Hoffman D.M., Singh B., Thomas J.H. (Eds). Handbook of Vacuum Science and Technology

Подождите немного. Документ загружается.

^24 Chapter S.I: High-Vacuum-Based Processes: Sputtering

during magnetron sputtering. A solution to this problem was proposed by Win-

dows [14], who developed a magnetron with unbalanced magnetic fields. This al-

lows electrons to stream away from the cathode plasma, which tended to pull ions

along toward the sample. Large substrate bias currents can then be measured at

the sample, consistent with bias sputtering or reactive sputter deposition.

5.1.3

MAGNETRON APPLICATIONS

Magnetrons are the most commonly used sputtering tool for the sputter deposi-

tion of thin films. Compared to RP or DC diodes, the voltages are more approach-

able (500V, not several keV) and the discharge currents can be huge. In addition,

the technology scales to very large dimensions: cathodes of lengths of several me-

ters are not unconmion. These cathodes might operate at many 10s of kW, with

ion currents of

to 100 amperes.

Magnetrons are used for mostly metal film deposition on a variety of sub-

strates. One industry that relies heavily on sputter deposition is the semiconduc-

tor industry, in which a large fraction of the metal layers used for interconnects

(on-chip wiring) use sputtering. The type of sputtering system used is described

as a "single-wafer tool" in which only one wafer is present in the deposition

chamber at a time. This is contrasted to a batch tool in which there might be many

wafers experiencing the same process concurrently. Single-wafer processing has

become the norm for the semiconductor industry partly due to the increasingly

large size of the wafers (200 mm and heading up) and the value of the sample,

which might approach $50,000 per wafer as the wafer nears completion. If any-

thing went wrong with a batch process, potentially $1 million in wafers could be

lost in a single run, whereas with a single-wafer tool the damage is contained to a

single sample.

Single-wafer tools generally operate with multiple process chambers and a cen-

tral handler and load-lock. An example of a single-level system is shown in Fig-

ure 8. The process modules might contain sputter deposition processes (magne-

trons),

or they could also be etch or Chemical Vapor Deposition (CVD) chambers.

A second large user of sputter deposition technology is the automotive indus-

try. With the advent in the 1970s and 1980s of plastics in automobiles, sputtering

has been broadly used for the metallization of these parts. Most of this is decora-

tive,

but in some cases it is also used to protect underlying parts from corrosion.

A third large user of sputtering technology is the glass-coating industry. Mag-

netron sputtering is used to put reflective, darkened, or decorative coatings on the

5.1.3 Magnetron Applications

625

Rg.8.

Load lock for wafer

cassettes

Module

Schematic of

multiple-level integrated processing sputter tool.

large windows used in office buildings. The cathodes in these cases can be many

meters long. The control of the film thicknesses in these applications is very crit-

ical: The human eye is very sensitive and can pick up intensity changes due to

thickness variations on the order of tens of angstroms.

Another large application for sputtering is the deposition of metal films on

plastic sheets. There are numerous uses for these metallized plastics, perhaps the

most common being on the insides of potato chip or peanuts bags. The volumes

of material here are very significant, and rolls of plastic sheet with lengths of

many miles are typically loaded into the vacuum system (known as a web coater)

for a single run.

Other applications of sputter deposition include decorative coatings for jew-

elry, watch bands, glasses frames, etc. These applications often use reactively

sputtered Ti-nitride because of its hardness and attractive golden color.

Another application is in the field of optical coatings for very-high-quality op-

tics applications. An application here is with laser ring gyroscopes, which are

used in inertial guidance systems. Other optical applications might include space-

craft windows and the various lithographic applications used in patterning semi-

conductors.

626 Chapter

5.1:

High-Vacuum-Based Processes: Sputtering

5.1.4

FUTURE DIRECTIONS IN SPUTTERING

In many ways, sputtering and sputter deposition are mature technologies. Sputter-

ing is widely used in a number of applications because it is versatile, inexpensive,

and well understood. Two recent areas of sputtering technology have been devel-

oped that may have implications for the future applications of sputtering. These

are (1) the ability to amplify or increase the sputter yield selectively for certain

types of cathodes, and (2) the in-flight ionization of sputtered atoms and the sub-

sequent deposition of films from ions, rather than neutrals.

