Silicon

Silicon is a chemical element with symbol Si and atomic number 14.

Silicon is the eighth most common element in the universe by mass, but very rarely occurs as the pure free
element in nature. It is most widely distributed in dusts, sands, planetoids, and planets as various forms of
silicon dioxide (silica) or silicates. Over 90% of the Earth's crust is composed of silicate minerals, making
silicon the second most abundant element in the Earth's crust (about 28% by mass) after oxygen.
[7]

Silicon is a solid at room temperature, with relatively high melting and boiling points of 1414 and 3265 C,
respectively.

Production
Metallurgical grade
Metallurgical grade silicon is commercially prepared by the reaction of high-purity silica with wood,
charcoal, and coal in an electric arc furnace using carbon electrodes. At temperatures over 1,900 C
(3,450 F), the carbon and the silicon undergo the chemical reaction SiO + 2 C Si + 2 CO. Liquid silicon
collects in the bottom of the furnace, which is then drained and cooled. The silicon produced in this manner
is called metallurgical grade silicon and is at least 98% pure.
2

As of September 2008, metallurgical grade silicon costs about US$1.45 per pound ($3.20/kg), up from
$0.77 per pound ($1.70/kg) in 2005.
[29]

Electronic grade
The use of silicon in semiconductor devices demands a much greater purity than afforded by metallurgical
grade silicon. Very pure silicon (>99.9%) can be extracted directly from solid silica or other silicon
compounds by molten salt electrolysis. This method, known as early as 1854 (see also FFC Cambridge
process, mentioned later), has the potential to directly produce solar-grade silicon without any carbon
dioxide emission at much lower energy consumption.

Solar grade silicon cannot be used for microelectronics. To properly control the quantum mechanical
properties, the purity of the silicon must be very high. Bulk silicon wafers used at the beginning of the
integrated circuit making process must first be refined to a purity of 99.9999999% often referred to as "9N"
for "9 nines", a process which requires repeated applications of refining technology.
The majority of silicon crystals grown for device production are produced by the Czochralski process, (CZSi) it was the cheapest method available. However, single crystals grown by the Czochralski process contain
impurities because the crucible containing the melt often dissolves. Historically, a number of methods have
been used to produce ultra-high-purity silicon.

Monocrystalline silicon ingot grown by the Czochralski process (find later here)

Early purification techniques

Early silicon purification techniques were based on the fact that if silicon is melted and re-solidified, the last
parts of the mass to solidify contain most of the impurities. The earliest method of silicon purification, first
described in 1919 and used on a limited basis to make radar components during World War II, involved
crushing metallurgical grade silicon and then partially dissolving the silicon powder in an acid. When
crushed, the silicon cracked so that the weaker impurity-rich regions were on the outside of the resulting
grains of silicon. As a result, the impurity-rich silicon was the first to be dissolved when treated with acid,
leaving behind a more pure product.
In zone melting, also called zone refining, the first silicon purification method to be widely used industrially,
rods of metallurgical grade silicon are heated to melt at one end. Then, the heater is slowly moved down the
length of the rod, keeping a small length of the rod molten as the silicon cools and re-solidifies behind it.
Since most impurities tend to remain in the molten region rather than re-solidify, when the process is
complete, most of the impurities in the rod will have been moved into the end that was the last to be melted.
This end is then cut off and discarded, and the process repeated if a still higher purity is desired.

A polycrystalline silicon rod made by the Siemens process

Modern purification techniques

Today's purification processes involve the conversion of silicon into volatile liquids, such as trichlorosilane
(HSiCl ) and silicon tetrachloride (SiCl ) or into the gaseous silane (SiH ). These compounds are then
separated by a distillation and transformed into high-purity silicon, either by a redox reaction or by chemical
decomposition at high temperatures.
3

In the late 1950s, the American chemical company DuPont patented a method for the production of 99.99%
pure silicon, using the metal zinc as a reductant to transform redistilled silicon tetrachloride into high-purity
silicon by a vapor phase reaction at 900 C. This technique, however, was plagued with practical problems,
as the byproduct zinc chloride (ZnCl ) solidified and clogged lines, and was eventually abandoned in favor
of more sophisticated processes.
2

[35]

Schematic diagram of the traditional Siemens and the Fluidized bed reactor purification process.

The best known technique is the so-called Siemens process. This technique does not require a reductant
such as zinc, as it grows high-purity silicon crystallites directly on the surface of (pre-existing) pure silicon
seed rods by a chemical decomposition that takes place when the gaseous trichlorosilane is blown over the
rod's surface at 1150 C. This process produces polycrystalline silicon, also known as polysilicon, at
typically 9N11N purities, that is, it contains impurity levels of less than one part per billion (ppb).
[36][37][38]

A more recent alternative for the production of polysilicon is the fluidized bed reactor (FBR) manufacturing
technology. Compared to the traditional Siemens process, FBR features a number of advantages that lead to
cheaper polysilicon demanded by the fast-growing photovoltaic industry.
Contrary to Siemens' batch process, FBR runs continuously, wasting fewer resources and requires less setup
and downtime. It uses about 10 percent of the electricity consumed by a conventional rod reactor in the
established Siemens process, as it does not waste energy by placing heated gas and silicon in contact with
cold surfaces. In the FBR, silane (SiH ) is injected into the reactor from below and forms a fluidized bed
together with the silicon seed particles that are fed from above. The gaseous silane then decomposes and
deposits silicon on the seed particles. When the particle has grown to larger granules, they eventually sink to
the bottom of the reactor where they are continuously withdrawn from the reaction process.
4

While the Siemens technology can produce polysilicon at purity levels at or above 9N to 11N, FBR outputs
polysilicon at 6N to 9N, purity suited for the photovoltaic industry and still higher than the 5N to 6N of
upgraded metallurgical silicon (UMG-Si). Currently most silicon for the photovoltaic market is produced by
the Siemens process and only about 10 percent by the FBR technology, while UMG-Si accounts for about 2

percent. By 2020, however, IHS Technology predicts that market shares for FBR technology and UMG-Si
will grow to 16.7 and 5.4 percent, respectively.
The company REC is one of the leading producers of silane and polysilicon using FBR technology. The
three-step chemical reaction involves (last step occurs inside the FB-reactor): (1.) 3 SiCl + Si + 2 H
4 HSiCl , followed by (2.) 4 HSiCl 3 SiCl + SiH , and (3.) SiH Si + 2 H . Other precursors such as
tribromosilane had been used by other companies as well.
4

Applications
Electronics
Main article: Semiconductor device fabrication

[40]

Silicon wafer with mirror finish

Most elemental silicon produced remains as ferrosilicon alloy, and only a relatively small amount (20%) of
the elemental silicon produced is refined to metallurgical grade purity (a total of 1.31.5 million metric
tons/year). The fraction of silicon metal which is further refined to semiconductor purity is estimated at only
15% of the world production of metallurgical grade silicon. However, the economic importance of this
small very high-purity fraction (especially the ~ 5% which is processed to monocrystalline silicon for use in
integrated circuits) is disproportionately large.
[28]

Pure monocrystalline silicon is used to produce silicon wafers used in the semiconductor industry, in
electronics and in some high-cost and high-efficiency photovoltaic applications. In terms of charge
conduction, pure silicon is an intrinsic semiconductor which means that unlike metals it conducts electron

holes and electrons that may be released from atoms within the crystal by heat, and thus increase silicon's
electrical conductivity with higher temperatures. Pure silicon has too low a conductivity (i.e., too high a
resistivity) to be used as a circuit element in electronics. In practice, pure silicon is doped with small
concentrations of certain other elements, a process that greatly increases its conductivity and adjusts its
electrical response by controlling the number and charge (positive or negative) of activated carriers. Such
control is necessary for transistors, solar cells, semiconductor detectors and other semiconductor devices,
which are used in the computer industry and other technical applications. For example, in silicon photonics,
silicon can be used as a continuous wave Raman laser medium to produce coherent light, though it is
ineffective as an everyday light source.
In common integrated circuits, a wafer of monocrystalline silicon serves as a mechanical support for the
circuits, which are created by doping, and insulated from each other by thin layers of silicon oxide, an
insulator that is easily produced by exposing the element to oxygen under the proper conditions. Silicon has
become the most popular material to build both high power semiconductors and integrated circuits. The
reason is that silicon is the semiconductor that can withstand the highest temperatures and electrical powers
without becoming dysfunctional due to avalanche breakdown (a process in which an electron avalanche is
created by a chain reaction process whereby heat produces free electrons and holes, which in turn produce
more current which produces more heat). In addition, the insulating oxide of silicon is not soluble in water,
which gives it an advantage over germanium (an element with similar properties which can also be used in
semiconductor devices) in certain type of fabrication techniques.
[59]

Monocrystalline silicon is expensive to produce, and is usually only justified in production of integrated
circuits, where tiny crystal imperfections can interfere with tiny circuit paths. For other uses, other types of
pure silicon which do not exist as single crystals may be employed. These include hydrogenated amorphous
silicon and upgraded metallurgical-grade silicon (UMG-Si) which are used in the production of low-cost,
large-area electronics in applications such as liquid crystal displays, and of large-area, low-cost, thin-film
solar cells. Such semiconductor grades of silicon which are either slightly less pure than those used in
integrated circuits, or which are produced in polycrystalline rather than monocrystalline form, make up
roughly similar amount of silicon as are produced for the monocrystalline silicon semiconductor industry, or
75,000 to 150,000 metric tons per year. However, production of such materials is growing more quickly than
silicon for the integrated circuit market. By 2013 polycrystalline silicon production, used mostly in solar
cells, is projected to reach 200,000 metric tons per year, while monocrystalline semiconductor silicon
production (used in computer microchips) remains below 50,000 tons/year.

FFC Cambridge process

The FFC Cambridge Process is an electrochemical method in which solid metal compounds, particularly oxides, are
catholically reduced to the respective metals or alloys in molten salts. It is thought that this process will eventually be
capable of producing metals or alloys more efficiently than by current conventional processes, such as titanium by the Kroll
process.

History

The FFC Cambridge process was developed by George Z. Chen, Derek J. Fray and Tom W. Farthing between 1996 and
1997 in the University of Cambridge. (The name FFC derives from the first letters of their last names.) They reduced oxide
scales on titanium foils, as well as small pellets of titanium dioxide powder, to the metal by molten salt electrochemistry.[1]
The technology has since been commercialized in the form of an intellectual property company known as Metalizes, which
is based in Sheffield, UK.[2] A similar process was patented in 1904 as German patent 150557.

