Fact-checked by Grok 2 weeks ago

Antifuse

An antifuse is an electrical device that operates inversely to a traditional fuse, initially exhibiting high resistance in an open-circuit state and transitioning to a low-resistance conductive state through dielectric breakdown when a high voltage is applied, enabling permanent one-time programmability.^[1] Structurally, it consists of two conductive electrodes separated by a thin insulating dielectric layer, such as silicon dioxide or amorphous silicon, which ruptures under voltage stress to form a filamentary conductive path.^[2] Antifuses have roots in non-electronic applications like lighting before their adoption in integrated circuits. Antifuse technology emerged in the late 1980s as a solution for programmable integrated circuits, with Actel Corporation pioneering its development and shipping the first antifuse-based field-programmable gate arrays (FPGAs) in 1988, such as the ACT 1010 model, which offered up to 1,010 gates.^[3]^[4] These devices utilize antifuses as interconnect switches within the FPGA fabric, allowing users to configure logic functions by selectively programming connections without the need for volatile memory like SRAM.^[5] The programming process involves applying a voltage pulse, typically around 18 V for early devices, to induce breakdown in the dielectric, resulting in a stable on-resistance of less than 100 ohms.^[4] Beyond FPGAs, antifuses serve as one-time programmable (OTP) non-volatile memory elements in various applications, including redundancy repair in dynamic random-access memory (DRAM), secure key storage for cryptography, and configuration in Internet of Things (IoT) devices and mobile payment systems.^[6] Their advantages include inherent security due to the lack of visible physical changes post-programming, low static power consumption in the unprogrammed state, and compatibility with standard CMOS processes without requiring additional masks.^[7] Compared to alternatives like eFuses, antifuses offer higher programming yields and resistance to reverse engineering, making them suitable for high-reliability environments such as aerospace and automotive electronics.^[8]

Overview

Definition

An antifuse is an electrically programmable two-terminal device that functions as an open circuit with high initial resistance until programmed by applying a sufficient voltage or current, at which point it irreversibly forms a low-resistance conductive path.^[9] Unlike a traditional fuse, which begins as a low-resistance conductor and permanently opens upon exposure to excessive current to protect circuits, an antifuse performs the opposite operation by transitioning from non-conductive to conductive.^[10] The general structure of an antifuse consists of two conductive electrodes separated by an insulating dielectric material, such as silicon dioxide or silicon nitride, which acts as a barrier in the unprogrammed state. Upon programming, the applied voltage causes dielectric breakdown, creating a conductive filament through the insulator that establishes a permanent connection between the electrodes.^[11] This one-time programmable (OTP) characteristic renders the antifuse non-volatile, as the programmed state persists without power and cannot be reversed, making it ideal for establishing permanent electrical interconnections in integrated circuits.

Operating Principle

An antifuse functions as the electrical inverse of a conventional fuse, starting in a high-impedance off-state and irreversibly switching to a low-impedance on-state when programmed.^[12] The core operating principle of an antifuse relies on the application of a high voltage pulse, typically ranging from 5 to 15 V depending on the device type and insulating layer thickness, to initiate programming. This voltage, often accompanied by a programming current of approximately 5 to 200 mA, induces either dielectric breakdown in the insulating layer or a thermally driven phase change, creating a permanent conductive filamentary path between the terminals. In dielectric-based antifuses, the mechanism involves progressive defect generation and percolation under the intense electric field, leading to catastrophic failure of the insulator and formation of a conductive channel. For amorphous material-based variants, the high current density generates localized heating, melting the insulator and promoting crystallization into a low-resistance structure upon cooling. The process is modeled statistically using time-dependent dielectric breakdown (TDDB) frameworks, where the time to breakdown follows a power-law relation T_{BD} = \alpha \cdot V^{-n}, with n typically between 43 and 46, reflecting voltage acceleration of defect formation.^[13]^[14]^[15] Post-programming, the conductive path exhibits a low on-state resistance, generally in the range of 10 to 100 ohms, enabling efficient signal propagation while remaining stable without continuous power supply. In the unprogrammed off-state, the antifuse presents an extremely high resistance exceeding $10^9 ohms, ensuring negligible leakage, and an on-state capacitance below 1 pF to minimize parasitic effects in high-speed circuits. The breakdown voltage for the insulating layer in dielectric antifuses can be approximated by the relation V_{breakdown} = E_{field} \times d, where E_{field} is the material's dielectric strength (often 5-10 MV/cm for thin oxides) and d is the insulator thickness, providing a simplified model for design predictions.^[15]^[16]^[17] The irreversible nature of antifuse programming stems from fundamental physical alterations, such as irreversible filament formation via defect percolation in dielectric breakdown or permanent crystallization in phase-change mechanisms, ensuring long-term retention without degradation under normal operating conditions. Reliability is enhanced by the one-time-use design, with post-breakdown paths showing minimal resistance drift over time and temperature, though careful control of programming parameters is essential to avoid incomplete linking or overstress.^[13]^[14]

