for the VLSI Era
`Volume 1
`- Process Technology
`Second Edition
`
`} Silicon Processing
`
`
`
`8. Wolf and RN. Tauber
`
`APPLIED MATERIALS EXHIBIT 1041
`
`A.
`
`'3
`Page 1 of 14
`
`Page 1 of 14
`
`APPLIED MATERIALS EXHIBIT 1041
`
`

`

`D1SCLA1l\1ER
`
`This publication is based on sources and information believed to be reliable, but the
`authors and Lattice Press disclaim any warranty or liability based on or relating to the
`contents of this publication.
`
`Published by:
`
`LATTICE PRESS
`Post Office Box 340
`Sunset Beach, California 90742, U.S.A.
`
`Cover design by Roy Montibon, New Archetype Publishing, Los Angeles, CA.
`
`Copyright © 2000 by Lattice Press.
`All rights reserved. No part of this book may be reproduced or transmitted in any form
`or by any means, electronic or mechanical, including photocopying, recording or by any
`information storage and retrieval system without written permission from the publisher,
`except for the inclusion of brief quotations in a review.
`
`Library of Congress Cataloging in Publication Data
`Wolf, Stanley and Tauber, Richard N.
`
`Silicon Processing for the VLSI Era
`Volume 1: Process Technology
`
`Includes Index
`1. Integrated circuits-Very large scale
`integration. 2. Silicon. I. Title
`
`ISBN 0-9616721-6-1
`
`9 8 7 6 5 4 3 2
`
`PRINTED IN THE UNITED STATES OF AMERICA
`
`Page 2 of 14
`
`

`

`438 SILICON PROCESSING FOR THE VLSI ERA
`
`11.2 SPUTTER DEPOSITION FOR ULSI
`Sputtering- is a term used to describe the mechanism in which atoms are ejected from the surface
`of a material when that surface is struck by sufficiently energetic particles. It has become the
`dominant technique for depositing a variety of metallic films in VLSI and ULSI fabrication,
`including aluminum alloys, titanium, titanium:tungsten, titanium nitride, tantalum, and cobalt.
`Sputtering displaced the original PVD technique for depositing metal films (evaporation) for
`the following reasons:
`1. Sputtering can be accomplished from large-area targets, which simplifies the problem of de(cid:173)
`positing films with uniform thickness over large wafers.
`2. Film thickness control is relatively easily achieved by selecting a constant set of operating
`conditions, and then adjusting the deposition time to reach it.
`3. The alloy composition of sputter-deposited films can be more tightly (and easily) controlled
`than that of evaporated films.
`4. Many important film properties, such as step coverage and grain structure can be controlled
`by varying the negative bias and heat applied to the wafers. Other film properties (including
`stress and adhesion), can be controlled by altering such process parameters as power & pressure.
`5. The surface of the substrates can be sputter-cleaned in vacuum prior to initiating film depos(cid:173)
`ition (and the surface is not exposed again to ambient after such cleaning).
`6. There is sufficient material in most sputter targets to allow many deposition runs before target
`replacement is necessary.
`7. Device damage from x-rays generated during electron-beam evaporation is eliminated (al(cid:173)
`though some other radiation damage may still occur).
`As is true with other processes, however, sputtering also has its drawbacks. They include:
`1. Sputtering processes involve high capital equipment costs;
`2. Since the process is carried out in low-medium vacuum ranges (compared to the high vacuum
`conditions under which evaporation is conducted), there is greater possibility of incorporating
`impurities into the deposited film.
`3. Better step coverage can generally be achieved using CVD.
`In general, the sputtering process consists of four steps:
`1. Ions are generated and directed at a target.
`2. The ions sputter target atoms;
`3. The ejected (sputtered) atoms are transported to the substrate.
`4. Upon reaching the substrate they condense and form a thin film.
`Although it is of interest to note that sputtering can be conducted by generating the energetic
`incident ions by other means (e.g., ion beams), in virtually all VLSI and ULSI sputtering
`processes their source is a glow-discharge. The discussion of sputtering in this section will be
`limited to glow-discharge sputtering.4.5,6
`11.2.1 Introduction to Glow Discharge Physics
`The energetic particles used to strike target materials to be sputtered in ULSI sputter deposition
`systems are generated by glow-discharges.4,5 A glow-discharge is a self-sustaining type of plas(cid:173)
`ma (a plasma is defined as a partially ionized gas containing an equal number of positive and
`negative charges as well as some number of neutral gas particles). In Fig. 11-3 a simple de(cid:173)
`diode type system that can be employed to study properties of glow discharges used in
`
`Page 3 of 14
`
`