Process
The process typically takes place between 9001100 C, with an anode (typically carbon) and a cathode (oxide being
reduced) in a bath of molten CaCl2. Depending on the nature of the oxide it will exist at a particular potential relative to the
anode, which is dependent on the quantity of CaO present in CaCl2. The cathode is then polarized to more negative voltages
versus the anode. This is simply achieved by imposing a voltage between the anode and cathode. When polarized to more
negative voltages the oxide releases oxygen ions into the CaCl2 salt, which exists as CaO. To maintain charge neutrality, as
oxygen ions are released from the cathode into the salt, so oxygen ions must be released from the salt to the anode. This is
observed as CO or CO2 being evolved at the carbon anode. In theory an inert anode could be used to produce oxygen.
When negative voltages are reached, it is possible that the cathode would begin to produce Ca (which is soluble in CaCl2).
Ca is highly reductive and would further strip oxygen from the cathode, resulting in calciothermic reduction. However, Ca
dissolved into CaCl2 results in a more conductive salt leading to reduced current efficiencies.
Cathode reaction mechanism
The electro-calciothermic reduction mechanism may be represented by the following sequence of reactions.
(1) MO
x+ x Ca M + x CaO

When this reaction takes place on its own, it is referred to as the "calciothermic reduction" (or, more generally, an example
of metallothermic reduction). For example, if the cathode was primarily made from TiO then calciothermic reduction would
appear as:
TiO + Ca Ti + CaO
Whilst the cathode reaction can be written as above it is in fact a gradual removal of oxygen from the oxide. For example, it
has been shown that TiO2 does not simply reduce to Ti. It, in fact, reduces through the lower oxides (Ti3O5, Ti2O3, TiO etc.) to
Ti.
The calcium oxide produced is then electrolyzed:
(2a) x CaO x Ca2+ + x O2
(2b) x Ca2+ + 2x e x Ca
And
(2c) x O2 x/2 O2 + 2x e
Reaction (2b) describes the production of Ca metal from Ca2+ ions within the salt, at the cathode. The Ca would then proceed
to reduce the cathode.
The net result of reactions (1) and (2) is simply the reduction of the oxide into metal plus oxygen:
(3) MO
x M + x/2 O2

Anode reaction mechanism`

The use of molten CaCl2 is important because this molten salt can dissolve and transport the O2 ions to the anode to be
discharged. The anode reaction depends on the material of the anode. Depending on the system it is possible to produce
either CO or CO2 or a mixture at the carbon anode.
C + 2O2 CO2 +4 e
C + O2 CO +2 e
However, if an inert anode is used, such as that of high density SnO2, the discharge of the O2 ions leads to the evolution of
oxygen gas. However the use of an inert anode has disadvantages. Firstly, when the concentration of CaO is low, Cl2
evolution at the anode becomes more favorable. In addition, when compared to a carbon anode, more energy is required to
achieve the same reduced phase at the cathode. Inert anodes suffer from stability issues.[citation needed]
Inert anode: 2O2 O2 + 4 e

Czochralski process
From Wikipedia, the free encyclopedia

Jump to: navigation, search

The Czochralski process is a method of crystal growth used to obtain single crystals of semiconductors (e.g. silicon, germanium and gallium
arsenide), metals (e.g. palladium, platinum, silver, gold), salts and synthetic gemstones. The process is named after Polish scientist Jan
Czochralski,[1] who invented the method in 1916 while investigating the crystallization rates of metals.[2]
The most important application may be the growth of large cylindrical ingots, or boules, of single crystal silicon. Other semiconductors, such as
gallium arsenide, can also be grown by this method, although lower defect densities in this case can be obtained using variants of the BridgmanStockbarger technique.

Production of Czochralski silicon

A puller rod with seed crystal for growing single-crystal silicon by the Czochralski process.
High-purity, semiconductor-grade silicon (only a few parts per million of impurities) is melted in a crucible at 1425 degree Celsius, usually made
of quartz. Dopant impurity atoms such as boron or phosphorus can be added to the molten silicon in precise amounts to dope the silicon, thus
changing it into p-type or n-type silicon, with different electronic properties. A precisely oriented rod-mounted seed crystal is dipped into the
molten silicon. The seed crystal's rod is slowly pulled upwards and rotated simultaneously. By precisely controlling the temperature gradients,
rate of pulling and speed of rotation, it is possible to extract a large, single-crystal, cylindrical ingot from the melt. Occurrence of unwanted
instabilities in the melt can be avoided by investigating and visualizing the temperature and velocity fields during the crystal growth process.[3]
This process is normally performed in an inert atmosphere, such as argon, in an inert chamber, such as quartz.

Size of crystals

Silicon crystal grown by Czochralski process at Raytheon, 1956. The induction heating coil is visible, and the end of the crystal just emerging
from the melt. The technician is measuring the temperature with an optical pyrometer. One of the earliest Si plants, the crystals produced by this
early apparatus were only one inch in diameter.
Due to the efficiencies of common wafer specifications, the semiconductor industry has used wafers with standardized dimensions. In the early
days, the boules were smaller, only a few inches wide. With advanced technology, high-end device manufacturers use 200 mm and 300 mm

diameter wafers. The width is controlled by precise control of the temperature, the speeds of rotation and the speed the seed holder is withdrawn.
The crystal ingots from which these wafers are sliced can be up to 2 meters in length, weighing several hundred kilograms. Larger wafers allow
improvements in manufacturing efficiency, as more chips can be fabricated on each wafer, so there has been a steady drive to increase silicon
wafer sizes. The next step up, 450 mm, is currently scheduled for introduction in 2018.[4] Silicon wafers are typically about 0.20.75 mm thick,
and can be polished to great flatness for making integrated circuits or textured for making solar cells.
The process begins when the chamber is heated to approximately 1500 degrees Celsius, melting the silicon. When the silicon is fully melted, a
small seed crystal mounted on the end of a rotating shaft is slowly lowered until it just dips below the surface of the molten silicon. The shaft
rotates counterclockwise and the crucible rotates clockwise. The rotating rod is then drawn upwards very slowly, allowing a roughly cylindrical
boule to be formed. The boule can be from one to two meters, depending on the amount of silicon in the crucible.
The electrical characteristics of the silicon are controlled by adding material like phosphorus or boron to the silicon before it is melted. The
added material is called dopant and the process is called doping. This method is also used with semiconductor materials other than silicon, such
as gallium arsenide.
Monocrystalline silicon grown by the Czochralski process is the basic material in the production of the large-scale integrated circuit chips used
in computers, TVs, mobile phones and all types of electronic equipment.[5] Monocrystalline silicon is also used in large quantity for producing
photovoltaic solar cells. The almost perfect crystal structure yields the highest light-to-electricity conversion efficiency for silicon.

Impurity incorporation
When silicon is grown by the Czochralski method, the melt is contained in a silica (quartz) crucible. During growth, the walls of the crucible
dissolve into the melt and Czochralski silicon therefore contains oxygen at a typical concentration of 1018
cm3
. Oxygen impurities can have beneficial effects. Carefully chosen annealing conditions can allow the formation of oxygen precipitates. These
have the effect of trapping unwanted transition metal impurities in a process known as gettering. Additionally, oxygen impurities can improve
the mechanical strength of silicon wafers by immobilizing any dislocations which may be introduced during device processing. It was
experimentally shown in the 1990s that the high oxygen concentration is also beneficial for radiation hardness of silicon particle detectors used
in harsh radiation environment (such as CERN's LHC/S-LHC projects).[6][7] Therefore, radiation detectors made of Czochralski- and Magnetic

Czochralski-silicon are considered to be promising candidates for many future high-energy physics experiments.[8][9] It has also been shown that
presence of oxygen in silicon increases impurity trapping during post-implantation annealing processes.[10]
However, oxygen impurities can react with boron in an illuminated environment, such as experienced by solar cells. This results in the formation
of an electrically active boronoxygen complex that detracts from cell performance. Module output drops by approximately 3 % during the first
few hours of light exposure.

Czochralski Crystal Growth Method

Created Jan 30, 2003 | Updated Aug 31, 2008

The Czochralski method is a technique for growing single-crystal silicon ingots for use in manufacturing
semiconductor devices. The finished crystals are called boules. The boules are later sliced into very thin, circular
wafers and then diced into the little silicon chips from which all silicon semiconductor LSI 1 chips are made.
Growing large monocrystalline silicon boules is a tricky business. The Czochralski method (abbreviated as CZ in
the technical literature) begins with silicon powder that is 99.999999999% pure 2. A lot of that ridiculously pure
silicon is placed in a large crucible that is made of fused quartz. The crucible is then put in a vacuum chamber,
which is sealed and then filled with some conveniently inert gas, usually argon.
The chamber is then heated up to 1500 Centigrade or so, to melt the silicon. When the silicon is nicely melted,
a small seed crystal mounted on the end of a rotating shaft is slowly lowered until it just dips below the surface of
the red-hot silicon melt. The shaft rotates counterclockwise and the crucible rotates clockwise. Now, the rotating
rod is drawn upwards very slowly, allowing a roughly cylindrical boule to form by some technical magic as it does
so3. The boule can be from one to two metres or so long, depending on how much silicon there is in the crucible.
In the early days of the technology, the boules were quite thin, only a few inches wide. However, the crystal
growers have had a lot of practice, and nowadays they can make nice, fat 300mm (12-inch) wide boules. The
thickness is controlled by precise control of the temperature, the speeds of rotation and how fast the seed holder
is withdrawn. Widths of 400mm (16 inches) are expected in the next several years. This is one reason for the
rapidly decreasing cost of chips that we have enjoyed over the years, because more LSI chips can be created
from a single wafer with the same number of fabrication process steps.
The electrical characteristics of the silicon are controlled by adding stuff like phosphorus or boron to the silicon
before it is melted. The stuff added is called dopant and the process is called doping. This method is also used
with semiconductor materials other than silicon, such as gallium arsenide.

As a necessary step in the production of large-scale integrated circuit chips, the Czochralski method is a basic
technique in the making of computers, TVs, cell phones and the advanced electronic equipment of all kinds that
shape modern life as we know it at the beginning of the 21st Century.
1

Large-scale Integrated Circuit.2Yes, that's no joke. This stuff contains less than a few parts per billion of
anything that is not silicon. That purity earns it a special name: electronic grade silicon. It has to be that pure for
use in making semiconductor devices, because the electrical characteristics of the silicon are controlled by the
intentional addition of special kinds of impurities at very low, precisely controlled concentrations, a process called
doping. Those special impurities are called dopants and include things like boron, arsenic and phosphorus. 3OK,
it's not really magic, of course. When the seed crystal is dipped into the melt and then pulled up slightly, surface
tension causes the liquid very near the seed to rise up above its normal level to meet the withdrawing seed,
forming a meniscus. (Just like water does around the edges of a tube.) The silicon in the meniscus cools down
and solidifies as crystalline silicon, thus forming the boule.

