1-2 How does a logic gate in a microchip work?

A gate seems like a device that must swing open and closed, yet microchips are etched onto silicon wafers that have no moving parts. So how can the gate open and close?

Larry Wissel, ASIC Applications Engineer at IBM Microelectronics, replies:

"Those of us who design logic gates for computers seldom reminisce on how the terms we use to describe technology came into use. The vision of a gate swinging back and forth clearly does not literally represent the structures on a silicon chip. But the reason for the usage of the term 'gate' for computer logic can be appreciated by examining the basic function of a gate: to control a flow."

"On a farm, gates may be used to control the 'flow' of sheep or goats between pens. In this case, the gate consists of a physical barrier whose position is controlled by a fanner. The farmer makes a decision about the flow of animals and then moves the physical barrier to permit the desired flow."

"In a computer, a gate controls the flow of electric current through a circuit. The gate consists of transistors; the transistors are selected by the chip designer from two basic types (PMOS and NMOS transistors) that are found in the ubiquitous CMOS (complementary metal-oxide semiconductor) technology.[1] The current that flows through a gate establishes a voltage at a particular point in the circuit. This voltage represents a single 'bit' of information. The voltage may either be high (representing the value '1’) or low (representing the value '0')."

"To establish a 1 on a circuit, the current flow is steered to the circuit (controlled) by 'turning on' a PMOS transistor connected between the circuit and the positive supply voltage. The supply voltage is usually an industry-standard value such as 3.3 or 5.0 volts. For the very brief interval that is required for a logic gate to switch (on the order of a nanosecond, or a billionth of a second), current will flow through a PMOS transistor from the positive power supply to the circuit."

"The current flow that charges the circuit node to a 0 is steered away from the circuit through a different kind of transistor (NMOS) connected between the circuit and the negative supply voltage, or electrical ground. Again, current will flow through the NMOS transistor for a very brief interval, but for the NMOS the current is between the circuit and the negative supply. In either case, the current flow results in a change in the circuit voltage that represents a bit of information. So, when a gate is controlling current flow, it is actually controlling the flow of information."

"Returning to the analogy between the farm and the computer chip, it is obvious that the flow is different (farm animals compared to information) and that the gate itself is different (a physical barrier compared to a transistor in the CMOS technology). But the most important difference is the means of controlling the flow. On the farm, the farmer resets the gate location by making a decision and then moving a physical barrier. A flow of animals through a complex maze of gates would require a farm hand at each gate."

"But in a computer chip, the control mechanism is the voltage on the control terminal of a transistor. This voltage turns on a transistor by changing its characteristics from that of an open circuit (the ‘off' position) to one that can conduct a small current. This control voltage, in turn, is already available within the chip as a voltage at a point on another circuit. And, being a voltage on a circuit, this control mechanism represents a different bit of information."

"The overwhelming computing power of logic gates stems from the fact that the output of any particular gate is a voltage, which can in turn be used to control another gate.[2] A computer chip therefore can be designed to make complex decisions about the information flow within itself. This ability enables sophisticated systems to be created by interconnecting as many as a million gates within a single chip. All of this with no farm hands and no moving parts."

TaK Ning of the IBM T.J. Watson Research Center adds some complementary details:

"A logic gate in a microchip is made up of a specific arrangement of transistors. For modem microchips, the transistors are of the kind called metal-oxide-semiconductor field-effect transistor (MOSFET), and the semiconductor used is silicon. A MOSFET has three components or regions: a source region, a drain region and a channel region having a gate over it. The three regions are arranged horizontally adjacent to one another, with the channel region in the middle."

"In a logic gate arrangement, each of the MOSFET works like a switch. The switch is closed, or the MOSPET is turned on, if electric current can flow readily from the source to the drain. The switch is open, or the MOSFET is turned off, if electric current cannot flow from the source to the drain."[3]

"The source and drain regions of a MOSFET are fabricated to be full of electrons which are ready to carry current. The channel region, on the other hand, is designed to be empty of electrons under normal condition, blocking the movement of current. Hence, under normal condition, the MOSFET is 'off' (or 'open') and no current can flow from the source to the drain."

"If a positive voltage is applied to the gate (which sits on top of the channel region), then electrons, which are negative charges, will be attracted toward the gate. These electrons are collected in the channel region of the MOSFET. The larger the gate voltage, the larger is the concentration of electrons in the channel region. The substantial concentration of electrons in the channel provides a path by which the electrons can move easily from the source to the drain. When that happens, the MOSFET is 'on' (or 'closed') and current can flow from the source to the drain freely".