`

`ALUMINUM THIN FILMS AND PHYSICAL VAPOR DEPOSITION IN ULSI 439
`
`GLASS ENVELOPE
`
`I ~~I
`Vb-------~'------~
`
`ANODE
`(+)
`
`ANODE DARK
`SPACE
`
`I
`I
`
`I
`I
`
`CATHODE
`NEGATIVE
`I DARK SPACE GLOW
`II
`I
`
`FARAfJAY
`!!ARK
`SPACE
`
`POSITIVE
`COLUMN
`
`n+ (-)
`
`n- (---)
`
`VOLTAGE
`
`I
`
`II
`tt
`
`II
`II
`
`oo
`
`(c)
`
`Fig, 11·3 (a) Structure of a glow discharge in a de diode system. (b) Charged particle concentration in a
`glow discharge. (c) Voltage variation in a de diode glow discharge.
`sputtering is shown. It consists of a glass tube that is evacuated and then re-filled with a gas at
`low pressure. Within the tube there are two electrodes (a positively charged anode and a
`negatively charged cathode) and a de potential difference is applied between them.
`11.2.2 The Creation of Glow Discharges
`Consider the system shown in Fig. 11-3 to examine the case when a tube is filled with Ar at an
`initial pressure of 1 torr, the distance between the electrodes is 15 cm, and a 1.5 kV potential
`difference is applied between them. At the outset no current flows in the circuit, as all the Ar gas
`atoms are neutral and there are no charged particles in the gas. The full 1.5 kV is thus dropped
`between the two electrodes. If a free electron enters the tube (most likely created from the
`ionization of an Ar atom by a passing cosmic ray), it will be accelerated by the electric field
`existing between the electrodes (whose magnitude is: E = V/d = 1.5 kV/15 cm= 100 V/cm).
`The average distance that a free electron will travel at P = l torr before colliding with an Ar
`atom (i.e., the mean free path A) is 0.0122 cm (Chap. 3). Most electron-atom collisions are
`elastic, in which virtually no energy is transferred between the electron and gas atom. Such
`elastic collisions occur because the mass of the electron is much smaller than that of the atom.
`Thus, the minumum distance an electron must travel before it can undergo an inelastic collision
`(in which significant energy is transferred to the atom, either by the excitation of an atomic
`electron to a higher energy level, or to cause its escape from the atom) is about ten times A, or
`0.122 cm. If this is the minimum distance that must be traveled by electrons between inelastic
`collisions, there must be a significant number of electron path lengths in the range of 0.5-1.0
`cm. If a free electron travels 1 cm in the 100 V /cm electric field, it will have picked up 100 e V
`of kinetic energy. With this amount of energy, the free electron can transfer enough energy to an
`Ar electron to cause it to be excited or ionized. If this transferred energy E is less than the
`ionization potential (e.g., 11.5 eV< E < 15.7 eV for Ar), the orbital electron will be excited to a
`
`Page 4 of 14
`
`