Fluidized bed reactor

A fluidized bed reactor (FBR) is a type of reactor device that can be used to carry out a variety of
multiphase chemical reactions. In this type of reactor, a fluid (gas or liquid) is passed through a granular
solid material (usually a catalyst possibly shaped as tiny spheres) at high enough velocities to suspend the
solid and cause it to behave as though it were a fluid. This process, known as fluidization, imparts many
important advantages to the FBR. As a result, the fluidized bed reactor is now used in many industrial
applications.

Basic diagram of a fluidized bed reactor.

Basic principles
The solid substrate (the catalytic material upon which chemical species react) material in the fluidized bed
reactor is typically supported by a porous plate, known as a distributor. The fluid is then forced through the
distributor up through the solid material. At lower fluid velocities, the solids remain in place as the fluid
[1]

passes through the voids in the material. This is known as a packed bed reactor. As the fluid velocity is
increased, the reactor will reach a stage where the force of the fluid on the solids is enough to balance the
weight of the solid material. This stage is known as incipient fluidization and occurs at this minimum
fluidization velocity. Once this minimum velocity is surpassed, the contents of the reactor bed begin to
expand and swirl around much like an agitated tank or boiling pot of water. The reactor is now a fluidized
bed. Depending on the operating conditions and properties of solid phase various flow regimes can be
observed in this reactor.

Monocrystalline silicon

A silicon ingot

Monocrystalline silicon (or "single-crystal silicon", "single-crystal Si", "mono c-Si", or just mono-Si) is the base material
for silicon chips used in virtually all electronic equipment today. Mono-Si also serves as photovoltaic, light-absorbing
material in the manufacture of solar cells.
It consists of silicon in which the crystal lattice of the entire solid is continuous, unbroken to its edges, and free of any grain
boundaries. Mono-Si can be prepared intrinsic, consisting only of exceedingly pure silicon, or doped, containing very small
quantities of other elements added to change its semiconducting properties. Most silicon monocrystals are grown by the
Czochralski process into ingots of up to 2 meters in length and weighing several hundred kilogrammes. These cylinders are
then sliced into thin wafers of a few hundred microns for further processing.
Single-crystal silicon is perhaps the most important technological material of the last few decadesthe "silicon era",[1]
because its availability at an affordable cost has been essential for the development of the electronic devices on which the
present day electronic and informatic revolution is based.
Monocrystalline silicon differs from other allotropic forms, such as the non-crystalline amorphous siliconused in thinfilm solar cells, and polycrystalline silicon, that consists of small crystals, also known as crystallites.

Mono-Si in solar cells

Global market-share in terms of annual production by PV technology since 1990

Monocrystalline silicon is used in the manufacturing of high performance solar cells. Since, however, solar cells are less
demanding than microelectronics for as concerns structural imperfections, monocrystaline solar grade (Sog-Si) is often
used.
Market-share
In 2013, monocrystalline solar cells had a market-share of 36 percent, that translated into the production of 12,600
megawatts of photovoltaic capacity,[3] and ranked second behind the somewhat cheaper sister-technology of polycrystalline
silicon.[4]
Efficiency
Lab efficiencies of 25.0 percent for mono-Si cells are the highest in the commercial PV market, ahead of polysilicon with
20.4 percent and all established thin-film technologies namely, CIGS cells (19.8%), CdTe cells (19.6%), and a-Si cells
(13.4%).[5]

Solar module efficiencieswhich are always lower than those of their corresponding cellscrossed the 20 percent mark
for mono-Si in 2012; an improvement of 5.5 percent over a period of ten years. The thickness of a silicon waver used to
produce a solar cell also decreased significantly, requiring less raw material and therefore less energy for its manufacture.
Increased efficiency combined with economic usage of resources and materials was the main driver for the price decline
over the last decade.[6]
..

VLSI devices (Intel) fabricated on a single-crystal silicon wafer)

Wafer (electronics)
From Wikipedia, the free encyclopedia

Jump to: navigation, search

Polished 12" and 6" silicon wafers. The notch in the left wafer and the flat cut into the right wafer indicates its
crystallographic orientation (see below)

VLSI microcircuits fabricated on a 12-inch (300 mm) silicon wafer, before dicing and packaging
A wafer, also called a slice or substrate,[1] is a thin slice of semiconductor material, such as a crystalline silicon, used in
electronics for the fabrication of integrated circuits and in photovoltaics for conventional, wafer-based solar cells. The wafer
serves as the substrate for microelectronic devices built in and over the wafer and undergoes many microfabrication process
steps such as doping or ion implantation, etching, deposition of various materials, and photolithographic patterning. Finally
the individual microcircuits are separated (dicing) and packaged.

History
On the 13th of December 1965, American-Engineers Eric O Ernst, Donald J. Hurd and Gerard Seeley filed Patent
US3423629A[2] while working under IBM. This Patent was for the first High-Capacity Epitaxial Apparatus and the methods
to create the, now, Silicon Wafer.

Formation

The Czochralski process.

2-inch (51 mm), 4-inch (100 mm), 6-inch (150 mm), and 8-inch (200 mm) wafers
Wafers are formed of highly pure (99.9999999% purity),[3] nearly defect-free single crystalline material.[4] One process for
forming crystalline wafers is known as Czochralski growth invented by the Polish chemist Jan Czochralski. In this process,
a cylindrical ingot of high purity monocrystalline semiconductor, such as silicon or germanium, is formed by pulling a seed
crystal from a 'melt'.[5][6] Donor impurity atoms, such as boron or phosphorus in the case of silicon, can be added to the
molten intrinsic material in precise amounts in order to dope the crystal, thus changing it into n-type or p-type extrinsic
semiconductor.
The ingot is then sliced with a wafer saw (wire saw) and polished to form wafers.[7] The size of wafers for photovoltaics is
100200 mm square and the thickness is 200300 m. In the future, 160 m will be the standard.[8] Electronics use wafer
sizes from 100450 mm diameter. (The largest wafers made have a diameter of 450 mm but are not yet in general use.)

Cleaning, texturing and etching

Wafers are cleaned with weak acids to remove unwanted particles, or repair damage caused during the sawing process.
When used for solar cells, the wafers are textured to create a rough surface to increase their efficiency. The generated PSG
(phosphosilicate glass) is removed from the edge of the wafer in the etching.

Wafer properties
Standard wafer sizes

Solar wafers on the conveyor

Completed solar wafer

Silicon wafers are available in a variety of diameters from 25.4 mm (1 inch) to 300 mm (11.8 inches).[10] Semiconductor
fabrication plants (also known as fabs) are defined by the diameter of wafers that they are tooled to produce. The diameter
has gradually increased to improve throughput and reduce cost with the current state-of-the-art fab considered to be 300 mm
(12 inch), with the next standard projected to be 450 mm (18 inch).[11][12] Intel, TSMC and Samsung are separately conducting
research to the advent of 450 mm "prototype" (research) fabs, though serious hurdles remain.

1-inch (25 mm)

2-inch (51 mm). Thickness 275 m.

3-inch (76 mm). Thickness 375 m.

4-inch (100 mm). Thickness 525 m.

5-inch (130 mm) or 125 mm (4.9 inch). Thickness 625 m.

150 mm (5.9 inch, usually referred to as "6 inch"). Thickness 675 m.

200 mm (7.9 inch, usually referred to as "8 inch"). Thickness 725 m.

300 mm (11.8 inch, usually referred to as "12 inch"). Thickness 775 m.

450 mm (17.7 inch, usually referred to as "18 inch"). Thickness 925 m (expected).[13]

Wafers grown using materials other than silicon will have different thicknesses than a silicon wafer of the same diameter.
Wafer thickness is determined by the mechanical strength of the material used; the wafer must be thick enough to support its
own weight without cracking during handling.
A unit wafer fabrication step, such as an etch step or a lithography step, can be performed on more chips per wafer as
roughly the square of the increase in wafer diameter, while the cost of the unit fabrication step goes up more slowly than the
square of the wafer diameter. This is the cost basis for shifting to larger and larger wafer sizes. Conversion to 300 mm
wafers from 200 mm wafers began in earnest in 2000, and reduced the price per die about 30-40%.[14] However, this was not
without significant problems for the industry.
There is considerable resistance to moving up to 450 mm despite the expected productivity improvement, mainly because
companies feel it would take too long to recoup their investment.[15] Machinery needed to handle and process larger wafers
results in increased investment costs to build a single factory. Lithographer Chris Mack claimed in 2012 that the overall
price per die for 450 mm wafers would be reduced by only 10-20% compared to 300 mm wafers, because presently over

50% of total wafer processing costs are lithography-related. Converting to larger 450 mm wafers would reduce price per die
only for process operations such as etch where cost is related to wafer count, not wafer area. Cost for processes such as
lithography is proportional to wafer area, and larger wafers would not reduce the lithography contribution to die cost.[16]
Nikon plans to deliver 450-mm lithography equipment in 2015, with volume production in 2017.[17][18] In November 2013
ASML paused development of 450-mm lithography equipment, citing uncertain timing of chipmaker demand.[19]
The time-line for 450 mm has not been fixed as of 2014. Mark Durcan, CEO of Micron Technology, said in February 2014
that he expects 450 mm adoption to be delayed indefinitely or discontinued. I am not convinced that 450mm will ever
happen but, to the extent that it does, its a long way out in the future. There is not a lot of necessity for Micron, at least over
the next five years, to be spending a lot of money on 450mm. There is a lot of investment that needs to go on in the
equipment community to make that happen. And the value at the end of the day so that customers would buy that
equipment I think is dubious.[20] As of March 2014, Intel Corporation expects 450 mm deployment by 2020 (by the end of
this decade).[21] Mark LaPedus of semiengineering.com reported in mid-2014 that chipmakers had previously set 2016 to
2018, though this has been delayed for the foreseeable future. According to this report some observers expect 2018 to
2020, while G. Dan Hutcheson, chief executive of VLSI Research, doesnt see 450mm fabs moving into production until
2020 to 2025.[22]
The step up to 300 mm required a major change from the past, with fully automated factories using 300 mm wafers versus
barely automated factories for the 200 mm wafers. These major investments were undertaken in the economic downturn
following the dot-com bubble, resulting in huge resistance to upgrading to 450 mm by the original timeframe. Other initial
technical problems in the ramp up to 300 mm included vibration effects, gravitational bending (sag), and problems with
flatness. Among the new problems in the ramp up to 450 mm are that the crystal ingots will be 3 times heavier (total weight
a metric ton) and take 2-4 times longer to cool, and the process time will be double.[23] All told, the development of 450 mm
wafers requires significant engineering, time, and cost to overcome.