"In summary, a MOSFET in a microchip is turned on by applying a voltage to the gate to attract electrons to the channel region, and turned off by applying a voltage to the gate to repel electrons away from the channel region. There is movement of charges in the silicon, but there are no mechanical moving parts involved."[4] 

What's a MOSFET?

MOSFET stands for metal-oxide-semiconductor field-effect transistor. It's a kind of transistor that clips gradually when overdriven, as most tubes do.

Both tubes and transistors amplify signals by passing current from one side of the device to the other, sculpting it along the way to the same shape as a much weaker input signal. It's like a movie or slide projector — a source of energy (the bulb) is shaped by the film, and projected on the screen, where we see a much bigger version of the image on the film (even though the actual light we see comes from the bulb, not the film).

There are basically three kinds of transistor that are used to amplify audio: the most common is a bipolar transistor. It is a sandwich of three layers of silicon, with the outer ones negatively charged and the middle one positively charged (NPN), on the other way around (PNP). A small signal on the middle layer controls a much bigger current passing between the two outer layers.

A later development was the field-effect transistor (FET). Here the current doesn't have to pass through the middle layer of the sandwich. It passes near it, and is controlled by the field effect exerted on it. This was more efficient in a number of ways. It also happens to clip more softly than a bi-polar transistor.

The third type is an FET where the element doing the controlling doesn't even contact the channel carrying the large current. It's insulated with a thin layer of silicon dioxide - a kind of glass. This is the MOSFET, and it clips very softly.

The clipping characteristics of individual vacuum tube or solid-state semiconductors are by no means the whole story in the behavior of a circuit. You've probably noticed by now that a circuit with a tube in it can produce a sound that's buzzer and harsher than another that's made up of bi-polar transistors. And the sound that formed the original criterion for what is desirable in overdrive, the sound of a cranked non-master-volume tube amp, has got to do with a lot things besides the tubes. There are transformers, speakers and the interaction of these with the tubes, to say nothing of the acoustic and psycho-acoustic byproducts of playing loud. Anyone interested in getting a repeatable sound that isn't dependant on playing at a certain sound pressure level would better off discarding the dogma surrounding tubes and transistors, and employing the only devices that can be trusted—the ears.

WORDS AND PHRASES

reminisce        缅怀往事，话旧

steer  掌舵，操纵，驾驶

ubiquitous  无处不在的

wafers  晶片，圆片；

Involved            繁杂的，受牵扯的

COMS (complementary mental-oxide semiconductor)

互补金属氧化物半导体

   MOSFET (mental-oxide-semiconductor field-effect transistor)

   金属氧化物半导体场效应晶体管

NOTES

[1]．The gate consists of transistors；the transistors are selected by the chip designer from two basic types（PMOS and NMOS transistors）that are found in the ubiquitous CMOS (complementary metal－oxide semiconductor)technology.

门电路由晶体管组成，而这些晶体管是由芯片的设计者从广泛使用的CMOS（互补金属氧化物半导体）技术中出现的两种基本类型的晶体管（PMOS晶体管和NMOS晶体管）选择确定的。

[2]．The overwhelming computing power of logic gates stems from the fact that the output of any particular gate is a voltage，which can in turn be used to control another gate．

逻辑门电路的强大计算能力源于这样一个事实：任何特殊门电路的输出都是一个电压信号，这个电压又可以用来控制另外的门电路。

[3]．In a logic-gate arrangement，each of the MOSFET works like a switch．The switch is closed，or the MOSFET is turned on，if electric current can flow readily from the source  to the drain．The switch is open，or the MOSFET is turned off, if electric current cannot flow from the source to the drain．

在逻辑门电路的排列中，每一个场效应晶体管就像一个开关一样工作。如果电流可以容易地从源极流向漏极，则开关处于闭合状态，或场效应晶体管处于开启状态。如果电流不能从源极流向漏极，则开关处于断开状态，或场效应晶体管处于关闭状态。

[4]．In summary, a MOSFET in a microchip is turned on by applying a voltage to the gate to attract electrons to the channel region，and turned off by applying a voltage to the gate to repel electrons away from the channel region. There is movement of charges in the silicon, but there are no mechanical moving parts involved．

总之，微芯片上的一个场效应晶体管通过给漏极加载一个电压把电荷吸引到沟道区域，使场效应晶体管处于开启状态；如果给栅极加载一个电压来阻止电荷，使之远离为道区域，则场效应晶体管处于关闭状态。硅片中有电荷的运动，但并没有涉及任何可移动的机械部件。