Welcome to Scribd, the world's digital library. Read, publish, and share books and documents. See more
Download
Standard view
Full view
of .
Look up keyword
Like this
1Activity
0 of .
Results for:
No results containing your search query
P. 1
Ion Implantation in Silicon Technology

Ion Implantation in Silicon Technology

Ratings: (0)|Views: 36|Likes:
Published by Mehdi Naderi
Ion Implantation in Silicon Technology
F E AT U R E

by Leonard Rubin and John Poate

on implanters are essential to modern integrated-circuit (IC) manufacturing. Doping or otherwise modifying silicon and other semiconductor wafers relies on the technology, which involves generating an Without ion beam and steering it into the substrate so that the ions come to rest implanters, beneath the surface. Ions may be allowed to travel through a beam line at the enertoday’s integrated gy at which they w
Ion Implantation in Silicon Technology
F E AT U R E

by Leonard Rubin and John Poate

on implanters are essential to modern integrated-circuit (IC) manufacturing. Doping or otherwise modifying silicon and other semiconductor wafers relies on the technology, which involves generating an Without ion beam and steering it into the substrate so that the ions come to rest implanters, beneath the surface. Ions may be allowed to travel through a beam line at the enertoday’s integrated gy at which they w

More info:

Published by: Mehdi Naderi on Feb 07, 2012
Copyright:Attribution Non-commercial

Availability:

Read on Scribd mobile: iPhone, iPad and Android.
download as PDF, TXT or read online from Scribd
See more
See less

02/07/2012

pdf

text

original

 
12
The Industrial Physicist
I
on implanters are essential to modern integrated-cir-cuit (IC) manufacturing. Doping or otherwise modify-ing silicon and other semiconductor wafers relies on thetechnology, which involves generating anion beam and steering it into the sub-strate so that the ions come to rest beneath the surface. Ions may be allowedto travel through a beam line at the ener-gy at which they were extracted from asource material, or they can be accelerat-ed or decelerated by dc or radio-frequen-cy (RF) electric fields.Semiconductor processors today useion implantation for almost all doping in silicon ICs. Themost commonly implanted species are arsenic, phospho-rus, boron, boron difluoride, indium, antimony, germa-nium, silicon, nitrogen, hydrogen, and helium.Implant-ing goes back to the 19th century, and has been continu-ally refined ever since. Physicist Robert Van de Graaff of the Massachusetts Institute of Technology and PrincetonUniversity helped pioneer accelerator construction, andthe high-voltage technology that emerged from this effort was instrumental in building High Voltage EngineeringCorp. (HVEC) in the late 1940s and 1950s. HVECserved as an incubator for the technology essential to building the first commercial ion implanters and theindividuals who pioneered the field. William Shockley first recognized the potential of ionimplantation for doping semiconductor materials, andhis 1954 patent application demonstrates a remarkableunderstanding of the relevant process issues long beforeimplantation entered mass production. However, thepatent expired in 1974, just as the commercial ion-implantation market began taking off. So althoughShockley demonstrated visionary insight, hispatent earned few royalties.Ion-implantation equipment and applica-tions gradually came together in the 1960s.Experience gained in building researchaccelerators improved hardware reliability and generated new techniques for purifyingand transporting ion beams. Theoristsrefined the hypothesis of ion stopping, which enabled the precise placement of ions based on the energy and angle of implanta-tion, and experimenters determined thathigh-temperature postimplant annealingcould repair implantation-induced crystaldamage. Initially, these anneals were done ata temperature of 500 to 700°C, but after several years, semiconductor processorsfound that the optimum annealing tempera-ture ranged from 900 to 1,100°C. After theresolution of process integration issues, ionimplantation rapidly displaced thermal dif-fusion of deposited dopants as the dominantmethod of semiconductor doping because it was more precise, reliable, and repeatable.IC manufacturers, especially IBM and Western Electric, designed and built many of the early ion implanters, almost exclusively for in-house use. But in the early 1970s, themarket for commercial ion implanters beganopening as start-up companies tapped thetechnology spun off from HVEC and the
Ion Implantation inSilicon Technology 
 by Leonard Rubin and John Poate
FEATURE
 JUNE/JULY 2003© American Institute of Physics
Withoutimplanters,today’s integratedcircuits would beimpossible
     C     O     N     C     E     N     T     R     A     T     I     O     N     (   a    t   o   m   s     /   c   m
     3
     )
DEPTH(
 µ
m)4 keV 1 MeV 2 MeV 3 MeV 0123410
21
10
20
10
19
10
18
10
17
10
16
10
15
Figure 1. Some of the most commonly implanted species highlightedon the periodic table, along with typical concentration-versus-depthtraces for various implant energies.
 