Wafers are grown from crystal having a regular crystal structure, with silicon having a diamond cubic structure with a lattice
spacing of 5.430710 (0.5430710 nm).[25] When cut into wafers, the surface is aligned in one of several relative directions
known as crystal orientations. Orientation is defined by the Miller index with (100) or (111) faces being the most common
for silicon.[25] Orientation is important since many of a single crystal's structural and electronic properties are highly
anisotropic. Ion implantation depths depend on the wafer's crystal orientation, since each direction offers distinct paths for
transport.[26] Wafer cleavage typically occurs only in a few well-defined directions. Scoring the wafer along cleavage planes
allows it to be easily diced into individual chips ("dies") so that the billions of individual circuit elements on an average
wafer can be separated into many individual circuits.

Polycrystalline silicon

Left side: solar cells made of multicrystalline silicon Right side: polysilicon rod (top) and chunks (bottom)
Polycrystalline silicon, also called polysilicon or poly-Si, is a high purity, multicrystalline form of silicon, used as a raw
material by the solar photovoltaic and electronics industry.
Large polysilicon rods, usually broken into chunks of specific sizes and packaged in clean rooms before shipment, are either
directly cast into multicrystalline ingots or submitted to a recrystallization process to grow monocrystalline silicon. The
resulting ingots are then sliced into thin silicon wafers and used for the production of solar cells, integrated circuits and
other semiconductor devices
Polysilicon is produced from metallurgical grade silicon by a chemical purification process. This process involves
distillation of volatile silicon compounds, decomposition at high temperatures, or refinement in fluid phase. When produced
for the electronics industry, polysilicon contains impurity levels of less than one part per billion (ppb), while polycrystalline
solar grade silicon (SoG-Si) is generally less pure. The photovoltaic industry also produces upgraded metallurgical-grade
silicon (UMG-Si), using metallurgical instead of chemical purification processes.

Polysilicon consists of small crystals, also known as crystallites, giving the material its typical metal flake effect. While
polysilicon and multisilicon are often used as synonyms, multicrystalline usually refers to crystalls larger than 1 mm.
Multicrystalline solar cells are the most common type of solar cells used in photovoltaics and consume most of the
worldwide produced polysilicon. Polysilicon is distinct from monocrystalline silicon and amorphous silicon.

Polycrystalline vs monocrystalline silicon

Comparing polycrystalline (left) to monocrystalline (right) solar cells

In single crystal silicon, also known as monocrystalline silicon, the crystalline framework is homogenous, which can be
recognized by an even external colouring.[1] The entire sample is one single, continuous and unbroken crystal as its structure
contains no grain boundaries. Large single crystals are rare in nature and can also be difficult to produce in the laboratory
(see also recrystallisation). In contrast, in an amorphous structure the order in atomic positions is limited to short range.

Polycrystalline and paracrystalline phases are composed of a number of smaller crystals or crystallites. Polycrystalline
silicon (or semi-crystalline silicon, polysilicon, poly-Si, or simply "poly") is a material consisting of multiple small silicon
crystals. Polycrystalline cells can be recognized by a visible grain, a "metal flake effect". Semiconductor grade (also solar
grade) polycrystalline silicon is converted to "single crystal" silicon meaning that the randomly associated crystallites of
silicon in "polycrystalline silicon" are converted to a large "single" crystal. Single crystal silicon is used to manufacture
most Si-based microelectronic devices. Polycrystalline silicon can be as much as 99.9999% pure.[2] Ultra-pure poly is used
in the semiconductor industry, starting from poly rods that are two to three meters in length. In microelectronic industry
(semiconductor industry), poly is used both at the macro-scale and micro-scale (component) level. Single crystals are grown
using the Czochralski process, float-zone and Bridgman techniques.

Siemens Reactor

Fig. 1. Siemens reactor

PolySim is a powerful tool for design and

optimization of reactors for polycrystalline
silicon deposition from chlorosilanes by
Siemens process. The tool is largely oriented
to engineers and researchers and does not
require expertise in Computational Fluid
Dynamics from the users. The unique features
of the tool are built-in chemical and heat
reactor models combined with simple and
convenient Graphical User Interface (GUI),
which works in "grower-friendly" terms similar
to those used in real reactor operation,
minimizing the possible errors and saving
user time needed to set-up the problem.
The growth simulation includes modeling of
numerous physicochemical processes such as
turbulent heat and mass transfer, radiative

heat transfer, gas-phase and surface chemical

reactions, and electrical heating of the silicon
rods. The software employs chemical models
involving gas-phase precursor decomposition
and interaction and an original surface
chemistry model. The approach suggested
allows description of the chemical processes
at the gas-solid interfaces in a wide range of
operating conditions.
The PolySim GUI includes everything required
for the problem specification, solution control,
and visualization of the results. The program
allows the user to find almost all the reactor
characteristics
depending
on
varying
operating conditions and main characteristics
of reactor design. It is also possible to
evaluate the effect of the conditions nonuniformity on the growth of the different rod
sections. For modeling the whole growth
process, a number of separate points for
different rod diameters are computed, and
then the instantaneous reactor characteristics

are time-integrated. The solution residuals are

visualized by the GUI, allowing easy
convergence
control.
The
computed
characteristics include reactor productivity,
energy consumption per 1 kg of silicon, silicon
conversion, electrical current parameters, gas
flow rates, rod center temperature, gas
depletion at the growth surface, the flow
criteria, the energy consumption content.
Run-time
and
post-processing
visualization
is
available
within
the
GUI,
presenting
dependencies
of
the
input
parameters
and
computed
characteristics.
Fig.
3.
PolySim
results:
support dependence of Silicon output on
provided temperature and pressure

Hot-line
can be
on request.

The

Fig.
2.
PolySim
results:
dependence of energy cost on
temperature and pressure

PV Manufacturing
Some of the manufacturing processes and resources for
photovoltaics are shared with other applications, especially
electronic chips for computers, mobile phones and any other
electronic device. This competition has caused a shortage in supply
of crystalline cells.
Manufacturing Silicon

The raw material of most solar cells today is crystalline silicon.

Luckily, silicon is one of the most widely available elements in the

form of sand. Before silicon can be cut into thin wafers, however, it
has to be purified, as otherwise the photoeffect will not be very
efficient. Purity levels for solar cells do not have to be as high as in
chip applications. Solar-grade purity is 99.999% (5N) as opposed to
electronic-grade silicon purity of up to 99.9999999% (9N).

There are three main categories of manufacturing processes,

resulting in different purity levels:

Electronic-grade Silicon: 9N
There are three main steps to produce high-purity polycrystalline
silicon.
1.

2.

3.

Coke reduction: Metallurgical-grade silicon with 98.5% purity is

produced from quartz sand in an arc furnace at very high
temperatures.
Distillation: In a second step, the metallurgical grade silicon
powder is disolved in hydrogen chloride and subsequently
distilled to form a silane gas. In most instances, this is the
trichlorosilane, but could be others.
Siemens Process: In the so-called Siemens Process the
polycrystalline silicon is grown at very high temperatures. It
requires hydrogen and produces more hydrogen-chloride as a

by-product.
Medium-grade Silicon: 6-7N
The big drawback of the standard process as above is that a
Siemens reactor is very expensive and the Siemens process itself
requires a lot of energy. A number of new proprietary processes
reduce the energy consumption and the capital costs for silicon
production, though they are still similar to the traditional Siemens
process.

Fluidized Bed Reactor (Renewable Energy Corporation):

operates at much lower temperatures and does not produce byproducts.
Vapour to liquid deposition (Tokuyama): similar to Siemens,
but faster extraction.

Upgraded Metallurgical-grade Silicon (UMG): > 5N

In an altogether different process, metallurgical-grade silicon is
chemically refined. By blowing gasses through the silicon melt, the
boron and phosphorous impurities are removed, followed by
directional solidification. Companies like Timminco, Arise or RSI
Silicon all have their own proprietary processes. However, they all
have in common that by avoiding high purification, manufacturing
costs are reduced significantly.

Manufacturing Wafers

There are mainly three

different silicon wafer types of different qualities:

Monocrystalline wafer: Silicon with a single, continuous

crystal structure is grown from a small seed crystal that is slowly
pulled out of a polysilicon melt into a cylindrical shaped ingot
(Czochralski process). The ingot is cut into wafers using a
diamond saw. Silicon waste from the sawing process can be re-

cycled into polysilicon.

Polycrystalline wafer:Polycrystalline silicon consists of small

grains of monocrystalline silicon. Cube-shaped ingots can be
made directly by casting molten polysilicon, which are then cut
into wafers similar to monocrystalline wafers.
Silicon ribbons: This is a continuous process whereby thin
ribbons or sheets of multicrystalline silicon are drawn from a
polysilicon melt. The subsequent cutting into wafers does not
produce waste, as the drawn sheets are already wafer-thin.
Silicon ribbons require around 5g of silicon per Watt rather than
8g/W using crystalline wafers.

Manufacturing Cells and Modules

Crystalline
Crystalline cells are made from silicon wafers by cleaning and
doping the wafer. In a separate manufacturing process, a number of
cells are wired up to form a module. As such the manufacturing
process of crystalline modules consists of four distinct processes:
Polysilicon production, Ingot & Wafer manufacturing, cell
manufacturing and module manufacturing.
Thin-Film
"Thin" means that the semiconductor layer is about 1/100th times
"thinner" than in crystalline cells. The manufacturing process starts
by depositing the thin photoactive film on the substrate, which
could be either glass or a transparent film. Afterwards, the film is

structured into cells similarly to the crystalline module. Unlike

crystalline modules, the manufacturing process of thin-film modules
is a single process that can not be split up..
For CdTe - thin-film, 220kg are required for 1MWp, which is 36 times
less weight per kWp than crystalline silicon.

Polysilicon Production
Polysilicon is a key ingredient in the manufacture of semiconductors and solar cells. Polysilicon is
grown into ingots and then sliced into wafers, which are then used either for solar cells or for
semiconductor chips. The most common manufacturing process of polysilicon is the Siemens
process, but other methods are also being developed. This blog post will discuss the Siemens
method, the fluidized bed reactor method, and UMG silicon production method.

