Hubbry Logo
search
logo
Integrated injection logic avatar
Integrated injection logic
Integrated injection logic
current hub
1845301

Integrated injection logic

logo
Community Hub0 Subscribers
Read side by side
Integrated injection logic
View on Wikipedia
from Wikipedia
Simplified schematic of an I2L inverter.

Integrated injection logic (IIL, I2L, or I2L) is a class of digital circuits built with multiple collector bipolar junction transistors (BJT).[1] When introduced it had speed comparable to TTL yet was almost as low power as CMOS, making it ideal for use in VLSI (and larger) integrated circuits. The gates can be made smaller with this logic family than with CMOS because complementary transistors are not needed. Although the logic voltage levels are very close (High: 0.7V, Low: 0.2V), I2L has high noise immunity because it operates by current instead of voltage. I2L was developed in 1971 by Siegfried K. Wiedmann and Horst H. Berger who originally called it merged-transistor logic (MTL).[2] A disadvantage of this logic family is that the gates draw power when not switching unlike with CMOS.

Construction

[edit]
I2L NOR gate with two inputs, two outputs, and one voltage input for the current injector transistor

The I2L inverter gate is constructed with a PNP common base current source transistor and an NPN common emitter open collector inverter transistor (i.e. they are connected to the GND). On a wafer, these two transistors are merged. A small voltage (around 1 volts) is supplied to the emitter of the current source transistor to control the current supplied to the inverter transistor. Transistors are used for current sources on integrated circuits because they are much smaller than resistors.

Because the inverter is open collector, a wired AND operation may be performed by connecting an output from each of two or more gates together. Thus the fan-out of an output used in such a way is one. However, additional outputs may be produced by adding more collectors to the inverter transistor. The gates can be constructed very simply with just a single layer of interconnect metal.

In a discrete implementation of an I2L circuit, bipolar NPN transistors with multiple collectors can be replaced with multiple discrete 3-terminal NPN transistors connected in parallel having their bases connected together and their emitters connected likewise. The current source transistor may be replaced with a resistor from the positive supply to the base of the inverter transistor, since discrete resistors are smaller and less expensive than discrete transistors.

Similarly, the merged PNP current injector transistor and the NPN inverter transistor can be implemented as separate discrete components.

Operation

[edit]
IIL circuit

The heart of an I2L circuit is the common emitter open collector inverter. Typically, an inverter consists of an NPN transistor with the emitter connected to ground and the base biased with a forward current from the current source. The input is supplied to the base as either a current sink (low logic level) or as a high-z floating condition (high logic level). The output of an inverter is at the collector. Likewise, it is either a current sink (low logic level) or a high-z floating condition (high logic level).

Like direct-coupled transistor logic, there is no resistor between the output (collector) of one NPN transistor and the input (base) of the following transistor.

To understand how the inverter operates, it is necessary to understand the current flow. If the bias current is shunted to ground (low logic level), the transistor turns off and the collector floats (high logic level). If the bias current is not shunted to ground because the input is high-z (high logic level), the bias current flows through the transistor to the emitter, switching on the transistor, and allowing the collector to sink current (low logic level). Because the output of the inverter can sink current but cannot source current, it is safe to connect the outputs of multiple inverters together to form a wired AND gate. When the outputs of two inverters are wired together, the result is a two-input NOR gate because the configuration (NOT A) AND (NOT B) is equivalent to NOT (A OR B) (per De Morgan's Theorem). Finally the output of the NOR gate is inverted by IIL inverter in upper right of the diagram, the result is a two-input OR gate.

Due to internal parasitic capacitance in transistors, higher currents sourced into the base of the inverter transistor result in faster switching speeds, and since the voltage difference between high and low logic levels is smaller for I2L than other bipolar logic families (around 0.5 volts instead of around 3.3 or 5 volts), losses due to charging and discharging parasitic capacitances are minimized.