Historical Development

Early Uses in Lighting

The concepts underlying antifuses first appeared in early 20th-century electrical engineering through mechanisms designed to maintain circuit continuity in series-connected incandescent lighting systems, particularly after a bulb failure disrupted the chain. In these setups, which powered multiple lamps from a single constant-current source, a failed lamp would otherwise open the entire circuit, extinguishing all lights; to counter this, devices known as film cutouts were integrated into lamp sockets to automatically shunt the faulty lamp by creating a permanent low-resistance path. These cutouts, introduced around 1900, consisted of a thin insulating film—often cellulose acetate or similar material—positioned between two contacts in the socket; upon lamp burnout, the resulting voltage surge across the socket heated and punctured or melted the film, shorting the circuit and allowing current to bypass the dead lamp while reducing overall illumination slightly.^[18]^[19]^[20] Such antifuse-like devices saw widespread adoption in urban street lighting grids across the United States and Europe during the 1920s to 1950s, especially in high-intensity discharge (HID) lamp series circuits like those using mercury vapor technology. In these systems, shunts integrated into the ballasts or special lamp designs with shorting provisions permanently bypassed failed lamps upon open-circuit failure, maintaining circuit continuity and preventing total outages in long strings of lights that could span miles. For instance, early mercury vapor installations from the 1930s onward relied on series operation with constant-current regulators, where the ballast or dedicated shunting device would activate to bypass failed lamps, ensuring continued operation of the remaining fixtures at a marginally lower voltage per lamp. This approach was particularly valuable for cost-effective illumination of highways and city streets, with examples including General Electric's Type H lamps deployed in series configurations post-1936.^[21]^[22]^[19] These shunting mechanisms began to be phased out starting in the 1960s as parallel-wired systems with individual ballasts for each lamp gained adoption, offering greater flexibility, easier maintenance, and compatibility with emerging high-pressure sodium and metal halide technologies. The transition eliminated the need for series shunting devices, as independent operation per fixture reduced vulnerability to single-point failures, though it increased wiring complexity and initial installation costs. For instance, cities like Milwaukee initiated systematic replacements in the late 1960s but retained a significant portion of series setups into the 2000s.^[19]^[23]^[24]

Adoption in Integrated Circuits

The adoption of antifuses in integrated circuits began in the mid-1980s, marking a shift from earlier conceptual and lighting applications to semiconductor-based programmable logic. Actel Corporation, founded in 1985, pioneered the first commercial antifuse-based field-programmable gate arrays (FPGAs), shipping its initial products in 1988. These early devices, such as the ACT family, utilized antifuse technology to enable reliable, one-time programmable interconnects, addressing limitations in volatile SRAM-based alternatives prevalent at the time.^[3]^[25] A key milestone in this adoption was Actel's development of metal-to-metal antifuse interconnects, which allowed for dense, low-resistance routing in FPGAs without requiring volatile memory for configuration retention. This patented approach, introduced in the late 1980s and refined through the 1990s, provided superior security and radiation hardness compared to reprogrammable technologies, making antifuses ideal for high-reliability applications. By enabling permanent, low-power connections post-manufacturing, these interconnects facilitated the integration of antifuses into complex IC designs, reducing power consumption and enhancing performance in gate arrays up to thousands of logic gates.^[26]^[27] During the 1990s, antifuse technology expanded beyond FPGAs into programmable read-only memories (PROMs), programmable logic devices (PLDs), and redundancy repair mechanisms in dynamic random-access memories (DRAMs). This period saw widespread implementation for post-package repairs, where defective memory cells could be bypassed using programmed antifuses, improving yield in high-density IC production. Influential innovations included early patent filings like US Patent 5,110,754 (1992), which described a DRAM capacitor structure adapted as a programmable antifuse for redundancy options. By the late 1990s, radiation-tolerant antifuse variants emerged for space applications, with Actel introducing its RT product line in 1998 to meet demands for single-event upset immunity in aerospace environments.^[28]^[29] In the 2000s and 2010s, antifuses gained prominence in one-time programmable (OTP) non-volatile memory for security-critical functions, such as storing cryptography keys in secure ICs, due to their resistance to hacking and non-volatility without power. This integration supported applications in embedded systems requiring tamper-proof data retention. The market for antifuse-based FPGAs has continued to grow, driven by demand in defense, aerospace, and automotive sectors, with projections indicating a compound annual growth rate (CAGR) of approximately 10.5% from 2025 to 2032.^[30]^[31]