5.1.4.1

Amplified Sputter Yields

Physical sputtering is a dynamic, coUisional process. As such, what usually mat-

ters is the mass of the target atom and the mass and energy of the projectile. Am-

plified sputtering [15] uses this effect to advantage. The target is composed of a

composite of the desired, to-be-sputtered species and generally

much heavier ele-

ment, such as Pt. Incident ions, usually Ar or Ne, are strongly reflected from these

heavy atoms, and rebound toward the near surface with most of their kinetic en-

ergy. This results in a sort of forward sputtering that can increase the emission

rate for the lower-mass species by as much as an order of magnitude. This is par-

ticularly useful for sputtering carbon, which has a very small yield. It is also use-

ful for other light atoms, such as Al or

Si.

Although there are no commercial appli-

cations for this technology yet, it creates an interesting capability for selectively

increasing the sputter yield.

5.1.4.2

ionized Sputter Deposition

Sputter deposition consists almost entirely of the deposition of neutral particles.

In most system geometries, this results in a wide angular distribution of the arriv-

ing particles: good for coating over steps and bumps on the surface, but poor for

filling up deep- or high-aspect ratio features. A recent development using the de-

position from mostly ions overcomes this limitation. By combining a dense, inert

gas plasma with a sputtering source, a large fraction of the sputtered flux can be

ionized in flight [16-18]. A simple bias on the sample then results in an electrical

acceleration of the sputtered ion species to the surface with controlled energy and

direction. This technique has been developed primarily for the semiconductor in-

dustry, but other applications, primarily in the area of low temperature reactive

sputter deposition, are being developed.

References

627

REFERENCES

p. Sigmund, Phys. Rev., 184 (1969) 383; 187 (1969) 768.

P. C. Zalm,

Surface

and

Interface

Analysis, 11 (1988) 1.

D. Hoffman, /

Vac.

Sci. Technoi, 8 (1990) 3707.

J. L. Cecchi, in Handbook of Plasma Processing Technology, edited by S. M. Rossnagel,

J. J. Cuomo, and

D. Westwood (Noyes, Park Ridge, NJ, 1989), chap. 2.

H. R. Kaufman and R. S. Robinson, in Operation of Broad Beam Ion Sources (Conmionwealth

Scientific, Alexandria, VA, 1987).

6. B. Chapman, Glow Discharge Processes (Wiley, New York, 1980).

S. M. Rossnagel and H. R. Kaufman, J.

Vac.

Sci. Technoi, AS (1987)

88-91.

8. J. A. Thornton and A. S. Penfold, in Thin Film Processes, edited by J. L. Vossen and W. Kern

(Academic Press, New York, 1978).

9. S. M. Rossnagel and H. R. Kaufman, J.

Vac.

Sci. Technoi, A5 (1987) 2276-2279.

10.

S. M. Rossnagel, J.

Vac.

Sci. Technoi, A6 (1988) 3049-3054.

11.

W. H. Class, Thin Solid Films, 107 (1983) 379-385.

12.

Darryl Restaino, IBM East Fishkill, private communication (January 1996).

13.

S. M. Rossnagel, J.

Vac.

ScL Technoi, A6 (1988) 19-24.

14.

B. Windows and N. Savides, J.

Vac.

Sci Technoi, A4 (1986) 196.

15.

S. Berg, I. V. Katardjiev, C. Nender, and R Carlsson, Thin Solid Films, 241 (1994)

-8.

16.

M. S. Barnes, J. C. Forster, and J. H. Keller, U.S. Patent 5,178,739 (January 12, 1993).

17.

S. M. Rossnagel and J. Hopwood, Applied Physics Utters, 63 (1993) 3285-3288.

18.

S. M. Rossnagel and J. Hopwood, J.

Vac.

Sci Technoi, B12 (1994) 449-453.