`

`ALUMINUM THIN FILMS AND PHYSICAL VAPOR DEPOSITION IN ULSI 443
`
`SHIELD
`
`SHIELD
`
`b)
`
`--- ---c==
`
`\\:7i, 11-1------
`
`\:...,.
`- - - -----
`
`~ i,~--1-1-i-r-----
`, --
`- -----
`, ....... - - - -- --
`~
`
`PLASMA
`
`\
`
`'
`0~~"'
`
`Fig, 11·7 (a) Potential distribution in vicinity of cathode shield (b) Reducing rim effect by extending
`cathode shield. (c) Reducing rim effect by wrapping shield around the cathode.8 From L. Maissel and R.
`Giang, Eds., Handbook of Thin Film Technology 1970. Reprinted by permission of McGraw-Hill Book Co.
`dark space however, the ion production rate becomes reduced, and the voltage across the
`electrodes must rise to increase the secondary electron emission. Such a glow is known as an
`obstructed glow. In most practical sputter deposition systems the glow is obstructed. That is, in
`order to most effectively collect the sputtered material onto the substrate, the anode (on which
`the wafers are sometimes mounted) is placed as close to the cathode as possible (typically just
`far enough away to avoid extinguishing the negative glow).
`It is typically necessary to insure that sputtering is allowed to occur only at the front side of
`the target, as the backside contains cooling coils and attachment fixtures which are definitely not
`to be sputtered. To guarantee that no sputtering takes place except from desired surfaces, a
`shield of metal (at a potential equal to that of the anode) is placed at a distance less than the
`Crookes dark space at all other cathode surfaces (Fig. 11-7). Since no discharge will occur
`between two electrode surfaces separated by less than this distance, such shielding (termed
`dark-space shielding) is effective in preventing sputtering from unwanted cathode surfaces.
`11.3 THE PHYSICS OF SPUTTERING
`When a solid surface is bombarded by atoms, ions, or molecules, many phenomena occur. The
`kinetic energy of the impinging particles largely dictates which are the most probable events.
`For low energy particles (<10 eV), most interactions occur only at the surface of the target
`material. At very low energies ( <5 e V) such events are limited to reflection or physisorption of
`the bombarding species. For low energies which exceed the binding energy of the target
`material (5-10 eV), surface migration and surface damage effects can take place. At much
`higher energies (> 10 ke V), the impinging particles travel well into the bulk of the sample before
`slowing down and depositing their energy. Thus, such particles are most likely to be embedded
`in the target, and this mechanism is the basis of ion-implantation. At energies between the two
`extremes, two other effects also arise: 1) some fraction of the energy of the impinging ions is
`transferred to the solid in the form of heat, and lattice damage; and 2) another fraction of such
`energy causes atoms from the surface to be dislodged and ejected into the gas phase
`(sputtering).
`
`Page 5 of 14
`
`

`

`444 SILICON PROCESSING FOR THE VLSI ERA
`
`~~~
`)
`
`(
`
`'i
`I
`\ __ 1
`
`SPUTIERED ATOM
`
`,,. ,,. ,,.
`
`/-,
`----, '--I GAS MOLECULE,
`~ f O
`j NEUTRALIZED
`REBOUNDED FAOM
`THE TARGET SURFACE
`·~. BOMBARDING IONIZED
`
`GAS MOLECULE
`
`,,..---
`
`--
`
`''(-) /1'
`/ y.-,
`- \--/)
`' "It
`1 J
`0
`
`...
`
`\ I
`
`·--
`
`.
`
`\
`'
`
`\
`
`\
`
`\
`
`\
`
`\
`
`\
`
`\
`
`\
`
`\
`
`a I
`
`-,;. INOICAlE DlnECJION OF
`MOMEfHUM TRANSi-ER.
`
`' ) - --..\
`) SPUTIEAEO ATOM
`
`:
`"- .. /
`
`Fig, 11·8 (a) Binary collision between atom A and B, followed by a binary collision between atom B and
`C. (b) Collision process responsible for sputtering and fast neutral generation.
`11.3.1 The Billiard Ball Model of Sputtering
`The exact mechanisms which lead to the ejection of atoms under ion bombardment are not
`known, and a comprehensive theory of sputtering is not likely to be developed in the near future
`since many parameters are involved. These include the kinetic energy of the ions, lattice
`structure, and binding energy of lattice atoms. Some of the details, however, are reasonably well
`understood and can be aptly described with a relatively simple momentum-transfer model. G.K.
`Wehner, whose theoretical work first established a solid scientific basis for sputtering, often
`described sputtering as a game of three-dimensional billiards, played with atoms.9 Using this
`analogy, it is possible to visualize how atoms may be ejected from a surface as the result of two
`binary collisions (Figs. ll-8b and 1 l-8c) when a surface is struck by a particle with a velocity
`normal to the surface (e.g., atom A in Fig. 11-8b). Note that a binary collision is one in which
`the primary incoming particle (e.g., atom A) strikes a single object (e.g., atom Bin Fig. 11-8b),
`and gives up a significant fraction of its energy to the struck atom, while retaining the remaining
`fraction. As a consequence of the collision, atom B may leave the point of impact at an angle
`greater than 45°. If atom B then undergoes a secondary collision with atom C, the angle at
`which atom C leaves the secondary impact point may again be greater than 45°. Thus, it is
`possible that atom C can have a velocity component greater than 90° (and thus be directed
`toward the surface). As a result, there is a finite probability that atom C will be ejected from the
`surface as a result of the surface being struck by atom A.
`When the directions of sputtered atoms from the surface of polycrystalline materials (and
`most cathode materials in sputter applications are polycrystalline) are measured for the case of
`normal incidence, it is found that the ejected atoms leave the surface in essentially a cosine
`distribution. A cosine distribution, however, does not describe the sort of small-angle ejectjons
`that would be expected from the simple collision processes described above. Evidently in actual
`sputtering events, more than two collisions are involved, and the energy delivered by impinging
`ions during normal incidence is so randomly distributed that the effect of the incident
`momentum vector is lost. Note that the energy range of sputtered atoms leaving the target is
`
`Page 6 of 14
`
`