Types of Antifuses

Dielectric Antifuses

Dielectric antifuses consist of two conductive electrodes, typically metal or polysilicon layers, separated by a thin insulating dielectric material, such as silicon dioxide (SiO₂) or an oxide-nitride-oxide (ONO) stack. The dielectric layer has an electrical thickness equivalent to approximately 9 nm of SiO₂, enabling compact structures suitable for high-density integration in sub-micron semiconductor processes. In the unprogrammed state, this configuration provides excellent isolation with extremely low leakage currents, on the order of nanoamperes, making it ideal for interconnect applications in integrated circuits.^[32] Programming occurs when a high-voltage pulse is applied across the electrodes, initiating Fowler-Nordheim tunneling of electrons through the thin dielectric, which generates hot carriers and leads to localized avalanche breakdown. This process creates a conductive silicon filament that bridges the electrodes, transforming the high-resistance insulator into a low-resistance conductor. Typical programming voltages range from 18-21 V in early implementations, with pulse durations of 150-300 μs applied in multiple cycles if needed, though optimized designs achieve completion in under 1 ms total. The filament formation ensures a permanent, one-time programmable connection without requiring additional heating elements.^[13]^[33]^[32] Once programmed, dielectric antifuses exhibit an on-state resistance of 50-500 Ω, depending on the specific electrode materials and process node, with via-link variants achieving lower values around 50-250 Ω for minimal signal delay. This low resistance, combined with a programming time under 1 μs in advanced configurations, supports high-speed reconfiguration in dense arrays. A representative application is in Actel's early field-programmable gate arrays (FPGAs), such as the ACT family, where via-link dielectric antifuses interconnect metal layers to form customizable routing between logic modules.^[32]^[34] These antifuses offer robust reliability, with the conductive filament providing resistance to electromigration due to its localized, non-metallic nature, unlike traditional metal interconnects. Post-programming failure rates are below 1 ppm, with projected lifetimes exceeding 40 years at operating voltages up to 5.5 V and temperatures of 125°C, as demonstrated in accelerated life testing with no observed failures over 1.8 million device-hours.^[32]

Amorphous Silicon Antifuses

Amorphous silicon antifuses feature a layered structure consisting of a thin undoped amorphous silicon (a-Si) dielectric sandwiched between two metal electrodes, often incorporating barrier metals such as titanium nitride to inhibit diffusion and ensure isolation.^[35]^[36] This configuration is typically integrated into metal layers of integrated circuits, with the a-Si layer deposited over a conductive plug like tungsten for vertical interconnects.^[35] The design leverages the high resistivity of amorphous silicon in its unprogrammed state to provide effective insulation between routing channels.^[37] Programming occurs irreversibly when a voltage pulse of 12-15 V, accompanied by a current of 15-25 mA, is applied across the electrodes, generating localized Joule heating within the a-Si layer.^[36] This thermal process drives a phase transition, crystallizing the amorphous silicon into polycrystalline form and facilitating metal silicidation or filamentary alloy growth that bridges the electrodes with a stable conductive path.^[36]^[35] The resulting link forms a linear resistor without catastrophic breakdown, contrasting with other antifuse types by relying on controlled thermal transformation rather than dielectric rupture.^[35] Post-programming, these antifuses demonstrate low on-state resistance, typically around 25 Ω, enabling efficient signal propagation in interconnects.^[38] Unprogrammed devices maintain leakage currents below 1-10 nA at operating voltages, with parasitic capacitance per antifuse on the order of 1-3 fF, minimizing loading effects in high-speed circuits.^[38]^[36] They have been widely employed in programmable logic devices, including QuickLogic's ViaLink technology and older Actel FPGAs, for routing channels between logic modules.^[35]^[37] Reliability is enhanced by the permanent, charge-free nature of the programmed filament, which exhibits negligible resistance variation across temperatures up to 150-250°C and withstands over 10^17 switching cycles equivalent to a decade of high-frequency operation.^[36]^[35] Additionally, their immunity to radiation-induced soft errors stems from the lack of volatile storage elements, making them suitable for harsh environments.^[36] Accelerated testing confirms lifetimes exceeding 40 years under typical conditions, with breakdown voltages consistently in the 8-11 V range for uniform programming.^[36]

Zener Antifuses

Zener antifuses are programmable devices based on Zener diodes, typically implemented as a PN junction with an avalanche region or back-to-back Zener diodes in bipolar or BiCMOS processes. The structure commonly features a circular N+ cathode diffusion, 1-3 µm deep with sheet resistance of 5-10 Ω/sq, overlapping a P-type anode diffusion, 3-5 µm deep with 100-200 Ω/sq, forming a subsurface lateral junction.^[1] In CMOS variants, the device uses P+ source/drain as the anode and N+ source/drain as the cathode within a P-well.^[1] Programming occurs by applying a high-current pulse to the reverse-biased junction, triggering avalanche breakdown and thermal runaway that leads to localized heating and metal electromigration across the junction. This forms a low-resistance conductive path through spiking or filamentation, irreversibly shorting the device. The pulse typically consists of 100-200 mA for 1-5 ms in conventional bipolar implementations, though optimized structures can reduce current to around 50 mA; a two-phase approach—initial high peak current followed by sustained lower current—ensures reliable fusing.^[1] Post-programming, the path exhibits a resistance of approximately 10 Ω.^[1] These antifuses have a pre-programming breakdown voltage of 6-7 V, set by the P-material doping level, making them suitable for precision analog trimming where gradual resistance adjustments are needed. By integrating multiple Zener antifuses into binary-weighted resistor networks, they enable fine control over circuit parameters, such as shunting specific resistors to achieve desired values with resolutions around 125 Ω and ranges up to 7875 Ω.^[1] A key application is in analog integrated circuits for tasks like operational amplifier offset voltage adjustment or voltage reference fine-tuning, serving as a non-laser alternative for post-fabrication calibration to boost yield in mixed-signal VLSI.^[1] Despite their effectiveness, Zener antifuses demand higher power during programming—potentially up to 30 V and 240 mA in driver circuits—posing challenges for on-chip integration without robust support electronics. Additionally, their size and bond pad requirements limit scalability to lower densities compared to advanced digital antifuse technologies.^[1]

