Embedded SRAM memory is a vital component in modern SoCs. More than 80% of the System-on-Chip (SoC) die area is often occupied by SRAM arrays. As such, system reliability and yield is largely governed by the SRAM's performance and robustness. The aggressive scaling trend in CMOS device minimum feature size, coupled with the growing demand in high-capacity memory integration, has imposed the use of minimal size devices to realize a memory bitcell. The smallest 6T SRAM bitcell to date occupies a 0.1um2 in silicon area. SRAM bitcells continue to benefit from an aggressive scaling trend in CMOS technologies. Unfortunately, other system components, such as interconnects, experience a slower scaling trend. This has resulted in dramatic deterioration in a cell's ability to drive a heavily-loaded interconnects. Moreover, the growing fluctuation in device properties due to Process, Voltage, and Temperature (PVT) variations has added more uncertainty to SRAM operation. Thus ensuring the ability of a miniaturized cell to drive heavily-loaded bitlines and to generate adequate voltage swing is becoming challenging. A large percentage of state-of-the-art SoC system failures are attributed to the inability of SRAM cells to generate the targeted bitline voltage swing within a given access time. The use of read-assist mechanisms and current mode sense amplifiers are the two key strategies used to surmount bitline loading effects. On the other hand, new bitcell topologies and cell supply voltage management are used to overcome fluctuations in device properties. In this research we tackled conventional 6T SRAM bitcell limited drivability by introducing new integrated voltage sensing schemes and current-mode sense amplifiers. The proposed schemes feature a read-assist mechanism. The proposed schemes' functionality and superiority over existing schemes are verified using transient and statistical SPICE simulations. Post-layout extracted views of the devices are used for realistic simulation results. Low-voltage operated SRAM reliability and yield enhancement is investigated and a wordline boost technique is proposed as a means to manage the cell's WL operating voltage. The proposed wordline driver design shows a significant improvement in reliability and yield in a 400-mV 6T SRAM cell. The proposed wordline driver design exploit the cell's Dynamic Noise Margin (DNM), therefore boost peak level and boost decay rate programmability features are added. SPICE transient and statistical simulations are used to verify the proposed design's functionality. Finally, at a bitcell-level, we proposed a new five-transistor (5T) SRAM bitcell which shows competitive performance and reliability figures of merit compared to the conventional 6T bitcell. The functionality of the proposed cell is verified by post-layout SPICE simulations. The proposed bitcell topology is designed, implemented and fabricated in a standard ST CMOS 65nm technology process. A 1.2_ 1.2 mm2 multi-design project test chip consisting of four 32-Kbit (256-row x 128-column) SRAM macros with the required peripheral and timing control units is fabricated. Two of the designed SRAM macros are dedicated for this work, namely, a 32-Kbit 5T macro and a 32-Kbit 6T macro which is used as a comparison reference. Other macros belong to other projects and are not discussed in this document.

The ever expanding range of applications for embedded systems continues to offer new challenges (and opportunities) to chip manufacturers. Applications ranging from exciting high resolution gaming to routine tasks like temperature control need to be supported on increasingly small devices with shrinking dimensions and tighter energy budgets. These systems benefit greatly by having the capability to operate over a wide range of supply voltages, known as ultra dynamic voltage scaling (U-DVS). This refers to systems capable of operating from nominal voltages down to sub-threshold voltages. Memories play an important role in these systems with future chips estimated to have over 80% of chip area occupied by memories. This thesis presents the design and characterization of an ultra dynamic voltage scalable memory (SRAM) that functions from nominal voltages down to sub-threshold voltages without the need for external support. The key contributions of the thesis are as follows: 1) A variation tolerant reference generation for single ended sensing: We present a reference generator, for U-DVS memories, that tracks the memory over a wide range of voltages and is tunable to allow functioning down to sub-threshold voltages. Replica columns are used to generate the reference voltage which allows the technique to track slow changes such as temperature and aging. A few configurable cells in the replica column are found to be sufficient to cover the whole range of voltages of interest. The use of tunable delay line to generate timing is shown to help in overcoming the effects of process variations. 2) Random-sampling based tuning algorithm: Tuning is necessary to overcome the in-creased effects of variation at lower voltages. We present an random-sampling based BIST tuning algorithm that significantly speed-up the tuning ensuring that the time required to tune is comparable to a single MBIST algorithm. Further, the use of redundancy after delay tuning enables maximum utilization of redundancy infrastructure to reduce power consumption and enhance performance. 3) Testing and Characterization for U-DVS systems: Testing and characterization is an important challenge in U-DVS systems that have remained largely unexplored. We propose an iterative technique that allows realization of an on-chip oscilloscope with minimal area overhead. The all digital nature of the technique makes it simple to design and implement across technology nodes. Combining the proposed techniques allows the designed 4 Kb SRAM array to function from 1.2 V down to 310 mV with reads functioning down to 190 mV. This would contribute towards moving ultra wide voltage operation a step closer towards implementation in commercial designs.

