"But what ... is it good for?"—

Engineer at the Advanced Computing Systems Division of IBM,

1968, commenting on the microchip.

RECENT TRENDS IN 300-MM PLASMA EQUIPMENT

Georgy K. Vinogradov

FOI Corporation 6-38-28, KAMI-ASAO, ASAO -KU, KAWASAKI, 215-0021 JAPAN

vinogradov@foi.co.jp

Introduction

Semiconductor equipment manufacturing is a

multi-billion dollar business strongly connected to

the markets of semiconductor devices and consumer

electronics through the fastest feedback loops

keeping them altogether. Since the manufacturers are

targeting exclusively their investors and customers,

“confidential” design and process characteristics are

zealously hidden from competitors and the public.

Industrial exhibitors screen visitors not to allow

competitors chances for a glance at demo machines

or process data. Therefore, it is rather difficult to

obtain solid references reviewing modern 300-mm

equipment.

A whole set of 300-mm equipment is available in

the market and several production lines made

operational start in the last three-four years

worldwide: Trecenti Technology (Hitachi), UMC,

Infineon, SONY Semiconductor, Micron, TSMC,

Philips Semiconductor-STMicroelectronics, IBM,

Texas Instruments, Intel, Samsung and others [1].

However, 300-mm machines shipped to R&D labs

operate conventionally with 200-mm wafers saving

on existing equipment and well-integrated processes.

Therefore, there is typical conservatism: no major

change in process technologies. Devices manufactures

demand similar, if not “same”, processes and

conditions for 300-mm tools. Therefore, the 300-mm

are mainly just scaled-up versions of 200-mm tools. In

this sense, there is no trend between 200- and 300-mm.

Different logic works for new materials, for example,

copper and a variety of Low-k. New processes must be

developed and delivered with new equipment. Such

model was practiced during the 200-mm span

exhibiting remarkable change in the balance of process

development loads between the device and equipment

manufacturers. Process development cooperation is to

be highly respected by all parties for mutual benefits.

Still, at even higher level of complexity, the

development moves mainly by a thorny path of trial

and errors. There are very few things among the basic

scientific ideas, which can be affected. In contrast,

there are much more unexpected and unclear

material/design factors affecting process technology.

That is why we discuss here more finalized

commercial solutions and less laboratory and patent

curiosities.

Plasma Etching Equipment

For the last three decades etching equipment is the

arena driving advanced plasma technology

development [2]. Other equipment: PVD, CVD,

cleaning, ashing, etc. use plasma techniques usually

after etchers in terms of plasma sources. Chemical

aspects of numerous plasma technologies represent

special subjects.

Every etcher accumulates substantial engineering

and process know-how rather than science. There is

enough room for R&D art just because of a great

complexity of machines and processes and

abundance of not yet tested combinations of well-

and hardly-known ideas.

There is basic similarity between different kinds of

plasma processing equipment considering process

chambers. Any chamber has its plasma source

converting electric power into the electric fields

generating fast free electrons, which start ionization,

excitation and dissociation of gas particles, surface

bombardment, not saying about a number of

secondary phenomena like modification and erosion

of surfaces, deposition and material transfer, ageing,

dust particles, arcing. The set of the basic processes is

determined by the regularities of plasma physics and

chemistry. Then what constitutes the main differences

of plasma sources and process chambers?

There are two kind of plasma processes: isotropic

and anizotropic, and hence, two kinds of process

chambers applying isotropic and anizotropic active

media to wafers. The isotropic, is almost electrically

neutral highly-dissociated gas or free radicals, which

does not generate electric currents into the wafers.

The second one, anizotropic, is electrically highly

active. It is aimed to supply not only neutrals but also

electrically charged particles (positive ions) with

predetermined kinetic energy and directionality onto

the wafer. We consider anizotropic etchers.

A pitfall to be avoided is pure technical consideration

of industrial equipment, which is basically highly

competing commercial products, where economical

parameters are as important as technical merits.

Recently, 300-mm equipment reveals several

converging tendencies.

Firstly, the price margins for different classes of

etchers are becoming narrower well before approaching

any maturity under very tense economic conditions.