Usage

[edit]

I2L is relatively simple to construct on an integrated circuit, and was commonly used before the advent of CMOS logic by companies such as Motorola (now NXP Semiconductors)[3] and Texas Instruments. In 1975, Sinclair Radionics introduced one of the first consumer-grade digital watches, the Black Watch, which used I2L technology.[4] In 1976, Texas Instruments introduced SBP0400 CPU which used I2L technology. In the late 1970s, RCA used I²L in its CA3162 ADC 3 digit meter integrated circuit. In 1979, HP introduced a frequency measurement instrument based on a HP-made custom LSI chip that uses integrated injection logic (I2L) for low power consumption and high density, enabling portable battery operation, and also some emitter function logic (EFL) circuits where high speed is needed in its HP 5315A/B.[5]

References

[edit]

Further reading

[edit]
Revisions and contributorsEdit on WikipediaRead on Wikipedia

Integrated injection logic

View on Grokipedia
from Grokipedia
Integrated injection logic (I²L), also known as merged-transistor logic (MTL), is a bipolar integrated circuit technology that employs multicollector transistors powered by direct minority carrier injection, eliminating the need for resistors and enabling self-isolation for high-density logic gates.[1] Developed in 1972 by Kees Hart and Arie Slob at Philips Research Laboratories, I²L operates with low supply voltages under 1 V and achieves a low power-delay product of approximately 0.4–1.75 pJ, making it suitable for large-scale integration (LSI) with up to 1000 gates per chip. The technology's core structure relies on a common n-emitter plane where p-n-p injectors supply minority carriers to drive n-p-n switching transistors, allowing logic functions through wired-AND connections at collectors with unlimited fan-in but limited fan-out of about four.[1] Fabrication uses a simplified five-mask bipolar process, with some variants requiring four masks, supporting packing densities of 400 gates per mm² with 5-µm feature sizes, which surpasses traditional bipolar logics like TTL in density and power efficiency while offering TTL-compatible noise immunity.[1] Propagation delays range from approximately 100 ns at low power to 10–20 ns with increased injection current, though output drive is relatively weak compared to MOS technologies.[2] By the early 1980s, I²L had demonstrated versatility in combining with other bipolar logics, analog circuitry, and sensors on the same chip, positioning it for applications in low-power digital systems despite challenges like neutron radiation susceptibility and design complexity.[3] Its speed-power product of around 0.5 pJ and power dissipation of 0.7 µW per gate at 1 µA bias made it competitive for LSI arrays, though it faced competition from advancing MOS families in speed and drive capability.[2]

History and Development

Invention and Origins

In the late 1960s, bipolar junction transistors (BJTs) formed the foundation of digital integrated circuits, powering logic families such as transistor-transistor logic (TTL) and diode-transistor logic (DTL). These technologies enabled reliable high-speed operation but suffered from relatively high static power dissipation due to resistor-based current limiting and saturation effects in BJTs, which constrained the achievable density of large-scale integration.[4] This limitation spurred research into more efficient bipolar alternatives that could maintain TTL-like speeds while reducing power and area requirements.[5] Integrated injection logic (I²L), also known as merged-transistor logic (MTL), was independently invented in 1971 by two teams: Siegfried K. Wiedmann and Horst H. Berger at IBM Research in Germany, who termed it MTL, and Kees Hart and Arie Slob at Philips Research Laboratories in the Netherlands, who termed it I²L.[6][7] Their work built on ongoing efforts to optimize bipolar processes for higher integration, evolving from the constraints of earlier families like TTL and DTL by introducing a fundamentally simpler circuit topology.[8] The core innovation lay in merging a p-n-p current source transistor and an n-p-n switching transistor into a single shared structure, enabling direct minority carrier injection without discrete resistors or complex isolation.[6] This design drastically cut power consumption and silicon area, as the integrated p-n-p provided a stable current source for the n-p-n inverter, avoiding the base current spikes and storage delays common in saturated bipolar logics. The concept was first published in 1972 in the IEEE Journal of Solid-State Circuits, where the authors emphasized its suitability for very large-scale integration (VLSI), promising TTL-equivalent propagation delays around 50 ns with power efficiencies approaching those of emerging CMOS technologies at roughly 100 µW per gate.[6]

