindrical pores alone. This study develops a coupled multi-physics model for water evaporation through silicon membranes containing cylindrical, right conical, and inverted conical nanopores under a common nanofabrication constraint. The framework integrates axial pore geometry, capillary-driven liquid replenishment, augmented Young-Laplace meniscus mechanics with disjoining pressure, interfacial phase-change kinetics, and Knudsen-scale vapor transmission. Simulations performed for nanopores with rmin = 40 nm and rmax = 150 nm resolve the transition from an outlet-pinned meniscus to interface recession and dry out, while quantifying the corresponding mass flux and heat-transfer coefficient. The results show that peak heat transfer occurs at the critical superheat separating the pinned and receding regimes; after recession begins, the reduced molecular escape probability suppresses evaporative transport. Among the investigated nanostructures, right conical nanopores provide the highest critical superheat tolerance and peak heat-transfer coefficient by alleviating liquid-phase hydraulic losses while stabilizing the meniscus. A smaller minimum fabrication radius, a thinner membrane, and a steeper wall taper angle can further enhance performance, subject to nanofabrication tolerances, sidewall roughness, and mechanical-integrity constraints. These findings establish axial nanopore shaping as valuable theoretical insights for guiding the future structural design of high flux phase change cooling membranes.
Regulation of stem cell behavior using nanoengineered platforms such as nanomaterials and nanostructured surfaces has emerged as a promising strategy to complement conventional biochemical induction. Graphene has been widely investigated because of its unique physicochemical properties and their influence on stem cell behavior. In this study, we investigated layer-tuned graphene interfaces (LGIs) consisting of one, three, and five graphene layers, denoted as LGI1, LGI3, and LGI5, respectively. The LGIs were fabricated by chemical vapor deposition (CVD) and transferred onto flexible polyethylene terephthalate (PET) substrates, exhibiting layer dependent differences in wettability and electrical conductivity. Human mesenchymal stem cells (hMSCs) cultured on the LGIs exhibited differences in cell elongation and cytoskeletal organization, accompanied by altered expression of focal adhesion-associated genes (i.e., vinculin (VCL) and protein tyrosine kinase 2 (PTK2)) and metabolic activity. While chondrogenic matrix deposition showed minimal variation among the groups, LGI3 exhibited the most pronounced osteogenic-related staining. These findings suggest that graphene interfaces with different layer numbers provide distinct interfacial environments associated with differential hMSC responses and may serve as a useful platform for investigating cell–material interactions.
This design divides the alternating current generated by the TENG into two independent output stages per cycle: the output of the negative half-cycle is stored in the upper series-connected internal capacitors of the capacitor-diode array, while the output of the positive half-cycle is stored in the lower series-connected internal capacitors. This configuration enables the nanogenerator to charge both the upper and lower sets of capacitors within a single cycle simultaneously.When the front-stage capacitors are fully charged, the next cycle will charge the rear-stage capacitors, thereby progressively increasing the voltage. With an n-stage capacitor-diode cascade, the output voltage can be amplified by n times. Utilizing a series connection of capacitors (each fully charged internal capacitor functions as a small-capacity battery) to power various devices can significantly enhance and control the output power. This micro-power generation technology ingeniously utilizes the positive and negative half-cycles as charging periods, storing energy by serially connecting internal capacitors. Each fully charged capacitor functions as a micro-battery, forming a series-connected battery pack. By configuring micro-battery packs that meet specific voltage requirements to power devices, the system achieves a arbitrary-Voltage, low current power supply solution based on nanogenerators. Our power management strategy utilizes a voltage multiplier circuit to boost the nanogenerator's voltage while storing the charge or the energy generated in the internal capacitors of the multiplier circuit. By employing the series-connected upper or lower capacitors of the voltage multiplier circuit (each fully charged internal capacitor acts as a small-capacity battery), power can be supplied to different load devices R. The method to enhance power output is as follows: We can control the charging time tch to obtaine Wc , while keeping the total energy Wc constant, the continuous power supply from the nanogenerator is converted into short-term power supply from a fully charged capacitor Ctotal to the load R . By controlling the discharging time tdis , the output power can be increased to meet the practical requirements of different power devices, significantly expanding the practical application range of nanogenerators. We have experimentally verified that this method can effectively enhance the output power of nanogenerators and their supporting circuits. Due to the system consisting solely of diodes and capacitors, energy loss is nearly negligible. It is regrettable that the system cannot power devices while charging the capacitors, as the capacitors essentially function like small batteries. This limitation could potentially be addressed by increasing the number of nanogenerators to alternately supply power to the load R .