Secondly, the huge price gap between high-end

etchers and, for example, advanced single-wafer ashers,

is shrinking.

Thirdly, hardware platforms in terms of basic

mandatory components are becoming similar: ashers

are getting turbo pumps, some etchers –

high-temperature platens, modular design leads to more

customizable tools.

Fourthly, large companies, manufacturing wide range

of equipment, acquire small, one-sector players. It

changes not only competition but also affects hardware

design, in particular cluster tools.

The tendencies are supported by unavoidable

maturity curve driving semiconductor industry into the

21st century, where the number of equipment

manufacturers is decreasing.

300-mm etchers

There are two approaches to classify etching tools:

considering plasma sources or principal applications.

Since they are meaningful and interconnected, both

should be taken into account.

The majority of 300-mm machines available in the

market are scaled up from 200-mm with minor

modifications. Therefore, the parallel-plate capacitive

systems from two major vendors: AMAT (Applied

Materials, USA) and TEL (Tokyo Electron, Japan)

are still the main working horses well known to

device manufacturers. LAM Research Corporation

(USA), known for pioneering pancake inductive

plasma sources, manufactures RIE capacitive tools

for oxide/dielectric etch as well. Microwave etchers,

for example, Hitachi’s ECR, also entered into the

300-mm competition. It is in the minority and not

considered here.

In spite of incredibly high R&D activity in the last

two decades around flat inductive plasma sources,

they failed yet to replace capacitive etchers.

Moreover, they practically did not enter into the most

difficult market sector of oxide etch.

AMAT and LAM sell inductive etchers for

polysilicon, metal, and some minor application.

AMAT, having the largest number of printed patent

pages and claims on inductive sources, is now

manufacturing tools with a dual-coil source mounted

above a flat or dome-shaped chamber roof.

There are several improvements in the process

chambers aimed to increase selectivity and suppress

particles and contamination. Silicon roofs were

inserted in several etchers. Not only roofs but also

silicon or SiC/SiOC walls of oxide etchers were

introduced in some cases. Hot wall of the process

chambers and better temperature control is a must as

well. However, such particular solutions do not touch

anyway basic structures of plasma processing units.

In order to reveal definite trends in the plasma

sources, we have to consider much longer

development term and some basic principles.

Why capacitive is good?

This long-liver has two principal advantages over

other discharges. First: it is the anisotropic discharge

exited by very uniform (equipotential metal electrode)

electric field perpendicular to the wafer surface. The

anisotropic etch was discovered in this discharge.

Second, the capacitive plasma generated by a uniform

flat electric capacitive sheath is perfectly conformal to

the wafer. It is the only discharge mechanism, which

does it so simply, precisely and cheaply. It is why

capacitive discharges were the first and main plasma

processing tools from early seventies. This is the secret

of their long life. This is a reason why other etchers

using a variety of discharges have so called RF-bias. In

fact, it is never only a ‘bias’ for accelerating ions.

Almost always it is a capacitive discharge near the

wafer surface. It not only affects the kinetic energy of

existing positive ions, it generates own sheath plasma.

The mandatory RF-bias on the wafer substantially

decreases but does not eliminate the difference between

plasma sources. The intrinsic features of parallel-plate

capacitive discharges, flatness and very narrow

inter-electrode gap, could not be realized with inductive

sources until very recently.

Where is a limit?

Capacitive discharges have some drawbacks. First,

they are very sensitive to the gas pressure and power.

This is because of a self-consistent nature of the double

electric sheath: it tunes itself depending on the gas

pressure and the RF power and tends to shrink or

spread along the surface instead of changing local

parameters. Second, they are all non-uniform at the gap

edge. Therefore, so-called guard rings are always used

in etchers but do not always mitigate the edge

non-uniformity. Third, capacitive discharges are

difficult to sustain belo w about 10-mTorr gas

pressure, where the recent process trend goes to.

Conventional capacitive discharges used one and

then two RF frequencies for three decades. One,

higher frequency, which is usually about 13.56 MHz,

is applied to the top electrode (“anode”), while

another, lower frequency, about 0.1-4 MHz, is

applied to the bottom wafer electrode (“cathode”).