`

`ALUMINUM THIN FILMS AND PHYSICAL VAPOR DEPOSITION IN ULSI 445
`
`typically 3-10 e V, and the bombarding species also recoil from the cathode face with some en(cid:173)
`ergy. Thus, the target surface is a source of sputtered atoms and energetic backscattered species.
`For the case when the surface is bombarded by ions at an oblique angle (i.e., 45°-90°), there
`is a higher probability that the primary collision between incident ions and surface atoms will
`lead to sputtering events. Furthermore, oblique incidence confines the action closer to the sur(cid:173)
`face, and thus sputtering is enhanced. In cases of oblique bombardment, the incident momentum
`vector becomes important, and sputtered atoms are ejected strongly in the forward direction. In
`addition, the sputter yield (defined as the number of atoms ejected per incident ion), may be as
`much as an order of magnitude larger than that resulting from normal incidence by bombarding
`ions (Fig. 11-9). This effect also leads to faceting, which is covered in Sect. 11.7.4.
`11.3 2 Sputter Yield
`The sputter yield is important because it largely (but not completely) determines the rate of
`sputter deposition. Sputter yield depends on a number of factors besides the direction of
`incidence of the ions, and these include: a) the target material; b) the mass of bombarding ions;
`and c) the energy of the bombarding ions. There is a minimum energy threshold for sputtering
`that is approximately equal to the heat of sublimation (e.g., 13.5 eV for Si). In the energy range
`of sputtering (10-5000 eV), the yield increases with ion energy and mass. Figure 11-10 shows
`the sputter yield of copper as a function of energy for various noble gas ions. The sputter yields
`of various materials in argon at different energies is given in Table 11-2.9
`Several matters related to sputter yield should be noted. First, although the sputtering yields
`of various materials are different, as a group they are much closer in value to one another than,
`for example, the vapor pressure of comparable materials. This makes the deposition of
`multilayer or multi-component films much more controllable by sputtering. The details of sput(cid:173)
`tering from multi-component targets are discussed in the section on Process Considerations in
`Sputter Deposition. Second, since the bombarding ions are by no means monoenergetic in glow(cid:173)
`discharge sputtering, it is not necessarily valid to use the sputter yield values for pure metals if
`alloys, compounds, or mixtures are being sputtered. Tabulations of sputtering yields, however,
`are useful for obtaining rough indications of deposition or etch rates of various materials.
`
`SLC6~/
`
`I
`I
`I
`I
`I
`
`/
`
`6max
`ANGlf Of INCIDfNCE 161
`
`7T/2
`
`Fig, 11 ·9 Schematic diagram showing variation of sputtering yield with ion angle of incidence.
`
`Page 7 of 14
`
`