CHAPTER

5.2

Plasma Etching

V. Patel, Ph.D.

Sarnoff Corporation

5.2.1

INTRODUCTION

The concept of using plasma processes for etching materials was first applied to

semiconductor processing in the 1960s for removing organic photoresist materi-

als.

Over the years, plasma etch processes have replaced most selective wet etch-

ing processes in semiconductor processing because of its environment friend-

liness,

cleanliness, uniformity, ease of automation and ability to allow faithful

transfer of submicron circuit patterns from the photolithographic masks to the sur-

face of semiconductor wafers. Today, plasma etch processes are critical to the fab-

rication of semiconductor integrated

circuits.

However, the development of plasma

etch processes has progressed on empirical grounds without much scientific foun-

dation, because of intense production demand and the complexity involved.

This chapter is devoted to the plasma etch technology with focus on various re-

actor configurations. In addition, some of the basic plasma concepts related to

etching and the basic plasma etching requirements are also reviewed.

5.2.2

REVIEW OF PLASMA CONCEPTS APPLICABLE

TO ETCHING REACTORS

5.2.2.1

Basic Physical Phenomena Occurring in RF Discharges

In plasma etching, a glow discharge is used to create chemically reactive species

from the feed gas that reacts with the material to be etched and creates volatile re-

$25.00 All rights of reproduction in any form reserved.

628

5.2.2 Review

Plasma Concepts Applicable

Etching Reactors

629

action by-products. A glow discharge is a plasma (defined as a partially ionized

gas composed of

ions,

electrons, and a variety of neutral species) that exists in the

pressure range of interest for plasma etching (0.1 mtorr to 10 torr) and contains

approximately equal concentrations of positively charged particles (positive ions)

and negatively charged particles (electrons and negative ions). A particular useful

aspect of plasma is its ability to enable high-temperature-type chemical reactions

to be performed at low substrate temperatures.

A simplified configuration of a basic plasma reactor used for etching applica-

tion is illustrated in Figure

1(a).

The semiconductor substrate(s) is normally placed

on the electrode to which the RF power is applied. The chamber walls are kept

at the ground potential and act as a grounded electrode in many configurations. A

radio frequency (RF) field of sufficient magnitude is applied across the electrodes

(typically through a coupling capacitor) to cause the feed gas to break down and

become ionized. A central bright region is created that is known as the plasma

region. The plasma region maintains charge neutrality (equal number of positive

and negative ionic species) and is equipotential (no electric field in this region).

Fig.l.

RF Power Supply RF Electrode "er Wall

Coupling Capacitor ^ „, . ^ . ^

^ ^ ^ Sheath Regions

(a)

RF Electrode

-Vdc

Ground Electrode

(Chamber WaU)

(b)

A simplified configuration of

typical plasma etch reactor in (a) and a typical time-averaged

potential profile across the reactor in (b).

630 Chapter 5.2: Plasma Etching

The potential of this region is known as plasma potential (V^) and is the highest

potential in the system. The region that separates the plasma region from the elec-

trodes (also chamber walls) is known as sheath or "dark space." Almost all po-

tential drop occurs in this region. In capacitively coupled systems, a negative DC

potential (V^DC) is induced on the powered electrode in order to balance the time-

averaged electron and ion current on electrodes. A simplified diagram of the

time-averaged potential profile across the reactor is shown in Figure 1(b).

In a plasma etching apparatus, the plasma is initiated by accelerating some free

electrons within the chamber volume by the means of applied electric or mag-

netic field. The feed gas is weakly ionized as a result of inelastic collisions be-

tween neutrals and energetic electrons. Electrons have smaller mass and can re-

spond to the RF field. Electrons can gain energy directly from acceleration in the

applied RP field and continue to gain energy through elastic collisions until an in-

elastic collision occurs. Electron impact ionization produces electron-ion pairs

that sustain the discharge against electron loss, which is normally dominated by

diffusion from the plasma volume to the electrodes and the reactor walls. Elec-

trons have sufficiently high energy and mobility to escape the plasma volume.

However, the coulombic forces between the electrons and positive ions create a

field, referred to as the "ambipolar field," resulting in a self-limiting process and

thereby maintaining charge neutrality in the plasma. This ambipolar field (assum-

ing electropositive plasma with negligible amount of negative ions) causes the

positive ions to diffuse much faster than the free diffusion condition and at the

same time slows the diffusion of the electrons.

