Silicon Crystal Structure and Growth (Plummer - Chapter 3)

Silicon Crystal Structure and Growth

(Plummer - Chapter 3)

Amorphous Atomic Structure

Figure 4.3

Atomic Order of a Crystal Structure

Figure 4.2

Unit Cell in 3-D Structure

Unit cell

Figure 4.4

Unit cell forms

Faced-centered Cubic (FCC) Unit Cell

Figure 4.5

Silicon Crystal Structure

•Planes and directions are defined using x, y, z coordinates.[ •111 ]direction is defined by a vector of 1 unit in x, y and z.

•Planes defined by “Miller indices” – Their normal direction (reciprocals of intercepts of plane with the x, y and z axes).

Crystals are characterized by a unit cell which repeats in the x, y, z directions.

Miller Indices of Crystal Planes

Z

X

Y

(100)

Z

X

Y

(110)

Z

X

Y

(111)

Figure 4.9

Silicon Unit Cell: FCC Diamond Structure

Figure 4.6

Basic FCC Cell Merged FCC Cells

Omitting atoms outside Cell Bonding of Atoms

Silicon has the basic diamond crystal structure – two merged FCC cells offset by a/4 in x, y and z.

Crystal Orientations in IC Fab.

There are two major principal Silicon crystal orientations that are used in manufacturing IC , (1 1 1) and ( 1 0 0 ).

The surface terminations on can make a difference in the surface electrical and physical properties.

CMOS-Based Tech. crystals uses ( 1 0 0 ).Bipolar-Based Tech. crystals historically

used ( 1 1 0 ) but recently uses ( 1 0 0 ).

Crystal Orientation Benchmark

<1 1 1> < 1 0 0 > Orientation

Largest Lowest Atom`s density

Fastest Slow Oxidation Rate

Easier -- Growing facility

Most -- Defects

111 Orientation

100Orientation

Various types of defects can exist in a crystal (or can be created by processing steps). In general, these cause electrical leakage and are result in poorer devices.

(Extra line of atoms)

Dislocations

Misplacement of the unit cells in a single crystal.Caused by: Growth conditions, Lattice strain and

physical abuse during fabrication.Typical value: 200-1000 per square centimeters.

Major Defects

Vacancy: atom missing from a location in the structure.

A natural phenomenon in every crystal.

Occurs when a crystal or wafer is heated and cooled suddenly.

Minimization of vacancies: desire for low temperature processing.

Vacancy defect

Interstitial defect

Contaminants jammed in the crystal structure .

Semiconductor-Grade Silicon

Steps to Obtaining Semiconductor Grade Silicon (SGS)

Step Description of Process Reaction

1 Produce metallurgical grade silicon (MGS) by heating silica with carbon

SiC (s) + SiO2 (s) Si (l) + SiO(g) + CO (g)

2

Purify MG silicon through a chemical reaction to produce a silicon-bearing gas of trichlorosilane (SiHCl3)

Si (s) + 3HCl (g) SiHCl3 (g) + H2 (g) + heat

3

SiHCl3 and hydrogen react in a process called Siemens to obtain pure semiconductor- grade silicon (SGS)

2SiHCl3 (g) + 2H2 (g) 2Si (s) + 6HCl (g)

• Si is purified from SiO2 (sand) by refining, distillation and CVD.• It contains < 1 ppb impurities. Pulled crystals contain O (~1018

cm-3) and C (~1016 cm-3), plus dopants placed in the melt.

Czochralski (CZ) crystal growing

Crystal seed

Molten polysilicon

Heat shield

Water jacket

Single crystal silicon

Quartz crucible

Carbon heating element

Crystal puller and

rotation mechanism

CZ Crystal Puller

Figure 4.10

All Si wafers come from

“Czochralski” grown crystals.

Polysilicon is melted, then held

just below 1417 °C, and a single

crystal seed starts the growth.

Pull rate, melt temperature and

rotation rate control the

growth

Silicon Ingot Grown by CZ Method

Photograph courtesy of Kayex Corp., 300 mm Si ingot

Photo 4.1

An alternative process is the “Float Zone” process which can be used for refining or single crystal growth.