Applications in Electronics

Programmable Logic Devices

Antifuses are employed in programmable read-only memories (PROMs) to program address lines or data bits, allowing one-time programmable (OTP) memory that functions as custom read-only memory (ROM) without requiring custom photomasks during fabrication.^[39] This approach enables cost-effective customization of memory contents post-manufacturing, where antifuses in each bit cell are selectively programmed by applying a high voltage to create a permanent low-resistance path, contrasting with traditional mask-programmed ROMs that demand upfront design commitments.^[39] In programmable logic devices (PLDs), antifuses serve as permanent vias to customize logic arrays by establishing fixed interconnections between gates, as seen in early desktop-configurable channeled gate arrays where they facilitate one-time logic configuration.^[39] Dielectric antifuses, valued for their high density, are particularly suited for such applications due to their compact size and reliability in forming stable connections.^[39] This one-time programming capability provides a secure, non-volatile alternative to fusible links, ensuring tamper-resistant logic implementations in simpler PLD architectures. Antifuses also play a critical role in redundancy repair for memory chips, such as dynamic random-access memories (DRAMs), where they bypass defective rows or columns identified after manufacturing by rerouting signals to spare elements. This post-package repair technique significantly enhances production yields—for instance, one-bit repairs can improve net yield by up to 2.4% in 0.16-μm processes—by salvaging otherwise unusable dies through targeted defect isolation.^[28] Programming occurs via on-chip high-voltage generators that deliver precise voltage pulses during final testing, enabling field-programmable or package-level corrections without external equipment.

Field-Programmable Gate Arrays

In field-programmable gate arrays (FPGAs), antifuses serve as the core programmable interconnect elements, forming switches within routing channels to connect configurable logic blocks and input/output pads. Unlike SRAM-based FPGAs, which rely on volatile memory cells for routing configuration, antifuse-based designs employ one-time programmable (OTP) switches that create permanent low-resistance connections upon programming, eliminating the need for external configuration memory and enabling non-volatile operation. This architecture, pioneered by Actel (now part of Microchip Technology), utilizes a sea-of-modules layout where antifuses are integrated directly into the interconnect fabric, providing predictable routing delays and reduced susceptibility to single-event upsets in radiation environments.^[40]^[26] A representative example of antifuse FPGA architecture is Actel's SX and Axcelerator families, which incorporate via-hole antifuses positioned between metal layers (typically Metal 2 and Metal 3) in a multi-layer CMOS process. These metal-to-metal antifuses consist of a thin dielectric layer that breaks down under a high-voltage programming pulse, forming a conductive path with resistance as low as 25-50 ohms, which minimizes signal propagation delays compared to pass-transistor switches in SRAM FPGAs. The compact size of each antifuse, which is significantly smaller than that of an SRAM-based switch, contributes to overall die area efficiency, allowing for denser logic integration without the overhead of distributed memory cells. Actel's historical adoption of this technology in the late 1980s established antifuse FPGAs as a reliable alternative for applications requiring permanence and security.^[40]^[41]^[42] Antifuse FPGAs find critical applications in space and security domains due to their radiation tolerance and inherent protection against reverse engineering. Since the 1990s, NASA has utilized Actel's antifuse devices, such as the A1020, in space missions for their immunity to total ionizing dose and single-event effects, as the programming elements lack charge-storage mechanisms vulnerable to radiation. In security contexts, the OTP nature of antifuses stores the configuration bitstream on-chip without external loading, preventing interception or cloning during power-up and offering resistance to side-channel attacks like power analysis. This makes them suitable for high-reliability sectors, including aerospace and defense, where design integrity is paramount.^[42]^[29]^[43] Performance-wise, antifuse FPGAs support high-speed operation with internal clock frequencies exceeding 500 MHz and system performance up to 350 MHz, facilitated by low-interconnect resistance and the absence of reconfiguration overhead. Configuration occurs instantly upon power-on, as the non-volatile antifuses retain the programmed state without boot-time delays typical of SRAM or flash alternatives. These devices maintain a niche presence in high-reliability markets, driven by demand in radiation-hardened and secure applications.^[44]^[45]

Applications in Consumer Products

Christmas Tree Lights

In miniature incandescent Christmas tree lights, bulbs are connected in series across standard household voltage, typically 120 V in North America, with each bulb designed to operate at 2.5–3.5 V. To maintain circuit continuity when a filament burns out, each bulb incorporates an internal antifuse shunt consisting of a thin wire coated in a low-melting-point insulating material, such as lacquer or enamel, wrapped around or near the filament support. Under normal operation, current flows through the low-resistance filament, but if the filament fails and opens the circuit, the full line voltage appears across the shunt, causing the insulation to melt and the wire to short-circuit the bulb permanently, bypassing the failure and preventing the entire string from darkening.^[46] This "shunt-type" design, introduced in the 1960s, significantly improved reliability for consumer holiday lighting by protecting against open-circuit failures from single bulb burnout, a common issue in series configurations that previously caused total string outage. The technology was patented in a 1964 filing for an electric shunt device specifically for series-connected decorative lamps like Christmas tree lights, enabling the shunt to activate via insulation breakdown under high voltage. These shunts became standard in Underwriters Laboratories (UL)-listed light strings, reducing user frustration during the holiday season by allowing the majority of bulbs to remain lit despite occasional failures.^[47] For added safety, many strings include parallel fuse bulbs at the ends or in the plug assembly, which are ordinary incandescent bulbs rated to blow under overcurrent conditions, such as when multiple shunts activate simultaneously and increase the overall current draw. This integration prevents excessive current from overheating wires or risking fire, as each activated shunt reduces string resistance and elevates voltage across remaining bulbs, potentially accelerating their burnout. The permanent shorting principle of these antifuses ensures one-time activation without reset, maintaining long-term circuit integrity.^[48]