Several gas-phase processes—such as impact ionization, dissociative ioniza-

tion, dissociation, recombination, relaxation, electron attachment, and resonance

charge transfer—also occur in the plasma. The plasma chemistry responsible for

creating reactive radicals and ions through these processes is largely determined

by the electron energy distribution function (EEDF), because it is a quantity on

which the rates of the electron impact processes depend [1]. Acceleration of sec-

ondary electrons emitted from the wafer/electrode surface by the sheath and en-

ergy gained by electrons in the oscillating sheath edge by so-called surf-riding

mechanism can also play significant role in determining the exact nature of EEDF

[2].

The steady-state electron energy distribution is a balance between energy

gain and energy loss as a result of inelastic collisions or in certain cases electron

beam plasma interaction.

The extensive use of plasmas to activate etch processes derives from two major

features of low-temperature nonequilibrium discharges. The first of these is the

existence of energetic electrons with average energies in the range of 2 to 10 eV

(—lO"* to 10^ Kelvin temperature) in the plasma volume. These electrons break

bonds to form chemically active etchant species or their precursors. The electrons

are also responsible for ionization, which sustains the discharge and creates ions.

These ions are often essential to the etch process. The second important feature of

plasma for etching applications is the acceleration of ions at the plasma boundary

5.2.2 Review of Plasma Concepts Applicable to Etching Reactors

631

known as sheath. In many configurations, the ions are accelerated by the electric

field in the sheath region between the plasma boundary and wafer. This field can

accelerate ions, normal to the wafer surface, with typical energies in the range

of 50 to 1000 eV. The ion bombardment often results in mechanisms that allow

lithographic patterns to be etched anisotropically with little or no lateral removal

of material. This feature is essential to realization of increasingly high device

density found in modern integrated circuits, and is perhaps the major reason why

plasma-activated etching is used so extensively. It should be noted here that

ions are rarely the etchant, and neutrals are responsible for almost all reactive

etching [3].

SHEATH FORMATION

An unique feature of plasma is its ability to shield out internal charge concentra-

tions from externally applied electric fields. Sheaths (defined as regions in which

the quasi-neutrality condition does not hold) are formed at all grounded and pow-

ered surfaces representing boundaries for the plasma. The sheaths insulate the

plasma from electron loss. In RF dischaiges, the dynamics of the sheaths are

more complex than for DC discharges and generally are not understood. The ear-

liest studies on the RF sheaths were made by Butler and Kino [4] and since their

pioneering work, numerous papers have been published outlining improved math-

ematical treatments

[5,6].

An approximate description of a RF sheath based on physical reasoning is

shown in Figure 2. The ions cannot respond to the instantaneous field. Instead,

Fig.

s(t)

The sheath structure showing the positive ion density («,), the time-averaged electron density

(t\) and the instantaneous electron density (n J at one time in an RF cycle.

632 Chapter 5.2: Plasma Etching

they move through a quasi-neutral presheath and acquire a directed average ve-

locity that is of the order of ion sound speed (see the next section) at the sheath

edge

S,„

where the electron density is equal to the ion density at all times. The re-

gion between the wall and the

S,„

is called the ion sheath, because it is a region of

net positive charge on time averaged basis, although this does not hold during the

entire RP cycle [7].

As positive ions move to the wall, they gain momentum in the average sheath

field and conservation of ion flux causes the ion density to fall toward the elec-

trode/wall to a value,

rij^^,,

at the wall. Since the response of an ion to the RF field

can be neglected, the ion density in the sheath is stationary. In contrast, the elec-

trons respond to the changing voltage across the sheath, and their motion can be

characterized most simply by the assumption of a moving sheath edge. The den-

sity profile of the electron at the moving sheath is a steplike profile [8]. Since the

average ion and electron currents to the wall in capacitively coupled systems must

balance, the sheath collapses for a small part of the RF cycle (when the RF volt-

age is maximum) enabling the electrons to reach the electrode. After the field re-

versal of the applied RF voltage at the electrode, the electron sheath edge moves

away from the electrode. Thus, the RF field causes the sheaths to expand and de-

cay, modulating the sheath voltage and length. An electron in the glow near the

sheath boundary comes under the influence of the repulsive field at the edge of the

moving sheath and can be reflected back into the plasma. This phenomenon is of-

ten termed as the "surf-riding" mechanism [2] or stochastic heating of electrons

[9].