With growing connectivity in the modern era, the risk of encrypted data stored in hardware being exposed to third-party adversaries is higher than ever. The security of encrypted data depends on the secrecy of the stored key. Conventional methods of storing keys in Non-Volatile Memory have been shown to be susceptible to physical attacks. Physically Unclonable Functions provide a unique alternative to conventional key storage. SRAM PUFs utilize inherent process variation caused during manufacturing to derive secret keys from the power-up values of SRAM memory cells. This thesis analyzes the effect of supply ramp-up times on the reliability of SRAM PUFs. We use SPICE simulations as the platform to observe the effect of supply ramp times at the circuit level using carefully controlled supply voltages during power-up. We also measure the effect of supply ramp times on commercially available SRAM ICs by performing reliability and uniqueness measurements on two commercial SRAM models. Finally, a hardware implementation is proposed in a commercial 16nm FinFET technology to establish the design flow for taping out a custom SRAM IC with separated peripheral and core power supplies that would allow for experimental evaluation of sequenced power supplies on the SRAM PUF.

The project deals with the testing of SRAM memories using method MBIST with the utilisation of sofware tool Tessent Memory BIST. The main purpose is to get familiar with memory testing and to create a design for testing on a specific chip which after its implementation on the chip will retain the original features and functions. Subsequently, the tool is evaluated on its usability.

Energy efficient memory designs are becoming increasingly important, especially for applications related to mobile video technology and machine learning. The growing popularity of smart phones, tablets and other mobile devices has created an exponential demand for video applications in today’s society. When mobile devices display video, the embedded video memory within the device consumes a large amount of the total system power. This issue has created the need to introduce power-quality tradeoff techniques for enabling good quality video output, while simultaneously enabling power consumption reduction. Similarly, power efficiency issues have arisen within the area of machine learning, especially with applications requiring large and fast computation, such as neural networks. Using the accumulated data knowledge from various machine learning applications, there is now the potential to create more intelligent memory with the capability for optimized trade-off between energy efficiency, area overhead, and classification accuracy on the learning systems. In this dissertation, a review of recently completed works involving video and machine learning memories will be covered. Based on the collected results from a variety of different methods, including: subjective trials, discovered data-mining patterns, software simulations, and hardware power and performance tests, the presented memories provide novel ways to significantly enhance power efficiency for future memory devices. An overview of related works, especially the relevant state-of-the-art research, will be referenced for comparison in order to produce memory design methodologies that exhibit optimal quality, low implementation overhead, and maximum power efficiency. / National Science Foundation / ND EPSCoR / Center for Computationally Assisted Science and Technology (CCAST)

abstract: Recent years have seen fin field effect transistors (finFETs) dominate modern complementary metal oxide semiconductor (CMOS) processes, [1][2], e.g., at the sub 20 nm technology nodes, as they alleviate short channel effects, provide lower leakage, and enable some continued VDD scaling. However, a realistic finFET based predictive process design kit (PDK) that supports investigation into both circuit and physical design, encompassing all aspects of digital design, for academic use has been unavailable. While the finFET based FreePDK15 was supplemented with a standard cell library, it lacked full physical verification (LVS) and parasitic extraction at the time [3][4]. Consequently, the only available sub 45 nm educational PDKs are the planar CMOS based Synopsys 32/28 nm and FreePDK45 (45 nm PDK) [5][6]. The cell libraries available for those processes are not realistic since they use large cell heights, in contrast to recent industry trends. Additionally, the SRAM rules and cells provided by these PDKs are not realistic. Because finFETs have a 3D structure, which affects transistor density, using planar libraries scaled to sub 22 nm dimensions for research is likely to give poor accuracy. Commercial libraries and PDKs, especially for advanced nodes, are often difficult to obtain for academic use, and access to the actual physical layouts is even more restricted. Furthermore, the necessary non disclosure agreements (NDAs) are un manageable for large university classes and the plethora of design rules can distract from the key points. NDAs also make it difficult for the publication of physical design as these may disclose proprietary design rules and structures. This work focuses on the development of realistic PDKs for academic use that overcome these limitations. These PDKs, developed for the N7 and N5 nodes, even before 7 nm and 5 nm processes were available in industry, are thus predictive. The predictions have been based on publications of the continually improving lithography, as well as estimates of what would be available at N7 and N5. For the most part, these assumptions have been accurate with regards to N7, except for the expectation that extreme ultraviolet (EUV) lithography would be widely available, which has turned out to be optimistic. / Dissertation/Thesis / Doctoral Dissertation Electrical Engineering 2019