•In the float zone process, dopants and other impurities are rejected by the regrowing silicon crystal. Impurities tend to stay in the liquid and refining can be accomplished, especially with multiple passes.(See the Plummer for models of this process)

Float Zone Crystal Growth

RF

Gas inlet (inert)

Molten zone

Traveling RF coil

Polycrystalline rod (silicon)

Seed crystal

Inert gas out

Chuck

Chuck

Figure 4.11

Dopant Concentration Nomenclature

Concentration (Atoms/cm3)

Dopant Material

Type < 1014

(Very Lightly Doped)

1014 to 1016

(Lightly Doped) 1016 to 1019

(Doped) >1019

(Heavily Doped)

Pentavalent n n-- n- n n+ Trivalent p p-- p- p p+

Table 4.2

Segregation Fraction for FZ Refining

Crystal GrowthCrystal Growth

ShapingShaping

Wafer SlicingWafer Slicing

Wafer Lapping and Edge

Grind

Wafer Lapping and Edge

Grind

EtchingEtching

PolishingPolishing

CleaningCleaning

InspectionInspection

PackagingPackaging

Basic Process Steps for Wafer Preparation

Figure 4.19

Flat grind

Diameter grind

Preparing crystal ingot for grinding

Ingot Diameter Grind

Figure 4.20

Internal diameter

wafer saw

Internal Diameter Saw

Figure 4.23

After crystal pulling, the boule is shaped and cut into wafers which are then polished on one side.

Wafer Notch and Laser Scribe

1234567890

Notch Scribed identification number

Figure 4.22

Polished Wafer Edge

Figure 4.24

Chemical Etch of Wafer Surface to Remove Sawing Damage

Figure 4.25

Wafer Dimensions & Attributes

Table 4.3

Diameter (mm)

Thickness (m)

Area (cm2)

Weight (grams/lbs)

Weight/25 Wafers (lbs)

150 675 20 176.71 28 / 0.06 1.5 200 725 20 314.16 53.08 / 0.12 3 300 775 20 706.86 127.64 / 0.28 7 400 825 20 1256.64 241.56 / 0.53 13

88 die200-mm wafer

232 die300-mm wafer

Increase in Number of Chips on Larger Wafer Diameters(Assume large 1.5 x 1.5 cm microprocessors)

Figure 4.13

Developmental Specifications for 300-mm Wafer Dimensions and Orientation

Parameter Units NominalSome Typical

Tolerances

Diameter mm 300.00 0.20

Thickness(center point)

m 775 25

Warp (max) m 100

Nine-Point ThicknessVariation (max)

m 10

Notch Depth mm 1.00 + 0.25, -0.00

Notch Angle Degree 90 +5, -1

Back Surface Finish Bright Etched/Polished

Edge Profile Surface Finish Polished

FQA (Fixed Quality Area –radius permitted on the

wafer surface)mm 147

Table 4.4

From H. Huff, R. Foodall, R. Nilson, and S. Griffiths, “Thermal Processing Issues for 300-mm Silicon Wafers:Challenges and Opportunities,” ULSI Science and Technology (New Jersey: The Electrochemical Society, 1997), p. 139.

Wafer Polishing

Double-Sided Wafer Polish

Upper polishing pad

Lower polishing pad

Wafer

Slurry

Figure 4.26

Improving Silicon Wafer Requirements

Year(Critical Dimension)

1995(0.35 m)

1998(0.25 m)

2000(0.18 m)

2004(0.13 m)

Wafer diameter(mm)

200 200 300 300

Site flatnessA (m)Site size (mm x mm)

0.23(22 x 22)

0.17(26 x 32)

0.1226 x 32

0.0826 x 36

MicroroughnessB of frontsurface (RMS)C (nm)

0.2 0.15 0.1 0.1

Oxygen content(ppm)D

24 2 23 2 23 1.5 22 1.5

Bulk microdefectsE

(defects/cm2) 5000 1000 500 100

Particles per unit area(#/cm2)

0.17 0.13 0.075 0.055

EpilayerF thickness( % uniformity) (m) 3.0 ( 5%) 2.0 ( 3%) 1.4 ( 2%) 1.0 ( 2%)

Adapted from K. M. Kim, “Bigger and Better CZ Silicon Crystals,” Solid State Technology (November 1996), p. 71.