Early Commercialization

Researchers at IBM and Philips developed merged-transistor logic (MTL)/integrated injection logic (I²L) in 1971, with the concept detailed in 1972 IEEE papers and subsequent patenting by IBM.[6][7] This bipolar logic family promised high-density integration compatible with standard fabrication processes, facilitating the transition from small-scale to large-scale integration (LSI). Early commercialization emerged in the mid-1970s as companies licensed the technology from IBM and Philips. By 1974, production of I²L-based chips for microprocessors, watch circuits, and control logic was underway at firms including Philips and IBM Deutschland, with applications in digital voltmeters, frequency counters, and consumer electronics.[9] Texas Instruments led U.S. adoption with the 1976 release of the SBP0400, a 4-bit bit-slice microprocessor featuring over 1,450 gates in a single 40-pin package, targeted at low-power embedded systems. This device represented a breakthrough in LSI density for bipolar technology. In parallel, the technology spread to other semiconductor manufacturers; Signetics incorporated Bipolar/I²L processes in linear LSI products by the late 1970s, while Motorola explored I²L for custom logic circuits.[10] RCA advanced mixed-signal applications with the CA3162 in the late 1970s, a monolithic 3-digit A/D converter using I²L for multiplexed BCD output and dual-slope conversion, enabling compact display interfaces in instruments. However, scaling I²L to larger LSI designs faced hurdles from fabrication yields, as the merged transistor structures demanded tight process controls to avoid inconsistencies in current injection and collector performance, limiting widespread adoption beyond niche uses.[11][1]

Technical Construction

Basic Components

Integrated injection logic (I²L) relies on a merged complementary bipolar transistor structure that combines a lateral p-n-p injector transistor functioning as a current source with a vertical n-p-n switching transistor featuring multiple collectors to enable high-density logic integration. The p-n-p injector operates by forward-biasing a p-n junction to inject minority carriers (holes) into the n-type epitaxial layer, which serves as the base for the n-p-n transistor, thereby providing the necessary base current without requiring separate load resistors or power supplies.[12] This lateral p-n-p structure uses a long, narrow p-type emitter region, often configured as a shared injector rail, with its collector shorted to the n-p-n base to minimize parasitic capacitances through region merging. The vertical n-p-n transistor acts as the primary switching element, constructed with an n+ buried layer as the common emitter, a p-type base (the epitaxial layer), and shallow n+ diffusions forming multiple collectors on the surface, allowing direct fan-out to subsequent stages.[12] In the basic inverter configuration, a single p-n-p injector (Q1) supplies current to the base of the n-p-n transistor (Q2), whose single collector output connects to the base of the next stage's n-p-n; when Q2 saturates due to input current, it diverts the injector current, inverting the signal with a typical 0.6 V logic swing. The merged collector-base regions of the p-n-p and n-p-n reduce interconnect parasitics and isolation needs, enabling the inverter to occupy roughly the area of a single transistor.[12] For NOR gate implementation, a single p-n-p injector supplies current to the base of an n-p-n transistor, which may feature multiple collectors for fan-out. Multiple inputs from preceding stages are wired via their collectors to the shared base node. If all inputs are high (previous stages off, no shunting), the base current saturates the n-p-n, producing a low output. If any input is low (previous stage on, shunting base current to the emitter), the n-p-n turns off, producing a high output, realizing NOR logic through current steering and wired connections at the base with unlimited fan-in.[12] This configuration supports open-collector outputs, facilitating OR/NOR functions without additional wiring layers. Interconnects in I²L utilize a single layer of aluminum metallization over the injector rail for uniform current distribution, with base and collector contacts patterned to minimize capacitance and allow flexible post-fabrication adjustments for current levels.[12]