`

`446 SILICON PROCESSING FOR THE VLSI ERA
`
`Table 11-2 SPUTIER YIELDS FOR METALS IN ARGON (ATOMS/ION)
`
`Target
`
`At.WtJDens.
`
`100eV
`
`300eV
`
`600eV
`
`1000 eV
`
`2000 eV
`
`Al
`Au
`Cu
`Ni
`Pt
`Si
`Ta
`Ti
`w
`
`10.0
`10.2
`7.09
`6.6
`9.12
`12.05
`10.9
`10.62
`14.06
`
`0.11
`0.32
`0.5
`0.28
`0.2
`0.07
`0.1
`0.08
`0.12
`
`0.65
`1.65
`1.6
`0.95
`0.75
`0.31
`0.4
`0.33
`0.41
`
`1.2
`2.8
`2.3
`1.5
`1.6
`0.5
`0.6
`0.41
`0.75
`
`1.9
`3.6
`3.2
`2.1
`
`0.6
`0.9
`0.7
`
`2.0
`5.6
`4.3
`
`0.9
`
`11.3.3 Selection Criteria for Process Conditions and Sputter Gas
`The information gleaned from sputter yields and the physics of sputtering can also be applied
`toward an understanding of how process conditions and materials are selected for sputtering
`including: a) type of sputtering gas; b) pressure range of operation; and c) electrical conditions
`for the glow discharge. 4•8 That is, in purely physical sputtering (as opposed to reactive
`sputtering) it is important that the ions or atoms of the sputtering gas not react with the growing
`film. This limits the selection of sputtering species to the noble gases. Furthermore, argon is
`generally the gas of choice, since it is easily available (hence low in cost), and its mass is a good
`match to those of the elements most frequently sputtered (Al, Cu, Si, and Ti), giving adequate
`sputtering yields for these elements. The pressure range of operation is set by the requirements
`of the glow discharge (lower limit -2-3 mtorr for magnetron sputtering) and the scattering of
`sputtered atoms by the sputter gas (upper limit 100 mtorr). In addition, a desired goal of sputter
`deposition is to obtain maximum deposition rates. As a result, electrical conditions are selected
`
`c
`0
`~
`E
`0 ....
`~
`;,-.
`
`Almen & Bruce
`(1961)
`
`20-
`
`15
`
`Xe
`
`Copper
`
`10~
`
`· -~
`
`--------N
`
`0
`
`10
`
`20
`
`30
`
`40
`
`50
`
`60
`
`70
`
`keV
`
`Ion Energy
`
`Fig. 11-10 Sputtering yields of the noble gases on copper, as a function ofenergy. 10
`
`Page 8 of 14
`
`