Historical Street Lighting

In the early 20th century through the 1950s, series circuits were widely employed for powering incandescent street lamps in urban areas, operating on constant current systems typically rated at 6.6 amperes to distribute power efficiently across multiple fixtures along long runs.^[19]^[18] These systems relied on thermal antifuses, such as film cutouts integrated into lamp sockets, to maintain circuit continuity when individual bulbs failed; upon outage, the sudden voltage surge across the defective lamp—often exceeding 2,000 volts—caused the insulating film in the cutout to vaporize or arc, permanently shorting the contacts and shunting current around the failure without interrupting the rest of the series.^[49]^[50] Designs like the Thomson and Jones film cutout sockets, introduced around 1900, exemplified this mechanism, where the antifuse operated as a one-time device triggered by electrical imbalance rather than overcurrent.^[49]^[18] This approach offered significant advantages in the era of expanding city infrastructure, particularly by minimizing wiring costs through a single-pair conductor system that spanned miles of urban streets, thereby saving on copper and simplifying installation compared to parallel circuits requiring multiple feeds per lamp.^[19] Additionally, the shunting action ensured partial illumination persisted even with multiple failures, avoiding total blackouts and supporting reliable nighttime visibility for public safety.^[19] Such systems were commonplace in major U.S. cities, including New York, where Edison's early municipal installations from the 1880s evolved into widespread series networks that remained in use until post-World War II electrification upgrades began favoring more flexible grid designs.^[18] By the 1970s, these series circuits with antifuse shunts were largely phased out in favor of high-intensity discharge lamps like mercury vapor and sodium vapor types, which operated on individual transformers and parallel wiring for better energy efficiency and easier maintenance.^[19]^[21] The transition accelerated urban modernization, though remnants persisted in remote or rural areas into the 1980s before full replacement by contemporary lighting technologies.^[19]

Performance Characteristics

Advantages

Antifuses provide high security and tamper resistance due to their one-time programmable (OTP) nature, which permanently alters the device through dielectric breakdown, preventing reconfiguration. This makes them suitable for storing secure keys or bitstreams, such as in field-programmable gate arrays (FPGAs) where anti-cloning measures are critical, though advanced failure analysis techniques can potentially extract data despite difficulties in detection with standard scanning electron microscopy.^[51]^[52]^[7]^[53] In harsh environments, antifuses exhibit high reliability, particularly immunity to radiation-induced soft errors that affect SRAM-based configurations, with no susceptibility to single-event upsets (SEUs) or alpha particle disruptions. This radiation tolerance, with SEU linear energy transfer (LET) thresholds exceeding 37 MeV·cm²/mg, has established their use in space and military applications, where devices like Microchip's RTAX series maintain configuration integrity without mitigation schemes required for volatile alternatives.^[54]^[55]^[56] Performance-wise, antifuses offer low on-state resistance, typically below 50 ohms (e.g., around 25 ohms in commercial implementations), and capacitance under 1 fF, minimizing parasitic effects to support high-speed signal propagation up to 270 MHz in system clocks without additional power for data retention. These characteristics enable efficient, low-power operation in programmable logic, surpassing the delays introduced by higher parasitics in alternative interconnect technologies.^[57]^[26]^[58] Antifuses achieve cost efficiencies and higher density through a smaller footprint compared to traditional laser fuses, occupying less silicon area while scaling effectively in advanced nodes like 28nm and below, including sub-10 nm processes with modern implementations such as I-fuse, which reduces overall die costs. This compact design facilitates post-fabrication customization, allowing device programming without the need to respin expensive masks, a key advantage for flexible manufacturing in secure or specialized electronics.^[8]^[59]^[51]^[60] Unlike SRAM or flash-based FPGAs that require configuration loading on power-up, antifuse devices provide instant configuration with no boot-up delay, powering on in a fully operational state for immediate functionality in time-critical systems. This "instant-on" capability eliminates the need for external non-volatile memory and reduces latency, enhancing suitability for applications demanding rapid activation.^[61]^[34]