Quality Measures

Physical dimensionsFlatnessMicroroughnessOxygen content Crystal defectsParticlesBulk resistivity

“Backside Gettering” to Purify SiliconPolished Surface

Backside Implant: Ar (50 keV, 1015/cm2)

The argon amorphizes the back side of the silicon. The wafer is heated to 550oC, which regrows the silicon. However, the argon can not be absorbed by the silicon crystal so it precipitates into micro-bubbles and prevents some damage from annealing. The wafer is held at 550oC for several hours, and all mobile metal contaminants are attracted to and then captured by the argon stabilized damage. Once captured, they never leave these sites.

Chapter Review (Wafer Fabrication)Raw materials (SiO2) are refined to produce

electronic grade silicon with a purity unmatched by any other available material on earth.

CZ crystal growth produces structurally perfect Si single crystals which are cut into wafers and

polished .Starting wafers contain only dopants, and trace

amounts of contaminants O and C in measurable quantities.

Dopants can be incorporated during crystal growth

Point, line, and volume (1D, 2D, and 3D) defects can be present in crystals, particularly after high

temperature processing.Point defects are "fundamental" and their

concentration depends on temperature (exponentially), on doping level and on other

processes like ion implantation which can create non-equilibrium transient concentrations of

these defects.

Measurement of Wafer CharacteristicsDarkfield and Brightfield Detection

Brightfield imaging

Two-way mirror

Light source

Lens

Viewing optics

Viewing optics

Darkfield imaging

Light s

ourceLens

Light reflected by surface irregularities

Figure 7.15

Schematic of Optical System

Phase and intensity detection

Phase and intensity detection

Data generation, processing, display are networked with factory management software

Data generation, processing, display are networked with factory management software

Lens

Light source

Video camera

CRT

Photo detector array

Objective lens assembly

Viewing optics

Split mirror

Vibration isolation pad

Wafer positioning stage

Three-axis piezo substage

Figure 7.16

DetectorPinhole

Wafer is driven up and down along Z-axis

Laser

Pinhole

Beam splitter

Objective lens

Center of focus+Z

-Z0

Principle of Confocal Microscopy

Figure 7.17

Particle Detection by Light Scattering

Incident lightBeam

scanning

Photo detector

Particle

Wafer motion Scattered light

Reflected light

Detection of scattered

light

Figure 7.18

Measurement of Wafer Characteristics

The hot point probe is a simple and reliable means to determine whether a wafer is N or P

type is the Hot Point Probe. The basic operation of this probe is illustrated in the

next slide. Two probes make ohmic contact with the wafer surface. One is heated 25-100°C hotter than the other. A voltmeter placed across the probes will measure a

potential difference whose polarity indicates whether the material is N or P type.

Hot Point Probe

Basic principle of the hot probe, illustrated for an N-type sample, for determining N- or P-type behavior in semiconductors.

Hot Point Probe

Consider an N-type sample. The majority carriers are electrons. At the hot probe, the thermal energy of the

electrons is higher than at the cold probe so the electrons will tend to diffuse away from the hot probe, driven by the

temperature gradient. If a wire were connected between the hot and cold probes, this would result in a measurable

current, whose direction would correspond to the electrons moving right to left. (The current by definition would be in

the opposite direction.) If we place a high-impedance voltmeter between the probes, no current flows, but a potential difference is measured, as illustrated. As the

electrons diffuse away from the hot probe, they leave behind the positively charged, immobile donor atoms that provided the electrons. The negatively charged mobile electrons tend to build up near the cold probe. This results in the hot probe

becoming positive with respect to the cold probe. By a similar set of arguments, if the material were P type, positively

charged holes would be the majority carriers and the polarity of the induced voltage would be reversed. The direction of

the current between the two probes would also be reversed in P-type material, if they were shorted with a wire. Thus a

measurement of either the short-circuit current or the open circuit voltage tells us the type of the material.

“Four-point probe” measurement method. The outer two probes force a current through the sample; the inner two probes measure the voltage drop.