`

`ALUMINUM THIN FILMS AND PHYSICAL VAPOR DEPOSITION IN ULSI 447
`
`to give a maximum sputter yield per unit energy. That is, as energy is increased, each energy
`increment gives a progressively smaller increase in sputter yield. This occurs because higher
`energy ions implant themselves, and thus end up dissipating a greater proportion of their energy
`via non-sputtering processes. The most efficient ion energies for sputtering are typically
`obtained for electrode voltages of several hundred volts.
`In general, the higher the current at the cathode, the higher is the film deposition rate, since
`more ions are striking the cathode (and thus cause more sputtering). The product of the cathode
`current and the electrode voltage gives the input power of the sputtering process. In magnetron
`sputtering, cathode current densities of 10-100 mA/cm2 at a few hundred volts are typical.
`From the foregoing discussion on the mechanism of sputtering one can see that sputtering is
`a highly inefficient process. In fact, - 70% of the energy consumed during the sputtering process
`is dissipated as heat in the target, and -25% by emission of secondary electrons and photons
`from the target. This heating can raise target temperatures to levels capable of damaging the
`target, associated vacuum components, or the material that bonds the target to the backing
`electrode. The target must therefore be cooled to avoid these problems. As we will discuss later,
`this is normally accomplished by bonding the target to a water-cooled Cu backing plate.
`11.3.4 Secondary Electron Production for Sustaining the Discharge
`Near the outset of the discussion on glow-discharges, it was observed that glow-discharges must
`be continuously provided with free electrons to keep them sustained. In most de-sputtering
`systems, the source of such electrons is secondary-electron emission from the target. An
`important mechanism that generates such secondary electrons is Auger emission, although other
`emission mechanisms are also responsible for their production. The Auger emission process
`occurs according to the sequence of events shown in Fig. 11-11. Target electrons require a
`minimum amount of energy q<P to escape into free space (where <I> is the work function of the
`target material, and q<P is typically 3-5 eV). However, when a positive ion (such as Ar+) comes
`close to the target surface it appears as a potential well of 15.67 eV to target electrons occupying
`energy states near the Fermi level. Thus, some of the target electrons can escape by tunneling
`into these potential wells. The energy difference between 15.67 eV and the mini-mum free(cid:173)
`space escape energy is released in the form of a photon. If this photon (e.g., of energy 10.67 eV
`= [15.67 eV - 5 eV]) is absorbed by another electron near the target surface, this electron (i.e.,
`the Auger electron) may posses enough energy to be emitted from the target into free space.
`Since this sequence of events is rather improbable, several ions must strike the cathode in order
`for a single Auger electron to be emitted. The secondary electron yield per bombarding ion Yi
`has been measured to be 0.05-0.1 for metal targets. Thus, each secondary electron needs to
`generate as much as 20 ion-electron pairs before it reaches the anode to sustain the discharge.
`Besides providing a source of electrons to sustain the discharge, electron emission is also
`important to the sputtering process in other ways. First, ions bombarding the cathode are
`neutralized. Thus, it is highly probable that each ion that closely approaches the target will
`extract an electron from the target, and return to the discharge as a neutral atom. Second, the
`total target current, I, is the sum of the ion flux striking the target, Ii, and the secondary electron
`current leaving the electrode, le. Since Yi is larger for dielectric materials, their le is also larger.
`This implies that for the same cathode current, I, dielectrics will sputter more slowly than
`metals. That is, in dielectric sputtering a larger fraction of I is due to electron emission, and thus
`for equal values of I, a smaller ion flux is striking the dielectric target.
`
`Page 9 of 14
`
`

`

`448 SILICON PROCESSING FOR THE VLSI ERA
`
`TRANSITION OF EXCITED ELECTRON (AUGER PROCESS)
`
`Fig, 11·11 Potential energy diagram for ion approaching a metal target.
`11.3.5 Sputter Deposited Film Growth
`Upon being ejected from the target surface, sputtered atoms have velocities of 3-6x 10 5 cm/sec
`and energies of 10-40 eV. It is desirable that as many of these sputtered atoms as possible be
`deposited upon the substrates to form the specified thin film. To accomplish this, the target and
`wafers are closely spaced, with spacings of 5-10 cm being typical. The mean free path, A, of
`sputtered atoms at typical sputter pressures is less than 5-10 cm (e.g., at 5 mtorr, A :l cm).
`Thus, it is likely that sputtered atoms will suffer one or more collisions with the sputter-gas
`atoms before reaching the substrate (Fig. 11-12). The sputtered atoms may therefore: a) arrive at
`the substrate with reduced energy (-1-2 e V); b) be backscattered to the target or the chamber
`walls; or c) lose enough energy so that thereafter, they move by diffusion in the same manner as
`neutral sputter gas atoms. These events imply that the sputtering gas pressure can impact various
`film deposition parameters, such as the deposition rate and composition of the film.
`The formation and growth of the thin film on the substrate proceeds according to the general
`discussion on thin film formation given in Chap. 4. Therefore, this discussion is restricted to the
`events that uniquely impact the formation and growth of glow-discharge sputtered films.
`The substrate surface ( onto which the desired film of the sputtered target material is
`deposited) is also subjected to impingement by many species. The sputtered atoms arrive and
`condense onto the substrate. For a typical deposition rate of 20 nm/min, a monolayer of
`deposited film will form approximately every second (assuming the size of a typical atom is
`-0.3 nm). For this case, and even for much higher deposition rates of Al (e.g., 620 nm/min), the
`heat of condensation is not necessarily the most important source of substrate heating.
`11.3.6 Species that Strike the Wafer During Film Deposition
`In addition to the sputtered atoms, the substrate is also struck by many other species (Fig. 11-
`13 ), the most important being: l)fast neutral sputter gas atoms, which retain significant energy
`after striking and recoiling from the cathode. As they impinge upon the substrate, some may
`embed themselves in the growing film; 2) negative ions, formed near the cathode surface by
`
`Page 10 of 14
`
`

