appliancerepairfirst

Digital logic and electronic circuits derive their functionality from electronicswitches called transistors. Roughly speaking, the transistor can be likened to anelectronically controlled valve whereby energy applied to one connection of thevalve enables energy to ?ow between two other connections. By combiningmultiple transistors, digital logic building blocks such as AND gates and ?ip-?opsare formed. Transistors, in turn, are made from semiconductors. Consult aperiodic table of elements in a college chemistry textbook, and you will locatesemiconductors as a group of elements separating the metals and nonmetals.They are called semiconductors because of their ability to behave as both metalsand nonmetals. A semiconductor can be made to conduct electricity like a metalor to insulate as a nonmetal does. These differing electrical properties can beaccurately controlled by mixing the semiconductor with small amounts of otherelements. This mixing is called doping. A semiconductor can be doped tocontain more electrons (N-type) or fewer electrons (P-type). Examples ofcommonly used semiconductors are silicon and germanium. Phosphorousand boron are two elements that are used to dope N-type and P-type silicon,respectively.
A transistor is constructed by creating a sandwich of differently dopedsemiconductor layers. The two most common types of transistors, thebipolar-junction transistor (BJT) and the ?eld-effect transistor (FET) areschematically illustrated in Fig. 2.1. This ?gure shows both the silicon structuresof these elements and their graphical symbolic representation as would beseen in a circuit diagram. The BJT shown is an NPN transistor, because it iscomposed of a sandwich of N-P-N doped silicon. When a small current isinjected into the base terminal, a larger current is enabled to ?ow from thecollector to the emitter. The FET shown is an N-channel FET; it is composedof two N-type regions separated by a P-type substrate. When a voltage is appliedto the insulated gate terminal, a current is enabled to ?ow from the drain to thesource. It is called N-channel, because the gate voltage induces an N-channelwithin the substrate, enabling current to ?ow between the N-regions.
Another basic semiconductor structure shown in Fig. 2.1 is a diode, which isformed simply by a junction of N-type and P-type silicon. Diodes act likeone-way valves by conducting current only from P to N. Special diodes can becreated that emit light when a voltage is applied. Appropriately enough, thesecomponents are called light emitting diodes, or LEDs. These small lights aremanufactured by the millions and are found in diverse applications fromtelephones to traf?c lights.
The resulting small chip of semiconductor material on which a transistor ordiode is fabricated can be encased in a small plastic package for protectionagainst damage and contamination from the out-side world. Small wiresare connected within this package between the semiconductor sandwichand pins that protrude from the package to make electrical contact withother parts of the intended circuit.
Once you have several discrete transistors, digital logic can be built by directlywiring these components together. The circuit will function, but any substantialamount of digital logic will be very bulky, because several transistors arerequired to implement each of the various types of logic gates. At the timeof the invention of the transistor in 1947 by John Bardeen, Walter Brattain, andWilliam Shockley, the only way to assemble multiple transistors into a singlecircuit was to buy separate discrete transistors and wire them together. In 1959,Jack Kilby and Robert Noyce independently invented a means of fabricatingmultiple transistors on a single slab of semiconductor material. Their inventionwould come to be known as the integrated circuit, or IC, which is the foundationof our modern computerized world. An IC is so called because it integratesmultiple transistors and diodes onto the same small semiconductor chip. Insteadof having to solder individual wires between discrete components, an ICcontains many small components that are already wired together in the desiredtopology to form a circuit.

A typical IC, without its plastic or ceramic package, is a square or rectangularsilicon die measuring from 2 to 15 mm on an edge. Depending on the level oftechnology used to manufacture the IC, there may be anywhere from a dozento tens of millions of individual transistors on this small chip.
This amazing density of electronic components indicates that the transistorsand the wires that connect them are extremely small in size. Dimensions onan IC are measured in units of micrometers, with one micrometer (1 ?m)being one millionth of a meter. To serve as a reference point, a human hairis roughly 100 ?m in diameter. Some modern ICs contain components andwires that are measured in increments as small as 0.1 ?m! Each year,researchers and engineers have been ?nding new ways to steadily reducethese feature sizes to pack more transistors into the same silicon area, asindicated in Fig. 2.2.
Many individual chemical process steps are involved in fabricating an IC. Theprocess begins with a thin, clean, polished semiconductor wafer most oftensilicon that is usually one of three standard diameters: 100, 200, or 300 mm.The circular wafer is cut from a cylindrical ingot of solid silicon that has aperfect crystal structure. This perfect crystal base structure is necessary topromote the formation of other crystals that will be deposited by subsequentprocessing steps. Many dice are arranged on the wafer in a grid as shown inFig. 2.3. Each die is an identical copy of a master pattern and will eventuallybe sliced from the wafer and packaged as an IC. An IC designer determineshow different portions of the silicon wafer should be modi?ed to create transistors,diodes, resistors, capacitors, and wires. This IC design layout can then beused to, in effect, draw tiny components onto the surface of the silicon.Sequential drawing steps are able to build sandwiches of differently dopedsilicon and metal layers.
Engineers realized that light provided the best way to faithfully replicate patternsfrom a template onto a silicon substrate, similar to what photographers havebeen doing for years. A photographer takes a picture by brie?y exposing?lm with the desired image and then developing this ?lm into a negative.Once this negative has been created, many identical photographs can bereproduced by brie?y exposing the light-sensitive photographic paper tolight that is focused through the negative. Portions of the negative that aredark do not allow light to pass, and these corresponding regions of the paperare not exposed. Those areas of the negative that are light allow the paper tobe exposed.