Measurement of Sheet Resistance

The most common method of measuring the wafer resistivity is with the four-point

probe. We measure the sample resistance by measuring the current that flows for a given applied voltage. This could be done

with just two probes. However, in that case, contact resistances associated with

the probes and current spreading problems around the probes are important and are not easily accounted for in the analysis. Using four probes allows us to force the

current through the two outer probes, where there will still be contact resistance

and current spreading problems, but we measure the voltage drop with the two

inner probes using a high-impedance voltmeter. Problems with probe contacts

are thus eliminated in the voltage measurement since no current flows

through these contacts.

Four Point Probe

Figure 7.3

Wafer

R

Voltmeter

Constant current source

V

Irs =

V

Ix 2ps (ohms-cm)

“Van der Pauw” Sheet Resistivity(similar to 4-point probe, but uses shapes on wafer)

I

(a)

(c) (d)

ContactConductive material

V(b)

Figure 7.4

Hall Effect Measurements

The Hall effect was discovered more than 100 years ago when Hall observed a transverse

voltage across a conductor subjected to a magnetic field .

The technique is more powerful than the sheet resistance method described above because it

can determine the material type, carrier concentration and carrier mobility separately.

The basic method is illustrated in the next slide. The left part of the figure defines the

reference directions and the various currents, fields and voltages; the right part of the figure

illustrates a top view of a practical geometry that is often used in semiconductor

applications.

Conceptual representation of Hall effect measurement. The right sketch is a top view of a more practical implementation.

Fourier Transform Infrared Spectroscopy (FTIR)

FTIR (Oxygen and Carbon Detection)The CZ crystal growth process introduces oxygen and carbon into the

silicon. These elements are not inert in the crystal. It is important is to be able to measure them and to control them. The method is Fourier

Transform Infrared Spectroscopy. FTIR measures the absorption of infrared energy by the molecules in a sample. Many molecules have vibrational

modes that absorb specific wavelengths when they are excited. By sweeping the wavelength of the incident energy and detecting which

wavelengths are absorbed, a characteristic signature of the molecules present is obtained. Oxygen in CZ crystals is located in interstitial sites in

the silicon lattice, bonded to two silicon atoms. Low concentrations of carbon are substitutional in silicon since carbon is located in the same

column of the periodic table as silicon and easily replaces a silicon atom. Oxygen exhibits a vibrational mode that absorbs energy at 1106 cm-1

(wavenumber), that is at a wavelength of about 9 microns; carbon absorbs energy at 607 cm-1.There are other wavelengths of IR light that

are absorbed by the silicon atoms themselves. By measuring the absorption of a particular wafer at 1106 or 607 cm-1, and comparing this absorption with an oxygen or carbon free reference, the FTIR technique

can be made quantitative .An IR beam is split by a partially reflecting mirror and then follows two separate paths to the sample and the detector. For pure silicon, if the

movable mirror is translated back and forth at constant speed, the detected signal will be sinusoidal as the two beams go in and out of phase.

The Fourier transform of this signal will simply be a delta function proportional to the incident intensity. If the frequency of the source is

swept, the Fourier transform of the resulting signal will produce an intensity spectrum. If we now insert the sample, the resulting intensity

spectrum will change because of absorption of specific wavelengths by the sample. The benefit of using the Fourier transform method as opposed to simply directly measuring the intensity spectrum is simply that the signal

to noise ratio is improved and as a result, the detection limit is reduced. With modern instruments, the detection limit for interstitial oxygen in

silicon is about 2x1015/cm3. Carbon can be detected down to about 5x1015/cm3. Oxygen precipitated into small SiO2 clusters can be detected by FTIR because in the SiO2 form, the oxygen does not absorb at 1106 cm-

1. As the precipitation occurs, the IR absorption at this wavenumber decreases.

Schematic of “TEM” Transmission Electron Microscope

}

Energy-loss spectrometer

Aperture

Sample stage

Detector

CCD video camera

Fluorescent screen

CRT

Condenser lens

Anode

Lenses

Electron gun

X-ray detector

Objective aperture

Displayed sample image

Liquid N2 Dewar

Wavelength of 1 MeVElectron ~ 1Angstrom

Electron Microscopy (TEM) of SiO2 on Si

Oxygen Contamination in Silicon

Oxygen is the most important impurity found in silicon. It is incorporated in silicon during the CZ growth process as a result of dissolution of the quartz crucible in

which the molten silicon is contained. The oxygen is typically at a level of about 1018 /cm3. It has recently become possible to use a magnetic field during CZ

growth to control thermal convection currents in the melt. This slows down the transport of oxygen from the crucible walls to the growing silicon interface and

reduces the oxygen concentration in the resulting crystal .Oxygen in silicon is always present at concentrations of ~10-20 ppm (5x1017-

1018/cm3) in CZ silicon. The oxygen can affect processes used in wafer fabrication such as impurity diffusion .