Limitations

Antifuse technology is inherently one-time programmable (OTP), meaning once an antifuse is programmed by forming a permanent conductive path, it cannot be erased or modified, which severely limits its use in iterative prototyping and design verification compared to reprogrammable alternatives like SRAM- or Flash-based field-programmable gate arrays (FPGAs).^[5]^[7] This irreversibility requires final designs to be fully validated before programming, often necessitating external programming services and increasing development timelines.^[5] Programming an antifuse demands precise application of high voltages, typically in the range of 5-15 V above the core supply, along with controlled current pulses to induce dielectric breakdown without compromising reliability.^[62]^[15] This process introduces complexity, as imprecise control can lead to over-programming, resulting in poor filament formation, increased resistance, or damage to adjacent structures through unintended program disturbs.^[62] Programming yields for antifuse arrays typically average 98-99%, but variations across production lots can occur, necessitating built-in redundancy and error-correcting codes to mitigate defects and ensure functionality.^[63] These measures, combined with rigorous pre- and post-programming testing, elevate initial validation costs compared to non-OTP technologies, where a single defect can propagate a 10x yield penalty in high-volume manufacturing.^[64] While historical challenges with variability in oxide breakdown limited scalability below 10 nm, modern antifuse implementations have achieved reliable operation in sub-10 nm nodes (e.g., 7 nm and 5 nm) through optimized designs, though reprogrammable options continue to dominate consumer-grade FPGAs due to flexibility needs.^[65]^[60] During programming, each antifuse consumes 10-100 mW due to the high-voltage pulses required for breakdown, making the process energy-intensive and impractical for battery-powered or low-power devices with large arrays.^[62] While operational power remains low post-programming, this programming overhead limits applications in power-constrained environments.^[7]