Fabrication Process

The fabrication of integrated injection logic (I²L) relies on a simplified bipolar semiconductor process that emphasizes compatibility with standard integrated circuit manufacturing techniques, enabling high-density integration without additional specialized steps. The process typically begins with an n⁺-type silicon substrate, onto which a thin n-type epitaxial layer (approximately 5 μm thick with variable resistivity) is grown to serve as the common collector region for the vertical n-p-n transistors.[1] This epitaxial structure allows for the formation of multiple collectors sharing the same epi layer, contributing to the technology's compact layout.[1] Key fabrication steps involve selective diffusions and ion implantations to define the transistor structures. The lateral p-n-p injector transistors are formed through p-type base implantation into the n-epi layer, creating the merging region where the p-n-p and n-p-n structures overlap without requiring separate isolation.[13] Emitter diffusions establish the n⁺ injectors and contacts, while base implantation precisely controls the p-region width (typically 2-3 μm) for the n-p-n inverters. Multiple shallow n⁺ collector diffusions are then performed to create the output terminals of the multicollector n-p-n transistors, often using up to four collectors per base for fan-in.[1] The process requires only five masks in its basic form: one for isolation, one for p-base diffusion, one for n⁺ emitters/collectors, one for contacts, and one for metallization, eliminating the need for resistors due to the current-mode operation.[1] This approach yields significant integration advantages, achieving gate densities around 400 gates per mm² through shared epitaxial layers and self-isolation, making I²L approximately 10 times denser than contemporary TTL circuits.[1] Typical feature sizes in the 1970s ranged from 5 to 10 μm, with minimum line widths of 5 μm enabling the small footprint.[1] Variations adapted the process for higher yields, such as isoplanar techniques employed by Texas Instruments and RCA, which incorporated oxide isolation and refined diffusion profiles to enhance manufacturability while maintaining bipolar compatibility.[12]

Operating Principles

Inverter and Gate Functionality

The inverter serves as the fundamental building block in integrated injection logic (I²L), consisting of a lateral p-n-p transistor functioning as a constant current injector and a vertical n-p-n transistor that performs the inversion. When the input voltage is low, approximately 0.2 V, the base-emitter junction of the n-p-n transistor is insufficiently forward-biased, turning it off; the injector current then flows through the load, resulting in a high output voltage around 0.7 V. Conversely, when the input is high at about 0.7 V, the n-p-n transistor saturates, sinking the injector current and pulling the output low to approximately 0.2 V. This current-steering mechanism enables efficient logic inversion with minimal power consumption, operating in a current-mode fashion where each gate typically draws 10–100 μA from the injector.[7] The NOR gate in I²L extends the inverter by incorporating multiple n-p-n transistors that share a common injector current source, allowing implementation of multi-input logic functions within a compact merged transistor structure. Each input connects to the base of a respective n-p-n transistor; if all inputs are low (0.2 V), all n-p-n transistors remain off, and the output remains high (0.7 V) as the injector current charges the load. However, if any input is high (0.7 V), the corresponding n-p-n transistor saturates, diverting the injector current and driving the output low (0.2 V). This configuration realizes the NOR function, expressed as $ Y = \overline{A + B} $ for a two-input gate, where the output is the logical complement of the OR of the inputs. The truth table for a two-input NOR gate is as follows:
ABY
LowLowHigh
LowHighLow
HighLowLow
HighHighLow
Logic levels in I²L are defined with a high state at approximately 0.7–1 V (unsaturated n-p-n collector) and a low state at about 0.2 V (saturated n-p-n collector), providing sufficient noise margin for reliable operation in dense circuits. Fan-out capability is typically up to 4 gates without additional buffering, achieved through multi-collector n-p-n transistors that distribute the sunk current across multiple outputs.[7][14]