The basic trends of capacitive discharges are as

follow. First add one more frequency in order to

separately excite main plasma discharge and the bias

sheath: done in seventies. Second, increase the higher

frequency in order to maintain the discharge at lower

and lower pressures. 300-mm capacitive etchers

come to 27; 60; and even 100 MHz RF. This way

capacitive etchers achieve the pressures limit down to

about 20 mTorr and higher plasma density.

The limit of this frequency trend is the wavelength

or transmission-line problem. A hundred megahertz

RF field has about 0.7 m quarter wavelength in free

space. Therefore, RF voltage along the radius of a

300-mm wafer surface is changing essentially. This is

a source of electrical non-uniformity, and hence,

wafer’s damage even without mentioning the edge

problems. Hence, it is hardly possible, that simple

capacitive parallel plate structure continues beyond

300 mm.

Oxide etch

Advanced silicon oxide etch is performed by

capacitive etchers since this process uses

fluorocarbon deposition mechanism accompanied by

intensive ion bombardment. Oxide etchers are

strong-bias narrow-gap machines with particularly

low gas residence time (down to a few milliseconds)

in order to achieve best quality. There were no new

narrow-gap etchers except capacitive in the market yet.

However, this status is about to change.

High-density plasma low-pressure etchers

Inductive coupled plasma (ICP) sources have a long

history and variety of applications such as powerful

free-radical or ion generators and atmospheric plasma

torches. ICP sources easily generate high-density

plasmas. The problem of their application to the wafer

processing is their basic radial non-uniformity, which

determines development trends.

About a decade ago, flat inductive plasma sources

occupied plasma laboratories []. It was accepted as a

long-awaited revolution in plasma equipment. Flat roof

inductive etchers have been built and tested. However,

they were large-gap plasma sources, because, of the

intrinsic radial non-uniformity. It appeared later that

large spiral inductive coils add substantial azimuthal

non-uniformity caused by transmission-line and

capacitive coupling effects. No one of them entered the

“prohibited garden” of advanced oxide etch.

Very popular in laboratories single-turn ICP sources

are not suitable. A narrow-gap single-coil inductive

discharge produces about a circular footprint on the

wafer. That is why there were several patents issued for

multi-coil sources. In a dual-coil source, the center

smaller coil increases plasma density, which otherwise

is too low for a 300-mm wafer size. However, two coils

are not enough for a narrow-gap etcher.

A large gap is unavoidable with several separate

coils, because the coils do not strictly define the

position of plasma generation in the flat-roof systems

and strongly interact. A multi-turn spiral coil does not

generate a sheath-plasma like a capacitive electrode just

because of different kind of symmetry of the electric

fields in these discharges.

Consequently, the main problem and development

trend of inductive plasma sources was toward

high-density uniform plasma in a narrower gap.

Major ICP sources (LAM, AMAT) have at least

about 100 -mm discharge gap. However, the problem

was solved in a GROOVYTM ICP plasma source

developed by FOI (Future Oriented Instruments,

Japan) Corporation.

GROOVY ICP: narrow-gap inductive plasma

source

GROOVY ICP source consists of two (200-mm)

to three (300-mm) or four (450-mm) inductive coils

built into the outside grooves in the flat dielectric or

silicon roof and surrounding inside groves enclosing

separate inductive plasma toroids. The separation of

plasma generation zones between the grooves allows

one to independently control plasma toroids avoiding

collapsing into one plasma generation area. The

collapse inevitably happens at the flat roofs of

conventional inductive sources.

The flat grooved roof makes a discharge gap of

about 40 mm, which is typical for advanced

capacitive oxide etchers. GROOVY generates

stable discharge in a wide pressure range of about

(0.5-1000) mTorr. This source is considered as the

only inductive candidate for below 100 nm deep

oxide etch. It shows an aspect ratio of about 20,

selectivity to resist of about 7, for 80 nm via etch

with typical uniformity of about 3% (range) over

300-mm wafers.

Low-k materials

The well-known merits of ICP sources over the

capacitively coupled are their ability to operate at

very low RF-bias voltage, very low gas pressure, and

high plasma density. Inductive discharges have wider

process range required by a variety of Low-k materials

from inorganic Si-containing films to organic polymers.