References

[1]
[PDF] Zener Zap Anti-Fuse Trim in VLSI Circuits - Semantic Scholar
which do not lend themselves to laser trimming. The term anti-fuse is used to describe an element which initially appears as an open circuit but can be made to ...
[2]
Long term storage reliability of antifuse field programmable gate arrays
Aug 10, 2025 · The antifuse is a one-time programmable (OTP) cell, and it consists of two electrodes and a thin dielectric film as the insulating layer ...
[3]
[PDF] AN 4535 : Programming Antifuse Devices - Microchip Technology
Antifuse technology is nonvolatile, therefore, it is live at power-up and inherently very secure. For more information on security types and implementation, ...Missing: definition | Show results with:definition
[4]
Metal-to-metal antifuse with low programming voltage and low on ...
Aug 6, 2025 · In addition, the antifuse technology enables the storage of secured information with respect to cryptography or else. The thesis focuses on ...
[5]
Reliable Antifuse One-Time-Programmable Scheme With Charge ...
Aug 7, 2025 · An antifuse EPROM and 3-V programming circuit has been demonstrated in an existing 0.22-μm DRAM process technology and is fully compatible with ...
[6]
The Benefits Of Antifuse OTP - Semiconductor Engineering
Dec 19, 2016 · In summary, an OTP memory with antifuse has better yield for programming, lower power consumption for unprogrammed bits, and overall superior ...
[7]
Review of fuse and antifuse solutions for advanced standard CMOS ...
Aug 14, 2025 · ... The antifuse is a one-time programmable (OTP) cell, and it consists of two electrodes and a thin dielectric film as the insulating layer ...
[8]
https://www.researchgate.net/publication/223234675_Review_of_fuse_and_antifuse_solutions_for_advanced_standard_CMOS_technologies
[9]
[PDF] nasa handbook nasa-hdbk 8739.23a
Feb 2, 2016 · Antifuse: An electrical device that performs the opposite function as a fuse. Antifuses are widely used to permanently program integrated ...
[10]
[PDF] Low Temperature Plasma Enhanced CVD for Antifuse Technology
The structure of the antifuse device studied here is three layers, metal- dielectric-metal. To make a permanent connection between the two metal electrodes, a ...
[11]
MOSFET anti-fuse structure and method for making same
Programming of the anti-fuse is performed by application of power to the gate and at least one of the source region and the drain region to break-down the gate ...
[12]
[PDF] Analysis of ultrathin gate-oxide breakdown mechanisms and ...
Mar 6, 2015 · physics of dielectric breakdown, the high voltage amplitude used in antifuse memo- ries is not in the range of their study as well as the ...Missing: formula | Show results with:formula
[13]
[PDF] One Time Programmable Antifuse Memory Based on Bulk ...
The breakdown occurs around WL=4.4 V and BL=0V at selected cell whereas. BL=2.5V suppresses the electric field to inhibit the programming to the unselected cell ...
[14]
https://ntrs.nasa.gov/api/citations/20190002597/downloads/20190002597.pdf
[15]
[PDF] Dielectric insulation and high-voltage issues
The maximum electric field achievable in a dielectric without the occurrence of an electrical breakdown is called dielectric strength, typically expressed in kV ...
[16]
[PDF] history of electric light - Smithsonian Institution
Open Flame Arc Lamp, 1898. 66. Enclosed Flame Arc Lamp, 1908. 66. Constant Current Transformer, 1900. 68. Series Incandescent Lamp Socket with Film Cutout, 1900.
[17]
Incandescent Lighting - Series Operation - Vintage Streetlights
1930s series streetlighting lamps. Street series lamps are designed to operate in series on most constant current circuits. A typical constant-current circuit ...
[18]
history of electric light - Project Gutenberg
Series Incandescent Lamp Socket with Film Cutout, 1900, 70. Nernst Lamp, 1900, 71. Diagram of Nernst Lamp, 72. Cooper-Hewitt Mercury Vapor Arc Lamp, 1901, 73.
[19]
Mercury-Vapor Streetlighting and Lamps - Light Sources
Among the earliest mercury lamp types was the "Type H". The discharge tube was made of aluminosilicate glass and operated within a tubular outer bulb, two ...
[20]
Early Street Light Systems, P.3 - KBR Horse Net
In series circuits The lamp ballasts also served as shunts that preserved continuity when mercury vapor lamps failed. Although mercury vapor lamps were more ...
[21]
[PDF] Street Lighting in Milwaukee:
Currently, most of Milwaukee's street light circuits are series circuits. Since the late 1960s, however, the City has been replacing series circuits with ...Missing: decline | Show results with:decline
[22]
Preserving Historic Street Lights, P.2 - KBR Horse Net
When providing 110 volt service to series luminaires that have removable Jones sockets, it is important to replace the film cutout disk in the socket bayonet ...
[23]
Actel - Wikipedia
History and competition Actel was founded in 1985 and became known for its high-reliability and anti-fuse-based FPGAs, used in the military and aerospace ...Missing: commercial | Show results with:commercial
[24]
How the FPGA Came To Be, Part 6: Actel's FPGA Story - EEJournal
Jul 22, 2024 · Actel shipped its first FPGAs in 1988. Xilinx had been shipping parts that the company called “Logic Cell Arrays,” based on SRAM programmability ...
[25]
From AMD to Actel (to Microchip) - SemiWiki
Sep 2, 2019 · I started at Actel in 1988. By the mid 90s my love affair with antifuses had all but ended. It had become clear to me that the SRAM ...
[26]
[PDF] Actel's Antifuse FPGAs
Ideal for integrating logic typically imple- mented in multiple CPLDs, PALs, and FPGAs, antifuse devices offer significant cost savings while maintaining high ...
[27]
[PDF] SX-A Family FPGAs Datasheet - Microchip Technology
Note: The A54SX72A device has four layers of metal with the antifuse between Metal 3 and Metal 4. ... against Actel antifuse FPGAs. Look for this symbol to ensure ...
[28]
Antifuse EPROM circuitry scheme for field-programmable repair in ...
Aug 9, 2025 · A highly reliable antifuse cell and its sensing scheme that can be actually adopted in DRAM are presented. A multi-cell structure is newly ...Missing: PLDs 1990s
[29]
[PDF] radiation tolerant antifuse fpga
The isolation transistor functions as a turn-off pass gate to isolate high voltage programming voltage (typically > 2Vcc) from low voltage (Vcc) transistors in ...
[30]
OTP Non-Volatile Memory Advantages | Synopsys IP
Oct 22, 2018 · AntiFuse OTP NVM replaces floating-gate or eFuse technologies for tasks such as secure key storage, device IDs, analog/sensor trimming and ...
[31]
Antifuse FPGA Market Analysis and Growth Projections for 2032 ...
The global Antifuse FPGA Market is projected to exhibit a robust compound annual growth rate (CAGR) of 10.5% from 2025 to 2032.
[32]
[PDF] ACT™ Family Gate Array - Bitsavers.org
... Dielectric-Based Antifuse for Logic and Memory ICs ... Hamdy, et aI, "Dielectric Based Antifuse for Logic and. Memory ICs", IEDM paper, p. 786-789,1988. 2 ...
[33]
Antifuse structure having an integrated heating element