Signal Propagation and Logic Levels

In integrated injection logic (I²L), signal propagation occurs through current steering between the p-n-p injector and n-p-n switching transistors, resulting in typical gate delays of 50-100 ns at low bias currents of 1-50 µA, which surpasses the performance of early CMOS technologies operating at similar power levels.[2] This speed arises from the minimal voltage swing of approximately 0.5 V, which minimizes the time needed to charge parasitic capacitances during transitions.[12] The open-collector outputs of I²L gates enable wired-OR configurations, facilitating AND-OR-INVERT logic functions across multiple inputs without additional circuitry. In multi-stage chains, pull-up injectors provide the necessary bias current to restore signal levels, ensuring reliable propagation over extended paths.[15] I²L demonstrates high noise immunity owing to its current-mode operation, where logic decisions depend on injected current rather than precise voltage thresholds, yielding typical input low (V_IL) and input high (V_IH) levels of about 0.5 V and 0.8 V, respectively. These levels exhibit a temperature coefficient of approximately -2 mV/°C, maintaining stability across operating ranges.[1] For multi-stage operation, the injector current establishes the switching threshold by controlling the base current in the n-p-n transistors. The output low voltage (V_OL) is given by the saturated base-emitter voltage of the n-p-n transistor:
VOL≈VBE(sat)≈0.2 V V_{OL} \approx V_{BE(sat)} \approx 0.2 \, \text{V}
This low V_OL supports efficient current sinking in subsequent stages.[15]

Performance Characteristics

Advantages

Integrated injection logic (I²L) offered significant power efficiency advantages over contemporary logic families like TTL, consuming approximately 10 µW per gate at typical operating currents of 10 µA and a supply voltage under 1 V, which is about 100 times lower than TTL's 10 mW per gate.[12] This low power dissipation, combined with the absence of resistors and reliance on current-mode operation for signal propagation, enabled dense large-scale integration (LSI) without the need for extensive cooling systems, supporting up to 400 gates per mm² with 5-µm features.[1] In terms of density and speed, I²L gates were 5 to 10 times smaller than equivalent TTL gates, achieving packing densities of 120 to 200 gates per mm² with 7-µm features compared to TTL's 20 gates per mm², while delivering propagation delays of 30 to 50 ns—approaching the performance of power-hungry ECL without its high dissipation.[12] The power-delay product (PDP) for I²L was approximately 0.1 to 0.7 pJ per gate, calculated as PDP = P × τ where P is the power dissipation per gate (e.g., 10 µW) and τ is the propagation delay (e.g., 30 ns), yielding PDP ≈ 0.3 × 10^{-12} J or 0.3 pJ; this represented an improvement of over 1,000 times compared to TTL's 100 pJ.[12] I²L facilitated ease of integration through its simple bipolar fabrication process requiring only 5 masks and a single-polarity supply of 1 to 5 V, eliminating the need for resistors and allowing seamless combination of digital and analog functions on the same chip. Limited fan-out of approximately 4 via multi-collector transistors simplified circuit design, though fan-in is unlimited.[12][1] Regarding cost, I²L's streamlined epitaxial process with fewer masking steps and high yields at LSI scales provided lower fabrication expenses than the multi-layer processes required for early CMOS, while leveraging mature bipolar manufacturing infrastructure.[2]

Disadvantages and Limitations

Integrated injection logic (I²L) gates exhibit static power consumption due to the continuous injection current required for operation, typically drawing 10 to 100 μA per gate even when idle, in contrast to CMOS logic which has near-zero static power.[12] This ongoing current flow results in appreciable heat dissipation in large-scale arrays, with estimates of around 1 mW per chip for moderate integration levels, limiting the feasible size of I²L circuits without advanced cooling.[2] The technology is sensitive to temperature variations due to the inherent characteristics of bipolar junction transistors, necessitating compensation circuits for reliable operation above 50°C.[1] Voltage constraints further restrict applicability, as the logic operates within a narrow swing of about 0.8 V and supply voltages limited to 0.8-5 V, while the lateral p-n-p structure proves inefficient for scaling to sub-micron processes, hindering further miniaturization.[1] I²L circuits are susceptible to noise from substrate coupling, which can degrade signal integrity in dense layouts, with thermal noise amplified in the current-steering mechanism approximated by $ V_n \approx \sqrt{4kTR \Delta f} $, where $ k $ is Boltzmann's constant, $ T $ is temperature, $ R $ is resistance, and $ \Delta f $ is bandwidth.[2] Scaling to VLSI faced yield challenges due to process complexities.[2] Additionally, I²L circuits are susceptible to radiation, such as neutrons, which can degrade performance in high-radiation environments.[2]