So far, there is nothing special in etching of organic

Low-k but essential problems of selectivity to resist and

difficult combination of hard mask and intermediate

inorganic layers: SiO2, SiOC, or SiC. Hence, Low-k

processing tools must be capable for different

chemistries in the same chamb er. One of the options is

hydrogen or ammonia plasma for etching and/or

material stabilization. Sidewall damage is the main

issue for the materials sensitive to free-radicals.

Consequently, low-pressure, low-temperature, and

short-time processing is the major trend.

Wide -gap inductive etchers

Inductive (wide-gap) etchers for polysilicon and

metal compete successfully with traditional capacitive

etchers. Etching of metals having non-volatile halides

(FeRAM, MRAM applications) represents very special

art, where the machine maintenance issues play

substantially important role in evaluation, because of

heavy deposition in the chambers. So far, there is no

market for 300-mm equipment, but is expected to

emerge in the near future.

Electrical damage issues

Electrical damage to processing devices mainly

come from process /discharge non-uniformities, which

are determined, in turn, by capacitive current/voltage

non-uniformities in the process chambers. Such

non-uniformities have several causes: edges of

capacitive platen/guard rings; non-uniformity of

conductive parts; asymmetric capacitive return

currents; parasitic local discharges and arcing;

capacitive coupling from ICP coils; etc.. These

problems persisted from 200-mm prototypes but are

even worsened because of increases size of pumping

ports, windows, and shutters .

Moreover, 300-mm equipment needs better

discharge uniformity than 200-mm for the following

reason. It is well known that a wafer provides a short

electric path between dielectric structures and

therefore, damage voltage applied to the devices

depends on the maximum voltage (potential)

differences. At the same constant gradient of

electric potential along the wafer surface, the voltage

difference between two distant points may become

about 1.5 times higher for 300-mm wafers. Therefore,

200-mm damage sensors are not perfectly suitable for

300-mm damage tests.

One of the common trends mitigating discharge

non-uniformities is dielectric or semiconductor

insulation of the process chamber walls contacting

active plasma. It is troublesome and difficult art

affecting RF capacitive bias and discharge ignition.

Process chambers have oxidized aluminum, alumina,

SiC, Si, SiOC, or quartz liners of high purity.

The discharge ignition damages can be more

severe in 300-mm equipment because of much higher

power of RF generators needed for sustaining

discharge and ignition in substantially DC/RF

insulated chambers. This is a concern for both

capacitive and inductive plasma sources. ICP etchers

use capacitive currents for discharge ignition as well.

The tradeoff between the discharge stability and

ignition is efficiently solved in capacitively balanced

transmission-line inductive plasma sources like

Gamma-Dipole Resonator [3].

Process chambers, especially pumping baffles,

become more critical for discharge uniformity

because of large wafer size: center to edge distance

along the wafer is becoming much longer than

edge-to-baffle distance. New materials in the chambers

bring about some new electrical effects as well.

Overall, there are a lot of particular discharge and

material improvements along the 200-300 transition to

improve process quality and drive the semiconductor

industry to higher efficiency.

Conclusion

The main development trend of 300-mm plasma

equipment, particularly etchers, is further increasing

competition between capacitive and inductive plasma

sources. Capacitive plasma sources increase excitation

frequency in order to attain low-pressure high-density

plasma conditions, while inductive sources are going to

essentially shrink the discharge gap and cover the low

gas residence time range previously occupied

exclusively by the capacitive counterpart. Both systems

seem to be converging at the very limit for advanced

oxide etchers.

Wide-gap inductive sources are successfully

replacing capacitive systems in polysilicon and metal

etch. Microwave systems are yet in minority and will

hardly occupy a noticeable place in the 300-mm market

in the near future.

There is yet plenty of work to do for mitigating

chamber and process non-uniformities along the way to

the “well-designed tools and well-designed processes ”.

References:

[1] D. Vogler, Solid State Technology, Oct. 2002, 27.

[2] J.W. Coburn, AVS 49th Symposium, Nov. 2002,

Abstracts, PS-MoM1.

[3] G.K. Vinogradov, Plasma Sources Science and

Technology, 9, 400 (2000).