The present invention provides antifuse structures having an integrated heating element and methods of programming the same, the antifuse structures ...
[34]
Antifuse field programmable gate arrays - Semantic Scholar
Antifuse field programmable gate arrays · J. Greene, E. Hamdy, S. Beal · Published in Proceedings of the IEEE 1 July 1993 · Engineering, Computer Science.
[35]
[PDF] Reliability of the Amorphous Silicon Antifuse | QuickLogic Corporation
The ViaLink's leakage mechanism is not catastrophic at the voltage levels seen in use. Programmed ViaLink Characteristics. The programmed ViaLink (see Figure 4) ...Missing: FPGA | Show results with:FPGA
[36]
[PDF] Actel's Quality and Reliability Guide
An antifuse structure is much more compact. Antifuses typically consist of a thin insulating layer between contact points that begins conducting under ...
[37]
US4758745A - User programmable integrated circuit interconnect ...
Another approach uses an array of uncommitted interconnect metal lines using anti-fuses consisting of an amorphous silicon alloy sandwiched into insulation ...
[38]
[PDF] eX Family FPGAs
Interconnection among these logic modules is achieved using Actel's patented metal-to-metal programmable antifuse interconnect elements. The antifuse.
[39]
https://ieeexplore.ieee.org/document/32929
[40]
https://ww1.microchip.com/downloads/aemdocuments/documents/fpga/ApplicationNotes/ApplicationNotes/low_power_an.pdf
[41]
[PDF] Design for Low Power in Actel Antifuse FPGAs - Microchip Technology
A special low-power pin is available on the eX devices as an additional means to reduce power. It shuts off all internal charge pumps, reducing static current ...
[42]
[PDF] Radiation Effects in Antifuse FPGA - CERN Document Server
Programming a Flash switch needs high voltage, approximately 17 V. Antifuse switch is programmed at a high voltage scaled with the operational VCC, about three.<|control11|><|separator|>
[43]
[PDF] Design and Test of Field Programmable Gate Arrays in Space ...
Of the FPGAUs available in 1990, only the 2000 gate Actel A1020, using antifuse technology, had demonstrated adequate radiation tolerance to make it a ...
[44]
[PDF] Implementation of Security in Microsemi Antifuse FPGAs
Microsemi nonvolatile antifuse FPGAs do not require a start-up bitstream, eliminating the possibility of configuration data being intercepted or copied.Missing: OTP | Show results with:OTP
[45]
Axcelerator FPGAs - Microchip Technology
Axcelerator, our antifuse family of FPGAs, offers high performance and unprecedented design security at densities of up to 2 million equivalent system gates.Missing: instant configuration
[46]
Actel to base high-speed FPGAs on antifuse technology - EE Times
Actel attributes the higher speed to its unique routing system. The AX architecture contains 53 million antifuse switches, seven times more switches than ...Missing: history commercial 1985
[47]
Antifuse-based Field Programmable Gate Array (FPGA) Market Size ...
Antifuse-based Field Programmable Gate Array (FPGA) Market size is estimated to be USD 1.2 Billion in 2024 and is expected to reach USD 2.
[48]
How Do Holiday Lights Work? | Department of Energy
Dec 16, 2015 · In incandescent holiday lights, shunts are small wires wrapped beneath the filament.
[49]
US3345482A - Electric shunt device - Google Patents
ELECTRIC SHUNT DEVICE Filed June 29, 1964 3 Sheets-Sheet 2 Oct. ... SHUNT DEVICE S Sheets-Sheet 5 United States Patent ,0 ... Christmas tree. To overcome such ...
[50]
Christmas Lights and How to Fix Them - Ciphers By Ritter
Nov 27, 2005 · Each activated shunt reduces the string resistance and causes the string current to increase, which shortens the life of the remaining bulbs. If ...
[51]
LED Christmas Lights and How to Fix Them - Ciphers By Ritter
May 24, 2007 · Shorting upon failure conveniently takes the place of the "shunts" added to series string incandescent mini bulbs, because those shunts often ...
[52]
Early Street Light Designs, P.2
Designs such as the Thomson and Jones film cutout sockets came into use. In these designs, when a lamp failed, the series voltage (typically less than 50 volts ...Missing: antifuses bimetal
[53]
Film Disc Cutouts - Meridian Electric
A film disc cutout, sometimes called an antifuse, can be thought of as a backward acting fuse. It blows on over voltage, not over current.
[54]
The Case For Antifuse OTP NVM For Secure & Reliable SoCs
Jan 10, 2019 · This article describes the advantages of antifuse OTP NVM as the most secure and reliable embedded NVM solution.
[55]
Comparing Anti-fuse Non-Volatile Memory in 28nm High-K Metal ...
Dec 13, 2011 · It is tamper resistance for secure data storage. It provides a wide variety of storage capacity options. And, it provides an extended ...
[56]
None
### Summary of Radiation Tolerance and Immunity of Antifuse FPGAs vs. SRAM FPGAs
[57]
Radiation-Tolerant FPGAs - Microchip Technology
High-reliability, radiation-tolerant antifuse-based FPGAs · Compact packaging for command and control applications · Flight heritage established on many programs ...
[58]
[PDF] High Reliability FPGAs in Fuze and Fuze Safety Applications
• Highest reliability. • Radiation configuration upset immunity. • Lowest power consumption. Flash FPGA Outlook. 2015. 2010. 2005. 2000. ProASIC3 / Igloo / ...
[59]
[PDF] eX Family FPGAs - Microchip Technology
The antifuse interconnect is made up of a combination of amorphous silicon and dielectric material with barrier metals and has an "on" state resistance of 25Ω ...<|separator|>
[60]
https://www.attopsemi.com/attopsemis-revolutionary-i-fuse-otp-silicon-proven-on-finfet-technology/
[61]
A Flexible, Field-programmable ROM Replacement
Mar 15, 2010 · Antifuse one-time programmable (OTP) provides a flexible, field-programmable alternative to ROM. An antifuse-based bit cell uses controlled, irreversible thin ...
[62]
One-Time Programmable - an overview | ScienceDirect Topics
Antifuse-based : Considered by many to offer the most security with regard to design IP, this also provides advantages like low power consumption, instant-on ...<|separator|>
[63]
Achieving Reliable and Secure SoC Designs with Advanced OTP IP
Jan 15, 2025 · Over programming an OTP can result in poor programming quality and unnecessary over-exposure of the cells to high voltage. Additionally, high ...
[64]
[PDF] Programming and Functional Failure Guidelines User Guide
If the security fuse is programmed (antifuse FPGAs), then Microchip will not be able to perform failure analysis. Programming of the security fuse restricts ...
[65]
I-fuse: Most Reliable and Fully Testable OTP - Design And Reuse
Feb 8, 2021 · In this case, any OTP programming failure or yield loss can be costly. There is a 10x rule as shown in Fig. 3. If a defect cannot be identified ...
[66]
Dispelling the Myths Surrounding Antifuse OTP - ChipEstimate.com
Jun 15, 2010 · Antifuse OTP has scaling and temperature limitations similar to those of floating-gate NVM. It's a Myth. Floating gate NVM bit cells detect ...