Oxygen has three principal effects in the silicon crystal .)1( In an as-grown crystal, the oxygen is believed to be incorporated primarily as

dispersed single atoms designated OI occupying interstitial positions in the silicon lattice, but covalently bonded to two silicon atoms. The oxygen atoms thus replace one of the normal Si-Si covalent bonds with a Si-O-Si structure. The

oxygen atom is neutral in this configuration and can be detected with the FTIR method. Such interstitial oxygen atoms improve the yield strength of silicon

by as much as 25%, making silicon wafers more robust in a manufacturing facility .)2( The formation of oxygen donors. A small amount of the oxygen in the

crystal forms SiO4 complexes which act as donors. They can be detected by changes in the silicon resistivity corresponding to the free electrons donated by

the oxygen complexes. As many as 1016/cm3 donors can be formed, which is sufficient to significantly increase the resistivity of lightly doped P-type wafers.

During the CZ growth process, the crystal cools slowly through ~500oC temperature and oxygen donors form. The SiO4 complexes are unstable at

temperatures above 500°C and so usually wafer manufacturers anneal the grown crystal or the wafers themselves after sawing and polishing, to remove the

oxygen complexes. These donors can reform, however, during normal IC manufacturing, if a thermal step around 400-500°C is used. Such steps are not

uncommon, particularly at the end of a process flow .)3( The tendency of the oxygen to precipitate under normal device processing

conditions, forming SiO2 regions inside the wafer. The precipitation arises because the oxygen was incorporated at the melt temperature and is therefore

supersaturated in the silicon at process temperatures.

Carbon Contamination in Silicon

Carbon is normally present in CZ grown silicon crystals at concentrations on the order of 1016/cm3.The carbon comes from the graphite components in the crystal pulling machine. The melt contains silicon and modest concentrations of oxygen. This results in the formation of SiO that evaporates from the melt surface. Generally, the ambient in the crystal puller is Ar flowing at reduced

pressure, and the SiO can be transported in the gas phase to the graphite crucible and other support fixtures. SiO reacts with graphite (carbon) to

produce CO that again transports through the gas phase back to the melt. From the melt, the carbon is incorporated into the growing crystal .

Four Effects of Carbon on Silicon(1) Carbon is mostly substitutional in the silicon lattice. Since it is a column IV

element, it does not act as a donor or acceptor in silicon. Carbon is known to affect the precipitation kinetics of oxygen in silicon. This is likely because

there is a volume expansion when oxygen precipitates and a volume contraction when carbon precipitates because of the relative sizes of O and

C. There is thus a tendency for precipitates that are complexes of C and O to form at minimum stresses in the crystal. Since precipitated SiO2 is crucial in

intrinsic gettering, this can have an effect on gettering efficiency.(2) Carbon is also known to interact with point defects in silicon. Silicon

interstitials tend to displace carbon atoms from lattice sites, presumably because this can help to compensate the volume contraction present when

there is carbon in the crystal .(3) Thermal donors (Oxygen Effects) normally form around 450°C. There is

also evidence that if C is present at ~1 ppm, these donors may also form at higher temperatures (650-1000°C) .

)4( Higher concentrations of C to Si (levels of a few percent) can change the bandgap of the silicon and may allow the fabrication of new types of

semiconductor devices in the future.

Chapter Review (Wafer Metrology)

Microscopic examination for particulates.

Hot Point Probe (wafer doping)Four Point Probe (wafer resistivity)Hall Effect (carrier mobility)FBIR (oxygen and carbon detection)TEM (atomic resolution of defects /

surface)Effects of Oxygen on IC fabricationEffects of Carbon on IC fabrication