Figure I: Polysilicon Manufacturing Steps (Source: Green Rhino Energy)

All these processes begin with metallurgical grade silicon (MG silicon): SiO2 sand is mixed with
carbon and undergoes coke reduction in an arc furnace at 1,800 degrees C. This product is then
either dissolved in HCl at 300 degrees C and distilled to produce high purity trichlorosilane
(HSiCl3), or is chemically refined.
Figure II: MG Silicon Production (Source: Open Stax)

The Siemens process is most commonly used due to the high purity polysilicon that is produced
through the process. The polysilicon produced is 9N quality, or 99.9999999% silicon. This is the
current standard for the semiconductor industry, and is known as electronic grade polysilicon.
After the MG silicon has been processed into HSiCl3, this is introduced to the Siemens deposition
reactor. The reactors contain two Si slim rods. The reactor is heated to 1100 degrees C, which
act as a nucleation point for the deposition of silicon. This results in a polycrystalline rod
consisting of columnar grains of silicon.
Figure III: Siemens Process (Source: Open Stax)

The fluidized bed reactor (FBR) method of producing polysilicon produced medium grade silicon,
6-7N or 99.999999%-99.9999999% Si. This process was developed by REC Silicon, and the
following information is from their website. FBR operates at much lower temperatures than the
Siemens process, and does not produce by-products. Instead of Si-rods, the process uses silicon
seed granules. These are fed into a chamber with heated silane gas, which fluidises the silicon
granules. As the silane gas breaks down, it deposits more silicon on the granules, which grow
larger and exit the chamber once they have reached a sufficient weight. This process uses much
less energy than the Siemens process and produces more silicon per cubic meter of reactor space
because the silicon crystals have a larger total surface area than the rods used in the Siemens
process.
Figure IV: FBR Method vs. Siemens Method of Polysilicon Production (Source: REC
Silicon)

Finally, UMG silicon is produced using chemical refining of MG silicon. This process produces
slightly lower grade silicon than the previous two methods discussed, but still> 99.99999% Si
product. UMG silicon is produced by blowing gasses through the silicon melt, removing boron and
phosphorous impurities. The material is then directionally solidified, creating an ingot of
polysilicon. There is no one method for producing UMG silicon: Each company producing this
product have their own processes and chemical mixtures that they use. This method is becoming
more popular as the costs associated with the Siemens process and FBR process are greatly
reduced by avoiding high purification.

Each of these processes have their own strengths and weaknesses. The
Siemens process is the most expensive method of polysilicon production, but
also produces the purest material. FBR polysilicon production is between
the Siemens process and UMG silicon in price and quality, making it a
potential replacement within the semiconductor industry. UMG silicon

cannot be used in semiconductor manufacturing, but is becoming more

popular in PV applications, due to reduced production cost and time.

Silicon Production ++++++

1.
Quartz mining

Mined at quartz deposits, mostly in rocks.

2.
Refining quartz grit to metallurgical silicon

Metallurgical Silicon (9899% Si) is produced in an arc furnace by mixing quartz grit, coal, chips and coke
at a high temperature.

During this stage, an electrode is placed into an arc furnace. Then the furnace is loaded with SiO and chips,
coke and carbon in the form of a coal. The mixture reacts at 1800, consuming 13 kWh. Some intermediate
reactions also take place in the furnace. Resulting reaction can be expressed as follows:
2

SiC(solid) + SiO (solid) Si(solid) + SiO (gas) + CO(gas) (1)

2

Metallurgical silicon produced with such method contains 9899% Si, 12% Fe, u, , , , Cr, Cu, Mg,
Mn, Ni, Ti, V, Zn and others.

3.
Chlorination

Trichlorosilane is usually produced by hydrochlorination (chlorination) of silicon: metallurgical silicon

interacting with hydrogen chloride or mixture of gases containing hydrogen chloride at 260400C.
Trichlorosilane synthesis is followed by side reactions forming silicon tetrachloride and other chlorosilanes
as well as metal halides i.e. ll , l , FeCl etc. Reactions producing silicon chlorosilanes are reversible
and exothermic:
3

Si(solid) + l(gas) SiHCl (gas) + H (gas) (2)

Si(solid) + 4l(gas) SiCl (gas) + 2 (gas) (3)
3

At the temperature higher than 300C TCS is almost completely absent in the resultant products. To increase
the TCS output the process temperature is decreased slowing down the reaction speed (3). Catalysts (copper,

iron, aluminum or others) are used for speeding up the reaction (2). For instance, by introducing up to 5%
of copper into initial silicon, the TCS content in the resultant products reaches 95%. at 265C.
TCS synthesis takes place in a fluidized bed reactor (FBR) continuously fed from above with MgSi powder
with the particle size of 0.011 mm. Fluidized bed of particles 200600 mm thick is created by a counter
flow of hydrogen chloride delivered to the reactors bottom part at a rate of 18 cm/s. This provides the
transition of heterogeneous chemical engineering process from diffusion area to kinetic one. As the process
is exothermic, in order to stabilize the mode in the specified interval, the heat is intensively removed with
temperature strictly controlled at different levels of the fluidized bed. Hydrogen chloride consumption and
pressure inside the reactor are monitored apart from temperature.
TCS output is significantly affected by water and oxygen content in raw materials. By oxidizing silicon
powder, these admixtures cause the formation of dense SiO beds at its surface preventing silicon and
hydrogen chloride interaction and thus decreasing TCS output. For instance, at increasing content
in HCl from 0.3 to 0.4% TCS output decreases from 90 to 65%. Therefore, hydrogen chloride and silicon
powder are thoroughly dried and purified from oxygen before TCS synthesis.
2

Steam-gas mixture formed during the TCS synthesis is delivered to a cooling area where it is quenched to
40130C, forming solid particles of admixtures (iron and aluminum chlorides etc.,) in the form of a dust.
These particles are then removed along with parts of unreacted silicon and polychlorides (Si Cl ) by filters.
After removing the dust (explosive material), steam-gas mixture at the temperature of 70C. is delivered
to a condenser. SiHCl and SiCl (boiling temperatures 31.8 and 57.2C correspondingly) are separated from
hydrogen and HCl (boiling temperature 84C). The mixture obtained after condensation consists mostly
of TCS (up to 9095%), the remainders are STC, which is then separated during the rectification process.
n

2n + 2

STC, formed during the separation, is then used for producing silicones, quartz glass and TCS by additional
hydrogenation in presence of a catalyst.

4.
Distillation

TCS is purified from admixtures (boron, phosphorous, carbon) via rectification (separation) and transition
to nonvolatile or complex compounds. Final testing of silicons quality is conducted by measuring its
specific resistance.

Produced TCS contains many admixtures hard to remove. The most effective way is rectification; however,
it is hard to conduct a deep and complete purification from admixtures having various physical and chemical
natures by using rectification alone. Therefore, some additional measures are taken in order to improve
purification efficiency.
For instance, admixtures, which are difficult to remove by crystallization (boron, phosphorous, carbon),
require the most thorough TCS purification. That is why these trace substances are transitioned
to nonvolatile and complex compounds in order to increase the purification efficiency. E.g. for boron
removal, TCS steams are flowed through aluminum chip at 120C. Chip surface consumes boron and
it is almost completely removed from TCS. Accessory aluminum chloride is then sublimated at 220250,
and separated by fractional condensation.
Silver, copper or stibium can be used apart from aluminum. Adding copper to aluminum allows removing
arsenic and stibium from TCS simultaneously. Introducing phosphorous pentachloride or oxychloride also
increases efficiency of borons removal from TCS. Nonvolatile complex compounds of phosphorous with
boron l l or l l , thus form, which are then removed during by rectification. Transition
of boron to nonvolatile compounds can also be carried out by adding trityl chloride (or trimethylamine,
ethane nitrile, amino acids, ketone etc.) to TCS, which leads to formation of a boron compound like
( )3l , which is then removed by rectification.
5

Removal of boron-containing admixtures is also conducted by adsorption in reactors filled with aluminogel
and other gels (TiO , Fe O , Mg(OH) ) with following TCS rectification.
2

For phosphorous removal, TCS is saturated with chloride, transitioning phoshourous trichloride
to pentachloride. Upon adding aluminum chloride to the solution, nonvolatile compound l ll forms,
which is then removed by rectification.
5

The purity of purified TCS is tested by infrared spectroscopy, chromatography, and measuring
of conductivity types and rates of silicon test patterns, obtained from TCS samples.
The content of remaining trace substances in TCS after purification should not exceed, mass %: boron
310 , phosphorous 110 , arsenic 510 , carbon (in the form of hydrocarbon) 510 .
-8

-7

-10

-7

The residual content of donors according to electrical measurements should provide silicons specific
resistance of n-type and 5000 Ohmcm, and by -type crystal acceptors no less than 8000 Ohmcm.

5.

Polysilicon production
Reduction

Polycrystalline silicon is produced by reducing purified TCS from gas into a solid substance
by depositing TCS at heated rods seeds made of monosilicon.
Reduction of purified TCS and resulting production of polysilicon are conducted in the
hydrogen atmosphere
SiHCl (Gas) + H (Gas) Si(Solid) + 3HCl(Gas) (5)
3

at the surface of heated silicon rods seed rods with diameter of 410 mm (sometimes up to 30 mm),
produced by growing from pedestal, cutting after Czochralski process or after special modes of reduction
process, molding. Some technologies use tabular rods (15 mm thick and 30100 mm wide) with greater
deposition surface instead of cylindrical ones. High quality polysilicon is used to grow roods. Rod surface
undergoes ultrasonic cleaning, etching in acids mixture (e.g., HF+ + HNO ), washing off and drying. Seed
rods for producing high quality polysilicon should meet strict purity requirements: they should have donor
resistance of >700 Ohmcm and boron resistance of >5000 Ohmcm.
3

Rods are used to produce electric heaters (e.g. of U-shape) and they are heated by conducting electric
current. Current strength is raised in proportion to rods diameter increase.
The choice of conditions for TCS hydrogen reduction is made based on optimal interrelation of the
following process parameters:
Equilibrium degree of SiHCl transition to Si, crystal structure of produced rods;
Process temperature;
Energy demands;
: SiHCl molar ratio;
Silicon deposition rate.
Optimal conditions for reduction process are considered to be as follows: temperature of 11001150C, :
SiHCl molar ratio within 515, TCS feed density of 0.004 mol/(hcm ). When the rod temperature is lower
than optimal, the degree of TCS turning to STC increases lowering silicon output. Increasing temperature
leads to significant increase of energy consumption. At optimal molar ratio of : SiHCl = 515, rods have
a dense fine-crystalline structure and relatively even surface. An uneven surface forms beyond the limits
of this ratio; rods structure becomes coarse-crystalline with pinholes, which result in swirling and fusion
spitting when polysilicon is melted in the process of crystal growing.