When the paper is developed in a chemical bath, portions of the paper that wereexposed change color and yield a visible image.
Photographic processes provide excellent resolution of detail. Engineersapply this same principle in fabricating ICs to create details that are fractionsof a micron in size. Similar to a photographic negative, a mask is created foreach IC processing step. Like a photographic negative, the mask does nothave to be the same size as the silicon area it is to expose because, withlenses, light can be fo- cused through the mask to an arbitrary area. Using atechnique called photolithography, the silicon surface is ?rst prepared with alight-sensitive chemical called photoresist. The prepared surface is thenexposed to light through the mask. Depending on whether a positive ornegative photoresist process is employed, the areas of photoresist that havebeen either exposed or not exposed to light are washed away in a chemicalbath, resulting in a pattern of bare and covered areas of silicon. The wafercan then be exposed to chemical baths, high temperature metal vapors, andion beams. Only the bare areas that have had photoresist washed away areaffected in this step. In this way, speci?c areas of the silicon wafer can bedoped according to the IC designers speci?cations. Successive mask layersand process steps can continue to wash away and expose new layers ofphotoresist and then build sandwiches of semiconductor and metal material.A very simpli?ed view of these process steps is shown in Fig. 2.4. Thesemiconductor fabrication process must be performed in a cleanroomenvironment to prevent minute dust particles and other contaminants fromdisturbing the li- thography and chemical processing steps.
In reality, dozens of such steps are necessary to fabricate an IC. The semiconductorstructures that must be formed by layering different metals and dopants arecomplex and must be formed one thin layer at a time. Modern ICs typicallyhave more than four layers of metal, each layer separated from others by athin insulating layer of silicon dioxide. The use of more metal layers increasesthe cost of an IC, but it also increases its density, because more metal wirescan be fabricated to connect more transistors. This complete process fromstart to ?nish usually takes one to four weeks. The chemical diffusion step (5)is an example of how different regions of the silicon wafer are doped to achievevarying electrical characteristics. In reality, several successive doping steps arerequired to create transistors. The metal deposition step (10) is an example ofhow the microscopic metal wires that connect the many individual transistors arecreated. Hot metal vapors are passed over the prepared surface of the wafer.Over time, individual molecules adhere to the exposed areas and form continu-ous wires. Historically, most metal interconnects on silicon ICs are made fromaluminum. However, copper has become a common component of leading-edge ICs.
As IC feature sizes continue to shrink, the physical properties of light can becomelimiting factors in the resolution with which a wafer can be processed. Shorter lightwavelengths are necessary to meet the demands of leading-edge IC processtechnology. The human eye can detect electromag- netic energy from about 700 nm(red) to 400 nm (violet). Whereas ultraviolet light (< 400 nm) was once adequatefor IC fabrication, deep UV wavelengths are now in use, and shorter wavelengthsbelow 200 nm are being explored.

Each of the process steps is applied to the entire wafer. The many dice on a singlewafer are usually exposed to light through the same mask. The mask is either largeenough to cover the entire wafer and therefore expose all dice at once, or themask is stepped through the dice grid (using a machine appropriately called astepper) such that each die location is exposed separately before the nextprocessing step. In certain cases, such as small-volume or experimental runs,different die locations on the same wafer will be exposed with different masks.This is entirely feasible but may not be as ef?cient as creating a wafer on whichall dice are identical. When an IC is designed and fabricated, it generally followsone of two main transistor technologies: bipolar or metal-oxide semiconductor(MOS). Bipolar processes create BJTs, whereas MOS processes create FETs.Bipolar logic was more common before the 1980s, but MOS technologieshave since accounted the great majority of digital logic ICs. N-channel FETsare fabricated in an NMOS process, and P-channel FETs are fabricated ina PMOS process. In the 1980s, complementary-MOS, or CMOS, becamethe dominant process technology and remains so to this day. CMOSICs incorporate both NMOS and PMOS transistors.