Electron reflection occurs at both electrode sheaths, and the electron velocity

after reflection depends on the sheath thickness and the applied voltage.

BOHM CRITERION FOR EXISTENCE OF A PRESHEATH

A smooth transition region exists between the quasi-neutral plasma and the sheath

and is known as the presheath region. This region has a relatively weak electric

field, compared to that of the sheath, that accelerates the ions from the quasi-neutral

plasma toward the sheath. This implies that the ions arriving at the sheath will ac-

quire a velocity, which is of the order of

the

ion acoustic speed in the presheath re-

gion. The existence of this enhanced velocity at the sheath edge and the resulting

criterion requiring the existence of a presheath was first demonstrated by Bohm

[10] based on energy and momentum conservation arguments. The necessity for

an ion-accelerating presheath has become known as the Bohm sheath criterion.

Chen [11] demonstrated the physical significance of this criterion by showing

that the acceleration of ions and repulsion of electrons in a typical sheath must be

such that the ion density decreases less rapidly than the electron density across

the sheath. In these classical theories, an electropositive gas was normally consid-

ered with equal concentration of electrons and positive ions in the bulk of the

plasma. The existence of negative ions has generally been ignored.

5.23 Basic Plasma Etching Requirements 633

For discharges containing electronegative gases, as is the case for most plasma

processing applications, the negative ions are present in copious amounts and their

concentration can exceed that of electrons [12]. This causes significant deviation

in the mathematical relations derived for electropositive discharges. The concept

of sheath, however, is not significantly changed for electronegative plasma, be-

cause negative ions experience only the time-averaged field and are repelled from

the sheath field. The negative ions do not play a significant role in the sheath dy-

namics. However, the presheath region in the electronegative plasma can be

significantly different from the presheath region in electropositive plasma be-

cause of the influence of negative ions on the presheath electric field. Braithwaite

[13] showed that as the ratio of negative ion concentration to electron concentra-

tion increases, the potential variation in the presheath becomes more rapid. For a

critical value of the ratio, an abrupt potential drop occurs between the sheath edge

and the plasma. This is accomplished by the formation of a "double layer" at the

plasma-sheath boundary [14]. This double layer plays an important part in many

plasma processing applications.

Negative ions are reflected by the sheath, and their generation and loss is dom-

inated by volume processes such as attachment balanced by detachment and ion-

ion recombination rather than by recombination at the wall. In an electronegative

plasma, the electron density (n^) can be much lower than the positive ion density

(n,),

and thus a much larger fraction of the random electron flux incident on the

sheath must overcome the sheath barrier in order to equalize the positive ion flux

to the electrode. This is ensured by a much lower sheath potential, and hence in-

duced DC bias.

5.2.3

BASIC PLASMA ETCHING REQUIREMENTS

Various physical and chemical processes occurring during plasma etching are ex-

tremely complex and not well understood. However, the basic processes respon-

sible for plasma etching can be described into

five

basic steps as shown in Figure 3.

These steps are as follows: (1) generation of reactive neutral and ionic species in

plasma; (2) diffusion of the reactive neutral species and acceleration of positive

ions through the sheath to the surface being etched; (3) reaction of these species

on the surface resulting in

volatile by-product; (4) desorption and diffusion of the

by-product from the surface into bulk plasma; and (5) removal of the by-products

from the etching chamber by vacuum pumps. No etching takes place if any of

these steps fails to occur. Note here that many reactive species can react readily

with a solid surface, but unless the by-product has a reasonable vapor pressure,

the etching cycle

ceases.

Also, the chemical reaction on the surface can be strongly

influenced by the mass, energy, and flux of the ions impinging on the surface.