technology developed by IC manufacturers, who became their customers.Some memory circuits now sell for less than 20nanodollars/transistor. Today, implanters and other fab-rication hardware must meet aggressive productivity tar-gets to achieve this minuscule cost. A large wafer fabri-cator may process up to 50,000 wafers/month, witheach wafer requiring 20 to 30 implants. This outputrequires the use of about 20 implanters, each with thecapacity to implant more than 200 wafers/h. In practice,maximum implanter throughput typically ranges from250 to 270 wafers/h, including placing the wafers intoand removing them from sealed cassettes used by auto-mated material-handling systems. This throughput isachieved for wafer sizes of 150, 200, and 300 mm.Depending on the configuration of the beam line and theend station (the wafer-processing chamber), an implanter occupies an area of 16 to 28 m
2
. Thus, fabrication spaceposes almost as significant a barrier as capital costagainst compensating for poor throughput by installingadditional implanters.
 Applications
 Among semiconductor-processing techniques, ionimplantation is nearly unique in that process parameters,such as concentration and depth of the desired dopant,are specified directly in the equipment settings for implant dose and energy, respectively (Figure 1). This dif-fers from chemical vapor deposition, in which desiredparameters such as film thickness and density are complex func-tions of the tun-able-equipmentsettings, whichinclude tempera-ture and gas-flowrate. The number of implants need-ed to complete anIC has increasedas the complexity of the chips hasgrown. Whereasprocessing a sim-ple n-type metaloxide semiconduc-tor during the1970s may haverequired 6 to 8 implants, a modern complementary-metal-oxide-semiconductor (CMOS) IC with embeddedmemory may contain up to 35 implants. The technique’s applications require doses and ener-gies spanning several orders of magnitude. Mostimplants fall within one of the boxes in Figure 2. The boundaries of each box are approximate; individualprocesses vary because of differences in design trade-offs. Energy requirements for many applications havefallen with increased device scaling. A shallower dopantprofile helps keep aspect ratios roughly constant as later-al device dimensions shrink. As energies drop, ion dosesusually, but not always, decline as well. The width of thestatistical distribution of the implanted ions decreases with energy, and this reduces the dose required to pro-duce a given peak dopant concentration. The result is
13
The Industrial Physicist
Bonded wafer splittingfor silicon on insulator (H, He)
10
17
10
16
10
15
10
14
10
13
10
12
10
11
0.11101001,00010,000ENERGY(keV)
Polysilicondoping (As, B)Source–draincontact(As, BF
2
, B)
     D     O     S     E     (   a    t   o   m   s     /   c   m
     2     )
Source–drainextension(As, BF
2
, B)Preamorphization(Ge, Si)Bipolar buriedsubcollector (P, As)Latch-up/electrostaticdischargeprotection(B)CMOSretrograde wells(P, B, As)Channel engineering(As, BF
2
, P, B, In, Sb)Threshold voltage adjust(As, BF
2
, B, P, In) Antipunch-through(As, B, In, Sb)Noise isolation wells (P, B)Charge-coupleddevice wells (B)
Dual-slit extractionelectrodeElectron confinementbeam guide Analyzer magnetedge focusingElectronconfinementbeam tunnelFlag Faraday SelectableresolvinghousingDeceleration groundextentionPlasmaelectronflood(xenon) Wafer 
Figure 2. Dose andenergy require-ments of majorimplantation appli-cations (speciesshown roughly inorder of decreasingusage).Figure 3.Schematic of theelectron confine-ment technologynecessary totransport severalmilliamperes of beam at energiesbelow 10 keV in amodern high-cur-rent beam line.
 