The quantity of rods installed in various commercial reactors ranges from 2 t 80with the length of each rod
to 3.2 m and finishing diameter 150250 mm. Due to mutual heating of the rods, silicon deposition rate

in multirod units is higher than in 2-rod installations; rate of rod diameter growth reaches 0.5 mm/h, with
energy consumption of 3000 kWh/kg.
In order to increase the purity of produced silicon, hydrogen is thoroughly purified, reactors are made
of special steel and their surface is protected against the impact of gas environment by introducing additional
quartz (silicon) jars, separating the reaction medium from reactor walls. Protective films like
polychlorosilane are good for protecting reactor walls.

Producing polysilicon from monosilane SiH4

Producing polycrystalline silicon rods by thermal decomposition of monosilane SiH is conducted by similar
method at 8001000C. Hydrogen formed during the decomposition process SiH (Gas) Si(Solid) +
2 (Gas) has a high degree of purity and is used in auxiliary production. Polysilicon obtained with this
technology is purer than silicon produced by TCS reduction.
4

Silicon is extracted from SiCl and SiJ by reducing STC with zinc of by thermal dissociation of tetraiodide.
4

Produced polycrystalline rods are broke into pieces for easier load of crucibles or cut into cut-to-length
sections before using them for growing monocrystals by Czochralski technology. For crucibleless zone
melting rods are grinded to the required diameter. Removing surface layers enriched with admixture and
gases also prevents spitting of silicon from the molten zone.
Modern process flows of polysilicon production include regeneration and reutilization of all components and
products of reduction reactions (pyrolysis) thus improving technical and economic parameters of the
process, decreasing silicon production cost and making the process more eco-friendly.

This polysilicon deposition process is also used for producing polycrystalline pipes with carbon expanders.
Due to high purity and durability quartz pipes are replaced with polycrystalline pipes in furnaces for high
temperature processes (higher than 1200C) used in technologies for producing semiconductors and
microelectronics. Silicon pipes are not subject to settling or any other deformation for some years
of exploitation despite of continuous temperature circulation ranging between 900 and 1250C, while quartz
pipes have a limited lifetime in the same conditions.

6.
Monosilicon production
Producing monosilicon by non-crucible zone melting
Polysilicon is purified from volatile and nonvolatile admixtures by pulling polysilicon rods through melted
zone in vacuum.
Polysilicon rods are exposed to crystal purification by vacuum zone melting for producing high quality
silicon. In addition, apart from crystal purification of silicon from nonvolatile admixtures (mostly acceptors),
it is significantly cleaned out of volatile donors due to their evaporation from the molten zone.
E.g. 15 passes through the molten zone at a rate 3 mm/min, result in producing p-type conductivity
monocrystals of silicon with residual concentration of the admixture of less than 1013 cm and specific
resistance (to boron) more than 104 Ohmcm.
-3

Producing monosilicon by Czochralski technology

Silicon monocrystals are mostly produced with Czochralski method (up to 8090% of electronic industry
consumption) and, to a lesser extent, by non-crucible zone melting.
Czochralski method implies growing monocrystals due to atoms transitioning from liquid to gaseous state
at the border of their division.

In regard to silicon, this process can be characterized as a single component liquid-solid growing system.
The growth rate (V) is determined by the number of points on the growing crystals surface for attachment
of the atoms coming from the liquid phase and by specifics of transitioning at the division borders.

Equipment for growing monocrystalline silicon

The unit consists of the following blocks:

Furnace, including a crucible (8), crucible container (14), heater (15), power supply
(12), high-temperature chamber (6) and insulation (3, 16);
Crystal pulling device including a seed rod (5), seed rotation (1) and fixing devices,
crucible rotation and lifting device (11);

Device for controlling atmosphere composition (4 gas inlet,, 9 exhaustion, 10

vacuum pump);
Control module, consisting of microprocessor, temperature detector and ingots
diameter data transmitter (13, 19), input devices;;
Auxiliary devices: inspection window 17, 2 shell.

Process technology
High quality rotating seed monocrystal is sunk into a silicon melt. Melt polysilicon is produced in a crucible
within the inert atmosphere (of argon at depression of ~104 .) at the temperature insignificantly exceeding
silicon melting point = 1415C. The crucible is rotated in the direction opposite to that one of the
monocrystal in order to mix the melt and minimize inhomogeneity of temperature allocation. Growing
at depression allows partial removing of volatile admixtures from silicon melt due to their evaporation and
decreasing the formation of silicon monoxide powder, which in case of getting into the melt results
in defects in the crystal and may disturb monocrystalline growth.

In the beginning of the monocrystal growth, a part of seed monocrystal is melted to liquidate sections with
increased density of mechanical tensions and defects in it. Then monocrystal is gradually pulled from the
melt.
There is highly efficient automated equipment developed and widely used for production of silicon
monocrystals with Czochralski technology. It provides reproductible production of dislocation free
monocrystals with diameter of up to 200300 mm. Upon increasing the load and the diameter
of monocrystals, their production cost decreases. However, in the melts of larger mass (60120 kg) the
character of convection streams becomes more complicated, creating additional troubles for providing
required parameters of the material. Apart from that, at larger masses of the melt the cost decrease becomes
insignificant due to high costs of quartz crucibles and decrease of crystal growth rate because of troubles
caused by removal of crystallization latent heat. Thus, in order to further increase the process efficiency and
decrease the volume of the melt used for growing crystals, the units of semicontinuous growth began
to be developed intensively. Additionally, crucibles in these units are continuously or periodically loaded
with silicon without cooling the furnace, e.g. by feeding the melt with liquid phase from other crucible,
which in its turn also can be periodically or continuously fed with solid phase. This improvement of the
Czochralski method allows reducing the cost of grown crystals for tens of percent. Along with that,
it is possible to conduct growing from the melts of small and permanent volume. This simplifies regulating
and optimizing the convection streams in the melt and liquidates crystals segregative inhomogeneity caused
by adjusting the melts volume during its growth.

7.
Multisilicon production
Multisilicon is produced by bottom-up direct crystallization of ingots from a melt in round quartz crucibles
due to cooling crucibles bottom.
Before each melting, the furnace is cleaned and prepared and polysilicon is loaded into the crucible. After
the furnace is closed and air removed, argon is fed and the heater is turned to the mode of raw materials
preheating. Then the heater is turned to the melting mode. After the fusion, the melt is soaked for some
period in order to average the composition and temperature, and then crucible with the melt are cooled
determining the process of crystal growth and being know-how of equipment suppliers.
During the crystallization, the temperature of the silicon melt in the crucible (container) gradually decreases
making the crystallites grow in the same direction, gradually expanding and pushing out smaller crystallites.
Such production provides producing ingots with columnar texture and of required seed size, and allows
moving the gases dissolved in the melt to the upper (cut) section. The size of the seed grown with this
method can reach 510 mm in a section perpendicular to the growth direction. Produced ingots (with mass
of 120200 kg) are cut into smaller ones, which are then wire-cut into wafers for solar elements.
Silicon Times, 20122013

Properties
Dispersion

30%

Melting temperature1417C
Boiling temperature2600C
Density2.33

g/cm

Chemical element
Silicon (lat. Silicium), Si, belongs to the group IV of Mendeleev system with atomic number 14 and atomic mass of 28.086.
Dispersion
Silicon is the second most abundant element in the crust after oxygen.
Crystalline structure
Dark grey silicon crystals form diamond cubic side-centered latitude.
Melting temperature
Melting temperature: 1417, boiling temperature: 2600.
Heat conductivity
Silicon has a high thermal conductivity ranging within 84126 W/mK (at 25)
Transparency

Silicon is transparent for long wavelength infrared ray; Refraction index (for = 6 um) equals to 3.42.

Silicon hardness
7.0 according to the Mohs scale, 2.4 henry/m (240 kgf/mm ) according to the Brinell scale.
2

Resiliency
Fragile material: elastic modulus equals 109 henry/m (10,890 kgf/mm )
2

Specific heat capacity

800 j/(kg), or 0.191 cal/(gdeg).
Thermal expansion coefficient
2.3310 , becomes negative when lower 120 .
-6

-1

Magnetic behavior
Silicon is diamagnetic.
Dielectric permittivity
= 11.7. Shows how much the electric field is weakened by dielectric, quantitatively characterizing the dielectric property
to polarize in the electric field.
Electrical resistivity
Pure silicon scarcely carries electrical current (specific resistance equals to 2.3105 Ohmcm). Depending on admixture,
it can have not only electron (n-type) but also hole conduction (p-type), that vary within wide limits.

Chemical valence
Silicon is 4-valent in compounds
Chemical activity
Crystalline silicon is low active, but more active than carbon.
PROPERTIES DETAILS

Silicon types

(Monosilicon)(Multisilicon)
(Polysilicon)
The difference between silicon types lies within the degree of regular and proper order of silicons crystalline structure. Thus, silicon can be
classified in accordance to size of crystals it consists of. For example, monosilicon ingot is a silicon crystal, whereas polysilicon ingot has more
crystals than multisilicon ingot has.
Name

Notation

Seed size

Growing technology

Mono-

sc-Si

> 100mm

Czochralski (CZ), floating zone

Name

Notation

Seed size

Growing technology

melting (FZ)

Multi-

mc-Si

1100mm Molding, thin sheets, bands

Poly-

pc-Si

1um1mm Chemical vapor deposition

Amorpho
us-

c-Si

< 1u

Plasma deposition

Monocrystalline silicon
Monocrystalline silicon differs from the polycrystalline modification by its crystalline structure oriented in a certain
crystallographic plane. Monosilicon has a homogeneous crystalline structure without grain line junctions (which
is noticeable even in appearance).