Indiana University-Purdue University Indianapolis (IUPUI) / Gopinath, Anoop. M.S.E.C.E., Purdue University, May 2018. Low-Power Hybrid TFET-CMOS Memory. Major Professor: Maher E. Rizkalla. The power consumption and the switching speed of the current CMOS technology have reached their limits. In contrast, architecture design within computer systems are continuously seeking more performance and e ciency. Advanced technologies that optimize the power consumption and switching speed may help deliver this e ciency. Indeed, beyond CMOS technology may be a viable approach to meeting the ever increasing need for low-power design. These technology includes devices such as Tunnel Field E ect Transistor (TFET), Graphene based devices such as GFET and GRNFET and FinFET. However, the low cross-sectional area of the channel asso- ciated with smaller technology nodes brings with it the challenges associated with leakage current below the threshold. Mitigating these challenges with devices such as TFETs may allow higher levels of integration, faster switching speed and lower power consumption. This thesis investigates the use of Gallium Nitride (GaN) TFET devices at 20nm for memory cells. These cells can be used in the L1 data cache of the Graphic Processing Units (GPU) thereby minimizing the static power and the dynamic power within these memory systems. The TFET technology was chosen since it has a low subthreshold slope of nearly 30mV/decade. This enables the TFET-based cells to function with a 0.6V supply voltage leading to reduced dynamic power consumption and leakage current when compared to the current CMOS technology. The results suggest that there are bene ts in pursuing an integrated TFET-based technology for Very Large Scale Integrated Circuit (VLSI) design. These bene ts are demonstrated using simulation at the schematic-level using Cadence Virtuoso.

Indiana University-Purdue University Indianapolis (IUPUI) / In this thesis, we investigated the 7 nm FinFET technology for its delay-power product performance. In our study, we explored the ASAP7 library from Arizona State University, developed in collaboration with ARM Holdings. The FinFET technology was chosen since it has a subthreshold slope of 60mV/decade that enables cells to function at 0.7V supply voltage at the nominal corner. An emphasis was focused on characterizing the Non-Ideal effects, delay variation, and power for the FinFET device. An exhaustive analysis of the INVx1 delay variation for different operating conditions was also included, to assess the robustness. The 7nm FinFET device was then employed into 6T SRAM cells and 16 function ALU. The SRAM cells were approached with advanced multi-corner stability evaluation. The system-level architecture of the ALU has demonstrated an ultra-low power system operating at 1 GHz clock frequency.

No description available.

This research details the design of an SRAM compiler for quickly creating SRAM blocks for Cal Poly integrated circuit (IC) designs. The compiler generates memory for two process technologies (IBM 180nm cmrf7sf and ON Semiconductor 600nm SCMOS) and requires a minimum number of specifications from the user for ease of use, while still offering the option to customize the performance for speed or area of the generated SRAM cell. By automatically creating SRAM arrays, the compiler saves the user time from having to layout and test memory and allows for quick updates and changes to a design. Memory compilers with various features already exist, but they have several disadvantages. Most memory compilers are expensive, usually only generate memory for one process technology, and don’t allow for user-defined custom SRAM cell optimizations. This free design makes it available for students and institutions that would not be able to afford an industry-made compiler. A compiler that offers multiple process technologies allows for more freedom to design in other processes if needed or desired. An attempt was made for this design to be modular for different process technologies so new processes could be added with ease; however, different process technologies have different DRC rules, making that option very difficult to attain. A customizable SRAM cell based on transistor sizing ratios allows for optimized designs in speed, area, or power, and for academic research. Even for an experienced designer, the layout of a single SRAM cell (1 bit) can take an hour. This command-line-based tool can draw a 1Kb SRAM block in seconds and a 1Mb SRAM block in about 15 minutes. In addition, this compiler also adds a manually laid out precharge circuit to each of the SRAM columns for an enhanced read operation by ensuring the bit lines have valid logic output values. Finally, an analysis on SRAM cell stability is done for creating a robust cell as the default design for the compiler. The default cell design is verified for stability during read and write operations, and has an area of 14.067 µm2 for the cmrf7sf process and 246.42 µm2 for the SCMOS process. All factors considered, this SRAM compiler design overcomes several of the drawbacks of other existing memory compilers.