14
The Industrial Physicist
the sloping lines in Figure 2. Implantation is actually extremely inefficient at modifying material composition. The highest ion dose implanted with an economicalthroughput is about 10
16
 /cm
2
, yet this corresponds to but 20 atomic layers. Only the extreme sensitivity of semiconductor conductivity to dopant concentrationmakes ion implantation practical.Ion energy requirements vary from less than 1 keV tomore than 3,000 keV. Accelerating ions to higher ener-gies requires a longer beam line, yet low-energy beamsare difficult to transport intact over longer distances because the beam cross section expands to a point whereit can no longer travel down the beam tube. This funda-mental physics makes it nearly impossible to construct a beam line capable of all required ion energies. Figure 2indicates that the largest magnitude in required dosesoccurs in the middle of the energy range. Because dose isessentially the beam’s charge multiplied by the implanta-tion time, available beam currents in the 5- to 200-keV range must vary by at least 4 orders of magnitude to per-form all required implants efficiently. This level is diffi-cult to reproduce repeatedly in a single ion-source/beam-line configuration.
Market segments
Consequently, the commercial ion-implanter marketlong ago evolved into three segments. The red, black,and blue regions of Figure 2 indicate high-current, medi-um-current, and high-energy applications, respectively. As the name suggests,
high-current implanters
producethe highest beam currents, up to 25 mA (Figure 3). For high-dose applications, the greater the beam current, thefaster the implantation, which means the output of more wafers per hour. Implanter makers have investeda greatdeal ofeffort in maximizing beam current, especially at thelowest energies, where Child’s law limits the flux of ionsextractable from a source.Although high-current implanterscan produce beams in the 10-µA range, source instabilitiesmake these beams unsuitable for low-dose applications. The short beam line of these implanters allows an energy range from <1 keV up to 100 to 200 keV.
Medium-current implanters
are designed for maxi-mum dose uniformity and repeatability. Their beam currents are in the range of 1 µA to 5 mA, at energies of 5 to ~600 keV. The wafer-processing end stations canimplant ions at angles up to 60°from theperpendicular to the wafer surface.This isessential for certain applications, such asanti-punchthrough implants, for example,in which dopants must be implanted par-tially underneath a previously formed gatestructure. The lower operational cost of medium-current implanters when used for lower-dose applications and their ability todo high-tilt implants distinguish them fromhigh-current implanters.Last, only 
high-energy implanters
cangeneratemegaelectron volt ion beams.Commercial high-energy implanters pro-duce beam currents for singly-charged ionsup to ~1 mA. Energies for multiply-chargedions can be up to ~4,000 keV, with beamcurrents of ~50 µA. High-energy implanterscan produce beams down to 10 keV, making them suit-able for many medium-current applications as well. Thisadditional functionality justifies the capital cost of thesemachines. High-energy implanters using both RF linear acceleration (Figure 4) and dc acceleration are used widely today in semiconductor manufacturing. A modern ion implanter costs about $2–5 million,depending on the model and the wafer size it processes. Of the three classes of implanters, the high-current machineshave traditionally been the biggest market in terms of rev-enue and unit volume. Revenue is increasing faster thanunit volume because implanters have become more expen-sive. However, both revenue and volume are subject to thesevere boom-and-bust cycles that have affected the entiresemiconductor capital-equipment industry in the pastdecade, a pattern that will likely continue.
Ten radio-frequency resonators(80 kV @ 13.56 MHz)Twelve quadrupolelenses (20 kV dc)Mass-analyzingmagnet(70°)Ten radio-frequency electrodes
Figure 4. Schematicof a radio-frequencylinear acceleratorused in a high-ener-gy ion implanter.

You're Reading a Free Preview

Download
scribd
/*********** DO NOT ALTER ANYTHING BELOW THIS LINE ! ************/ var s_code=s.t();if(s_code)document.write(s_code)//-->