Ordered arrangement of silicon atoms in the silicons monocrystalline latitude creates an accurate zone structure. Each atom
of silicon has 4 electrons at the outer shell. Electrons of neighbor atoms form couples, belonging to both atoms
simultaneously; thus, each atom has 4 bonds with neighbor atoms.
Behavior of monosilicon is well predicted, however, because of low growing speed and production process complexity
it is the most expensive silicon type. Monosilicon is the base for modern electronics. High requirements regarding purity
and structure perfection are imposed to it. Concentration of electrically active alloying admixtures usually lies within 1013
1018 cm , electrically active background impurities less than 1015 cm , electrically inactive impurities less than
10181019 cm . The main types of structure defects are so called microdefects. As a rule, these are small dislocation loops
and clusters of own and admixtures point defects.
-3

-3

-3

Multicrystalline silicon
Multicrystalline silicon holds the intermediate position between poly- and monocrystalline silicon by size and amount
of crystals. Growing of silicon multicrystals is a lot easier than monocrystals, which makes their cost lower. However, the
quality of multicrystals in comparison with monocrystal is lower as well because of lots of grain junction lines
of monocrystal containing in multicrystal.
Grains line junctions create additional defect levels in a semiconductor band gap, being local centers with high speed
of recombination, which leads to decrease in general lifetime of minority carriers. In addition, grain line junctions decrease
productivity blocking carriers current and creating bridging ways for current going through p-n transition.
To prevent high recombination losses at grain line junctions, the grain size should be at least few millimeters. This condition
also means that the measurements of a seed will exceed the thickness of the solar element, which will decrease resistance
to carriers current and general length of border zones in the solar element. Such multicrystalline silicon is widely used
in commercial manufacture of solar elements.

How to produce polysilicon?

Polycrystalline silicon
Polysilicon is high-purity silicon containing less than 0.0001% of admixtures, which consists of significant number of small
crystalline seeds chaotically oriented in respect to each other.
In essence, technical silicon is also polycrystalline; however, in order to prevent confusion, conception polycrystalline
silicon is applied only to especially pure semiconductor silicon.
Polysilicon is the purest form of industrially produced silicon and the main material for microelectronics and solar
energetics; it is an intermediate product obtained by purification of technical silicon with chlorides, which is used for
producing mono- and multicrystalline silicon.
Now polysilicon is distinguished into electronic (semiconductor) grade (with less than 11010% of admixtures) and
solar grade (with 1105% admixtures content).
Most of silicon in the world is produced in a form of grey-colored wires with grained dendritical surface. In the wires
center, there is a mono- and polysilicon fuse of round or square section with diameter (by side) of 810 mm. Densepacked crystallites in form of short needles with section of less than 1 mm grow from the fuse transversely to generatrix.
Polysilicon is a raw material for production of more advanced silicon types like multicrystalline silicon (multisilicon) and
monocrystalline silicon (monosilicon), and in some application areas can be used in its pure form.

Amorphous silicon
Amorphous silicon is a brown or brownish black moisture-retentive powder chemically more active than crystalline silicon.
At usual temperature reacts with fluorine directly forming fluorine silicon SiF ; at high temperature reacts with almost all
non-metals and with many metals.
4

Silicons coordination number equals four, and that is why each atom of silicon is bond with four neighbor atoms.
In crystalline silicon, these tetrahedral structures last in a wide range forming well-arranged crystalline latitude. In contrast,
in an amorphous structure the order in atomic positions is limited to short range. Most likely, atoms form continuous
random chains. In addition, not all the atoms in amorphous silicon are coordinated 4 times. In connection with unarranged
character of the material, some atoms have cutoff chains. Physically these cutoff chains are defects in the continuous
random chain and change silicon properties significantly.
In semiconductor instruments, amorphous silicon is usually used in a form of thin films deposited on a base coat. Layers
of hydrogenized amorphous silicon with most of the cutoff chains filled with atoms of hydrogen are used in solar elements.
Such silicon shows the best coefficient of light conversion to electric energy in comparison with pure amorphous silicon.

Market overview
Polysilicon production volume
> 501050< 10 K Tons

PV installation
EstablishedFuture trends markets
RussiaNorwayEstoniaCanadaUSABrazilGreat BritainGermanyUkraineSpainItalyAlgeriaGCCSouth AfricaChinaIndiaMalaysiaSouth
KoreaJapanTaiwanAustralia

Overview
Throughout the duration of a long period of time semiconductor (SC) industry was almost the only consumer of high-purity
silicon: as of the beginning of the century its consumption share was 6080% of global polysilicon market:
In 2006 the dynamic increase of demand for Photovoltaics (PV), driven by European governments subsidizing
PV installations mainly in Germany, Spain and Italy led to increased necessity of new capacities construction. First results
of the investment boom appeared in 2008, when the first factories were launched. At that moment, more than 60 projects
were still on different development stages. Capacity utilization exceeded 90%.
Capacities growth was happening via expanding already operating factories and launching of new manufacturers, using both
traditional and innovative technologies.
Innovative technologies were following the trends of lowering production costs and producing higher quality polysilicon
in order to increase its performance in PV.
In 2008-2009, financial and economic crisis had a bad impact on polysilicon investors. In 2008 capacity utilization has
reached 75%, in 2009 it has only reached 51%. The year of 2010 was a record for PV: the previous year installations
volume increased more than twice, which became a reason for consumption growth.
The world electronics market started to recover, and the wafers production in volume and value has grown by 40%. Despite
new factories were launched with total capacity of 56,000 tons, the capacity utilization rate has increased to 60%.

Market volume and market value

History
Market Volume Dynamics, K Tons
200620072008200920102011201220132014201520162017201820192020020040060080047627810513918324532743256
974543464953576065Show all
Solar Electronic
Polysilicone Market Value Dynamics, Bn $
20062007200820092010201120122013201420152016201720182019202001020303.5510.48.89.79.710.210.711.312.112.9
13.714.515.116.22.97.84.63.3322.93.95.27.19.612.717.122.7Show all
Solar Electronic
We can easily distinguish the growth rate of solar quality polysilicon market from the electronic one. The latter is quite
stable as opposed to the solar grade polysilicon. Photovoltaics, driven by European feed-in tariffs was booming. The prices
that solar companies were able to pay for raw materials were incredible. In 20062008 polysilicon production was a highly
profitable and promising market. This period was exactly the one when new capacities were being built.
Solar installations sponsorships were not indefinite: they started drying in 2009 when the growth rates declined.
Nevertheless, photovoltaic industry is still supported by European governments and is considered one of the most efficient
renewable energy sources. The industry has to start being self-regulatory and find new points of balance being costs, prices
and contractual terms. Solar installations growth will stabilize at a rate of ~33% per year. Electronic grade polysilicon
production growth will stagnate at a rate of ~7% focusing more on product quality.

We can see the boom in market value in 2008 that is a result of increased capacity. As we were saying before, market value
will reach ~$20 Bn in 2015 and reach ~$40 Bn in 2020. We expect the solar market to repeat the situation of 20082010 in
20171018. The LTA will still exist on the market with flexible prices though. Spot prices will be negotiated monthly,
following the same path as solar modules do.

Forecast
Based on various forecasts and official information provided by worldwide governments we expect the polysilicon market
to grow 24% CAGR from 2012 to 2020. The growth will be driven by solar grade silicone demand increase which will be
led mainly by PV installations growth announced by governments worldwide.
Increased attention to renewable energy in general and solar energy in particular is explained by the number of various
factors:

Government goals to reduce carbon dioxide burst

Increased climate change concerns

High dependence on oil and gas energy provided by certain countries only

Fossil energy price growth

Necessity to meet future energy needs via new energy sources

Cost of renewable energy generation and solar energy in particular is becoming more compatible
with traditional energy generation.

According to various data sources, in 3040 years the word economy will need additional 5000 GW of electrical generators
capacity, which is approximately twice more of the present capacity of all worlds energy stations.

At the present moment, despite global financial and economic crisis, photovoltaic industry is still developing. The main
drivers for this development are governmental programs and federal financing. This development results in:

Increased number of self-sustainable private houses in metropolitan areas

Usage of solar batteries for street and traffic lighting in Europe

Solar plants are being built in order to meet energy need of nearby towns

According to global Industry analysts, global silicone market will reach $17.2 billion in 2017. While PV is the main driver
for such a growth, we need to understand that the 20 GW installed in 2011 are just 1% of the total amount of electricity that
was generated in 2011 by all sources. The forecast for 2012 of long term growth estimates range from 20% to 30%.
As we can see from value graph above, the period of chaotic growth rates finishes after year 2012. The growth speed will
slowly decrease. Nevertheless, solar market will experience double-digit growth until 2020. Electronic market,
as mentioned previously, will be growing slowly at 7% rate.

Prices
Polysilicon prices are normally distinguished into:

contract (LTA) prices which are stated in long-term contracts between producers and consumers;

spot prices that are used if polysilicon is sold without preliminary arrangements.

The main indicator of price dynamics are spot prices which represent polysilicon as a commodity and which are the main
indicator of supply/demand balance.
To illustrate the historical price development below is the graph for spot and contract price development from 2006 until
today.

Polysilicon spot prices, Dec 17 , $/kg

th

Polysilicon

Max

1 grade
quality

22.00 14.50 15.550 -0.20 -1.27%

2 grade
quality

15.40 13.50 14.710 -0.21 -1.41%

st

nd

Min

Av.

As we can see from the graph, pricing boom affected both solar and electronic grade of polysilicon equally until the
beginning of 2009. After 2009, solar grade silicone prices experienced a moderate decrease as opposed to electronic grade
silicone. The process illustrated supply/demand balance in the industry, which became stable in the electronic industry
as opposed to solar.
Considering the fact that that semiconductors (electronic) industry is quite stable at the moment we will consider prices for
electronic grade silicone to balance around $250 per kg. From now onwards we will analyze solar grade silicone prices
because:

the weight of electronic grade silicon in the overall polysilicon is small but stable;

the prices are considered to be stable;

as a result, the influence of electronic grade silicone on the market dynamics is negligible.

Forecast
As we have mentioned in the paragraph above for forecast purposes we will use solar grade silicone prices. In order
to do so we have to consider both spot and LTA prices.
LTA prices are based on long-term relationships between contractors and are often confidential. During the period
of undersupply, contract prices were a good indicator of market dynamics. Contract prices were aimed to hedge volatility
risk of spot prices enabling both contractors to forecast revenues and costs better.
However, currently all producers are cancelling their long term contracts as the market is becoming a better tool for price
formation. Therefore, we can identify the trend that LTA prices will lean towards spot prices and polysilicon becoming
a commodity.

Polysilicon market in Russia

Russia is by far not the leading importer of polysilicon. The majority of silicone imported to the country is metallurgical
silicon of less than 99.99% purity. Russian largest aluminum producer is the major importer of this type of product.
Nevertheless, as we can see from the graph, semiconductors industry owns 10% share in value in overall silicon imports.
High technology, space and military technology has no option besides importing, as Russia does not produce high purity
silicon.
Besides, as can be seen from the graph, the volume/value ratio of imported product is very low, determined by high price,
especially for electronic grade silicone.