`

`ALUMINUM THIN FILMS AND PHYSICAL VAPOR DEPOSITION IN ULSI 461
`
`Fig, 11·23 An spent aluminum circular-planar-magnetron target (right) is being changed.
`reported that deposition rates of 1200-nm/min can be achieved when depositing Al:(0.5%)Cu
`films on 200-mm wafers.30 In fact, the maximum sputter deposition rates with such sources are
`usually not limited by any aspect of the plasma (as in de and rf diode cases), but rather by the
`ability to cool the cathode target to keep it from melting.
`The fact that the annular shape of the plasma in a circular planar magnetron results in a non(cid:173)
`uniform erosion of the target also causes the deposition onto nearby wafers to be non-uniform.
`In fact the deposition profile mirrors the race track when the wafer is close to the target, but this
`deposition profile becomes more smeared as the sample is moved further from the target (see
`Fig. 11-24a). Thus, a tradeoff is often made between deposition rate and thickness uniformity
`(i.e., uniformity improves as the target-to-wafer spacing is increased, but the deposition rate
`decreases). Note also from Fig. 11-24 that the target diameter is made larger than the wafer
`diameter to improve thickness uniformity. As an example, for 200-mm (8 in) wafers, the target
`diameter is typically 330 mm (14 in). Figure 11-24b shows such a target sputtering onto a
`smaller wafer. In practice, thickness uniformities of ±5% (3cr) within a wafer can be achieved.
`11.6 VLSI AND ULSI SPUTTER DEPOSITION EQUIPMENT
`A number of different sputtering systems have been designed for commercial use in IC
`fabrication. Only the dominant sputtering systems designs that were used in VLSI and ULSI
`generations will be described (i.e., those used for IC fabrication on 125-mm, 150-mm, and 200-
`mm wafars). Sputtering systems used for earlier IC technologies are described in the first edition
`of Vol. 1. In this section the generic components of a sputtering system are discussed first. Then
`the sputtering systems used in 125-mm and 150-mm wafer processing are briefly examined. The
`focus of this discussion is on the sputtering systems used for 200-mm wafers (and larger).
`11.6.1 The Components of a Generic Sputtering System
`The schematic of a basic sputtering system is shown in Fig. 11-25. It consists of the following
`subsystems: a) the sputter chamber, in which the substrate holder and sputtering source reside
`(the source also includes the target); b) vacuum pumps; c) power supplies (de and/or rf); d)
`sputtering gas supply and flow controllers; e) monitoring equipment (pressure gauges, volt-
`
`Page 11 of 14
`
`