Applications and Legacy

Historical Usage in Devices

Integrated injection logic (I²L) found practical application in various consumer and industrial devices during the 1970s, leveraging its low power consumption and high density for battery-operated and compact systems. One notable example was the Sinclair Black Watch, a digital calculator wristwatch introduced by Sinclair Radionics in 1975 in the UK. This device utilized an I²L chip to perform arithmetic functions while driving the LED display, enabling a compact, low-power design suitable for portable wearables.[16] In computing and embedded control, Texas Instruments' SBP0400, released in 1976, represented an early LSI processor based on I²L technology. This 4-bit slice processor, expandable to 8-bit or wider configurations, was designed for high-performance controllers and computers, offering TTL/MOS compatibility and operation times around 300 ns at low power levels (e.g., 200 mW). It supported microprogrammable architectures with 459 microinstructions, making it ideal for embedded applications in industrial systems. Similarly, Hewlett-Packard's 5315A and 5315B frequency counters, introduced in 1979, incorporated a custom I²L-based multi-register counter (MRC) chip for data acquisition. The I²L implementation in the MRC provided high packing density for serial logic functions like decade chains, complementing emitter function logic (EFL) for efficient 100 MHz measurements in portable instruments.[17][18] Beyond computing, I²L appeared in analog-digital interfaces and entertainment devices. RCA's CA3162, a 3.5-digit A/D converter chip from the late 1970s, employed I²L for its multiplexed BCD output and dual-slope conversion, facilitating low-power digital displays in panel meters and instrumentation. In early video games, Texas Instruments' SN76477 complex sound generator chip (1978) used I²L for its digital control logic, enabling versatile audio effects in arcade titles like Space Invaders while maintaining compatibility with TTL systems. By the late 1970s, I²L enabled integration of up to a few thousand gates per chip, as seen in portable calculators and similar devices, where its bipolar density approached MOS levels for cost-effective logic. It also found use in some early pocket calculators and Soviet-era clones like the KR582 series processors.[19][20][21]

Decline and Modern Relevance

By the mid-1980s, the rise of complementary metal-oxide-semiconductor (CMOS) technology marked the beginning of I²L's decline, as CMOS offered near-zero static power dissipation and superior scalability for large-scale integration compared to bipolar-based approaches like I²L.[22] Bipolar logics, including I²L, suffered from inherent current consumption even in idle states, limiting their efficiency as integration densities increased.[23] Additionally, I²L circuits were constrained to roughly thousands of gates per chip due to thermal management challenges and manufacturing yield limitations arising from the shared substrate current injection mechanism.[12] I²L saw its last significant commercial applications in hybrid analog-digital circuits during the 1980s, particularly in low-power devices like watches and calculators, before being largely phased out for general digital logic by the early 1990s in favor of more versatile CMOS processes.[24] The technology's static power drawbacks, while innovative for its era, could not compete with CMOS's energy efficiency as VLSI demands escalated.[25] Research efforts in the early 2010s, such as adaptations of I²L in high-speed SiGe bipolar processes, explored its potential in low-power scenarios, but no active commercial production exists today.[26] Its legacy endures in historical studies of low-power bipolar design principles and influenced the power-delay product optimizations that shaped early BiCMOS hybrid technologies for combining bipolar speed with CMOS efficiency.[22]
User Avatar
No comments yet.