Import of silicone to Russia

Country

Volume, tons

Price, $

Total

510.8

24,931.
9

China

473.7

17,212.0

Germany

18.1

3,103.7

ROTW

6.2

2,111.4

USA

5.9

992.1

Ukraine

3.2

398.2

Country

Volume, tons

Price, $

Canada

1.5

61.7

Italy

0.9

1,001.1

Netherla
nds

0.8

22.5

Norway

0.7

29.1

Imported silicon, 2011

Product

Volume, %

Price, $

MgSi

97.50%

70%

Solar grade

2.30%

20%

Electronic
grade

0.20%

10%

In total, the amount of imported polysilicon of more then 99.99% purity is 510 tons in 2011 out of which 473.7 is imported
from China. The main reason for working with Chinese vendors is mainly price.
Silicon Times, 20122013

Two growth techniques for mono-crystalline silicon: Czochralski vs Float

Zone.
Silicon is the most abundant solid element on earth, being second only to oxygen and it makes up more than
25% of the earths crust. However, it rarely occurs in elemental form, virtually all of it is existing as compounds.

In this lecture the question will be answered how very pure sand (SiO2) is converted into mono-crystalline silicon
and later on into silicon detectors. After a description of the different growth techniques for mono-crystalline
silicon with special interest in the material used in this work it is shown which kind of detectors have been
used and how they have been produced.
The material requirements for the manufacturing of silicon particle detectors used for high energy
physics applications have to meet two basic demands: high resistivity and high minority carrier lifetime. A
very high resistivity (> l KOhm/cm) is needed in order to fully deplete the detector bulk with a thickness of about
200 - 300 um by an adequate voltage below about 300 V. Together with the demand for a reasonable price and a
homogeneous resistivity distribution, not only over a single wafer but also over the whole ingot, Float Zone
silicon is the best choice of material and is therefore exclusively used for detector applications today. Further
requirements for detector grade silicon are often a high minority carrier lifetime and a very low bulk generation
current in order to avoid detector noise.
However, these requirements should not be taken too strictly for particle detectors that will be exposed to
severe radiation levels since already after small radiation fluences the lifetimes are reduced by orders of
magnitude and therefore the good initial lifetime qualities are of no use any more. In the search for radiation
harder material and in order to perform radiation tests on an as wide as possible range of material also silicon
grown by the Czochralski has been investigated in this lecture. While for the epitaxial technique the price and
the substrate problem might rule out largely its application as detector material the Czochralski method could
become of interest for the production of radiation hard material if it is possible to make high resistivity (> 1
KOhm/cm) CZ commercially available.

In this section the production of silicon with the two growth techniques mentioned above will shortly be reviewed
with special interest in the high resistivity silicon production and the possibilities of defect engineering
respectively the controlled incorporation of impurities into the crystal.

Czochralski silicon (Cz)

The vast majority of the commercially grown silicon is Czochralski silicon due to the better resistance of the
wafers to thermal stress, the speed of production, the low cost and the high oxygen concentration that offers the
possibility of Internal Gettering. The industrial standard crystals range in diameter from 75 to 200 mm, are
typically l m long and of < 100> orientation. In the following a short review is given.
Standard CZ
The Czochralski method is named after J. Czochralski, who determined the crystallisation velocity of metals by
pulling mono- and polycrystals against gravity out of a melt which is held in a crucible. The pull-from-melt method
widely employed today was developed by Teal and Little in 1950 . A schematic diagram of a Czochralski-Si
grower, called puller, is shown in Fig. 2.1. The puller consists of three main components:
1. a furnace, which includes a fused-silica crucible, a graphite susceptor, a rotation mechanism (clockwise as
shown), a heating element, and a power supply;
2. a crystal-pulling mechanism, which includes a seed holder and a rotation mechanism (counter-Clockwise);
and
3. an ambient control, which includes a gas source (such as argon), a flow control and an exhaust system.

The Czochralski method begins by melting high purity polysilicon (SGS) with additional dopants as required for
the final resistivity in the rotating quartz crucible. A single crystal silicon seed is placed on the surface and
gradually drawn upwards while simultaneously being rotated. This draws the molten silicon after it which
solidifies into a continuous crystal extending from the seed. Temperature and pulling speed are adjusted to first
neck the crystal diameter down to several millimetres, which eliminates dislocations generated by the seed/melt
contact shock, and then to widen the crystal to full diameter.
During the production process the quartz crucible (SiO 2) gradually dissolves, releasing large quantities of oxygen
into the melt. More than 99% of this is lost as SiO gas from the molten surface, but the rest stays in the melt and
can dissolve into the single crystal silicon. Another impurity, however with smaller concentrations, that is also
introduced into the melt by the production process itself is carbon. The silicon monoxide evaporating from the
melt surface interacts with the hot graphite susceptor and forms carbon monoxide that re-enters the melt. As the
crystal is pulled from the melt, the impurity concentration incorporated into the crystal (solid) is usually different
from the impurity concentration of the melt (liquid) at the interface. The ratio of these two concentrations is
defined as the equilibrium segregation coefficient k0 =Cs/c1 where Cs and C1 are the equilibrium concentrations
of the impurity in the solid and liquid near the interface, respectively.

Figure 2.1: Schematic setup of a Czochralski crystal puller

Oxygen is always the impurity with the highest concentration in CZ silicon. Typical oxygen and carbon
concentrations are [O] 5 - 10 10 ^17cm-3 and [C] 5 - 10 10^15cm-3, respectively. The solubility of O in Si is
10^18 cm-3 at the melting point but drops by several orders of magnitude at room temperature, hence there is a
driving force for oxygen precipitation. Furthermore the high oxygen concentration can lead to the formation of
unwanted electrically active defects. These are oxygen related thermal double donors (TDD) and shallow

thermal donors (STD) which can seriously change the resistivity of the material. However, oxygen has also good
properties.
Oxygen acts as a gettering agent for trace metal impurities in the crystal (Internal Gettering) and it can pin
dislocations which greatly strengthens the crystal. Oxygen precipitates in the wafer core suppress stacking
faults, and oxygen makes the Si more resistant to thermal stress during processing. This is the reason why CZSi is used for integrated circuit production, where there are many thermal processing steps.
However, the most important property of a high oxygen concentration from the point of view of this work is the
improved radiation hardness. The main problem for the application as detector grade material arises from the
resistivity of CZ silicon. Due to contamination with boron, phosphorus and aluminum from the dissolving quartz
Crucible the highest commercially available resistivity is about l00 Ohmcm for n-type and only slightly higher for
p-type material. Therefore standard CZ silicon is not suitable for detector production. However, first experiments
to compensate the natural p-type background doping by adding a small quantity of phosphorus to the melt have
been performed.
Magnetic Field Applied Cz (MGZ)
MCZ may be the future standard CZ technology since todays approaches to solve the challenge of the 300 mm
and later on also the 400 mm crystal diameter are based on this technology The method is the same as the CZ
method except that it is carried out within a strong horizontal (HMCZ) or vertical (VMCZ) magnetic field. This
serves to control the convection fluid flow, allowing e.g. with the HMCZ method to minimise the mixing between
the liquid in the center of the bath with that at the edge. This effectively creates a liquid silicon crucible around
the central silicon bath, which can trap much of the oxygen and slow its migration into the crystal. Compared to
the standard CZ a lower oxygen concentration can be obtained and the impurity distribution is more
homogeneous. This method offers also the possibility to produce detector grade silicon with a high oxygen

concentration. Since the technology is still a very young one, it is hard to get such material with reproducible
impurity concentrations on a commercial basis. However, a first test material of 4 Kcm p-type with an oxygen
concentration of 7 - 8 l017 cm-3 and a carbon concentration below 2xl016 cm-3 was obtained.
Continuous Cz (CCZ)
With the CCZ method a continuous supply of molten polycrystalline silicon is achieved by using a double quartz
crucible. In the first one the crystal is grown and in the second one, connected to the first one, a reservoir of
molten silicon is kept, that can be refilled by new polysilicon during the growth process. This allows for larger
crystal length and improves the throughput and operational costs of the CZ grower. Furthermore the resulting
single crystals have a uniform resistivity and oxygen concentration and identical thermal history. In combination
with the magnetic field method the Continuous Magnetic Field Applied CZ technique (CMCZ) offers the
possibility to grow long and large diameter CZ. However, silicon produced by this technology has so far not been
used for radiation damage experiments.

Float zone silicon (FZ)

Float-zone silicon is a high-purity alternative to crystals grown by the Czochralski process. The concentrations of
light impurities, such as carbon and oxygen, are extremely low. Another light impurity,nitrogen, helps to control
microdefects and also brings about an improvement in mechanical strength of thewafers, and is now being
intentionally added during the growth stages.
The float zone method

The float Zone (FZ) method is based on the zone-melting principle and was invented by Theuerer in 1962. A
schematic setup of the process is shown in Fig. 2.2. The production takes place under vacuum or in an inert
gaseous atmosphere. The process starts with a high-purity polycrystalline rod and a monocrystalline seed crystal
that are held face to face in a vertical position and are rotated.

Figure 2.2: Schematic setup for the Float Zone (FZ) process
With a radio frequency field both are partially melted. The seed is brought up from below to make contact with
the drop of melt formed at the tip of the poly rod. A necking process is carried out to establish a dislocation free

crystal before the neck is allowed to increase in diameter to form a taper and reach the desired diameter for
steady-state growth. As the molten zone is moved along the polysilicon rod, the molten silicon solidifies into a
single Crystal and, simultaneously, the material is purified. Typical oxygen and carbon concentrations in FZ
silicon are below 5 10 15 cm-3. FZ crystals are doped by adding the doping gas phosphine (PH 3) or diborane
(B2H6) to the inert gas for n- and p-type, respectively. Unlike CZ growth, the silicon molten Zone is not in contact
with any substances except ambient gas, which may only contain doping gas. Therefore FZ silicon can easily
achieve much higher purity and higher resistivity.
Additionally multiple zone refining can be performed on a rod to further reduce the impurity concentrations. Once
again the effective segregation coefficient k plays an important role. Boron, for example, has an equilibrium
segregation coefficient of k0 = 0.8. In contrast to this phosphorus cannot only be segregated (k0 = 0.35) but also
evaporates from the melt at a fairly high rate. This is the reason why on the one hand it is easier to produce
more homogeneous p-type FZ than n-type FZ and on the other hand high resistivity p-type silicon can only be
obtained from polysilicon with low boron content. Dopants with a small k0 like Sn can be introduced by pill
doping - holes are drilled into the ingot into which the dopant is incorporated - or by evaporating a dopant layer
on the whole ingot before the float zoning process.