`

`462 SILICON PROCESSING FOR THE VLSI ERA
`
`5cm THROW
`
`10000
`
`w
`~
`a:
`z
`0
`j:: 5000
`iii
`0
`0.
`w
`0
`
`-R
`
`(a)
`
`--(cid:173)R
`
`( b)
`
`Argon
`
`Thermocouple
`
`Fig, 11-24 a) The deposition profile and deposition rate on a sample in front of a magnetron cathode as a
`function of sample distance.21 Reprinted with permission of Academic Press. b) Sputter deposition from a
`planar magnetron target whose diameter is larger than that of the wafer.
`(pressure gauges, voltmeters, and residual-gas analyzers); t) wafer holders and handling
`mechanisms; and g) microcomputer controller.
`In modern sputter tools the sputter chambers are isolated from the ambient by a load lock.
`That is, a cassette of wafers to be sputter deposited is placed into the loadlock. Then the valve
`connecting the loadlock to the outside ambient is closed. The loadlock is pumped down, and
`upon reaching the base pressure of the transfer chamber the vacuum valve to the transfer
`chamber is opened. A robotic arm in the transfer chamber removes a wafer from the cassette
`and brings it into the transfer chamber. The loadlock:/transfer chamber valve is then closed. The
`sputter chamber awaiting the wafer has also been pumped to the appropriate low base pressure
`(e.g., 10-6-10-8 torr). The vacuum valve to the sputter chamber is opened, and the wafer is then
`DC/RF
`RF VOLTAGE
`
`GAS SUPPLY
`
`DC/RF/GROUND
`Fig, 11·25 Schematic drawing showing some of the components of a sputtering system.
`
`·:M~2fti.?
`PUMP
`
`Page 12 of 14
`
`

`

`ALUMINUM THIN FILMS AND PHYSICAL VAPOR DEPOSITION IN ULSI
`
`463
`
`fig. 11·26 Examples of sputtering targets for planar circular magnetron sources.
`inserted into the sputter chamber by the robot arm. After the arm withdraws, the vacuum valve
`between the sputter chamber and the transfer chamber is closed, and sputter gas is flowed into
`the chamber to the appropriate pressure (e.g., 5 mtorr) and flow rate. The sputter process can
`then be started.
`11.6.1.1 Sputtering Targets: Sputtering targets consist of the material that is to be sputter(cid:173)
`deposited onto the wafer. Targets in circular magnetrons are circular discs 3 to 10 mm thick, as
`shown in Fig. 11-26. These are bonded to a water-cooled Cu backing plate for good thermal
`contact, usually by soldering with a low-melting-temperature metal (or with a conductive
`epoxy). Since over 70% of the energy incident onto the target goes into target heating, targets
`must be adequately cooled to prevent the following problems: warpage; deposition-rate changes
`.due to thermal restructuring of the target material; delamination from the backing plate; or even
`target melting. De-ionized water is used as the cooling liquid to prevent electrolytic corrosion
`between the electrically biased backing plate and the grounded water supply. The entire
`cathode assembly is floated off ground by a ceramic insulator ring.
`Targets consisting of metals with relatively low melting points (e.g., Al, Cu, and Ti) are
`fabricated by melting and casting the metals in either vacuum or protective ambients. The
`starting materials to fabricate the targets must be very pure (e.g., up to 99.9999 pure [or six
`nines] for aluminum targets). In the case of cast alloy targets, fine grain size is preferred to
`minimize segregation of constituent elements into nodules during ion bombardment (since such
`nodules are a source of large particle ejection from the target during sputtering). Targets of
`high-melting-point refractory metals (e.g., W:Ti) and compound materials are generally formed
`from powders by hot-pressing. Although there may be no alternative, this fabrication method is
`less desirable than the vacuum-melting process. Less-than-theoretical densities are achieved
`with powder processing. As a result, such targets release trapped gases, particulates, and
`exhibit occlusions as their surfaces are eroded. The latter can give rise to non-uniform target
`erosion, arcing, and deposited films of inferior quality. Some device applications (e.g., memory
`circuits) also demand material very low in alpha-particle emitting elements (i.e., less than 0.01
`ppb U and Th content). 37
`Pre-sputtering of targets is done to clean their surfaces prior to film deposition. Once the
`target surface is clean, sputter deposition onto product wafers can take place. Arcing may occur
`at the target surface in de sputtering until the surface is cleaned. Thus, it is common practice to
`
`Page 13 of 14
`
`

`

`464 SILICON PROCESSING FOR THE VLSI ERA
`
`condition freshly installed targets by slowly increasing the applied power, until all the material
`that could give rise to the arcing has been sputtered away.
`11.6.1.2 Vacuum Pumps for Sputtering Systems: Vacuum pumps are also an important part of
`the sputtering system (as will be described in more detail in Sect. 11.6.3). Modern sputtering
`systems use cryopumps to keep the sputter chambers under high ( or ultra-high) vacuum between
`