Plasma Etching
`
`An Introduction
`
`Edited by
`
`Dennis M. Manes
`Plasma Plzysics Laborafonz
`Princeton Um'(3er.s‘1'fy
`Princeton, New Jersey
`
`Daniel L. Flamm
`A T& T Bell Laboratories
`
`Marray Hill, New Jersey
`
`\
`a‘
`
`/—‘
`
`ACADEMIC PRESS, INC.
`Harcourt Brace Jovanovich, Publishers
`Boston San Diego New York
`Berkeley London Sydney
`Tokyo Toronto
`
`LAM Exh 1015-pg 1
`
`

`
`Com
`
`Pref:
`
`7*-4?-H
`
`/w/«,4»--o<»~<>-4i—»~4
`
`I
`
`I»)
`
`
`
`
`
`y~»,«’,<_r-4/'/./'/>~‘1>-<*-**-*54>-.4
`
`Copyright © 1989 by Academic Press, Inc.
`All rights reserved.
`No part of this publication may be reproduced or
`transmitted in any form or by any means, electronic
`or mechanical, including photocopy, recording, or
`any information storage and retrieval system, without
`permission in writing from the publisher.
`
`ACADEMIC PRESS, INC.
`1250 Sixth Avenue, San Diego, CA 92101
`
`United Kingdom Edition published by
`ACADEMIC PRESS INC. (LONDON) LTD.
`24-28 Oval Road, London NW1 7DX
`
`Library of Congress Cataloging—in—Publication Data
`
`Plasma etching.
`(Plasma: materials interactions)
`Bibliography: p.
`Includes index.
`I. Manes, Dennis M.
`1. Plasma etching.
`Daniel L.
`III. Series: Plasma.
`TAZOZOPS
`1988
`621.044
`87-37419
`ISBN O-12-469370-9
`
`II. Flarnm,
`
`Alkaline paper
`
`
`
`
`
`
`
`Printed in the United States of America
`89909192
`987654321
`
`
`99
`I
`I
`_ -
`.
`I
`
`LAMeExh l0l5=pg} e— e— — — —
`
`LAM Exh 1015-pg 2
`
`

`
`
`
` neth Herb
`Plasma Etching Technol0gy--An Overview
`
`SUBTRACTWE
`
`ADDYFIVE
`
`SW C:
`
`
`
`SUBSTRATE _§, ‘
`
`AFTER
`LITHOGRAPHY
`
`DEPOSIT ——>
`
`REMOVAL
`
`AFTER
`MASK
`
`
`
`FIGURE 7. The two basic methods of pattern transfer are: subtractive, where the planar
`film is removed from areas not protected by the mask; and additive, where the film is
`deposited over the mask pattern and the mask with undesired film deposits is later removed.
`
`In the additive process, the pattern is formed at the same time the film is
`deposited. A resist pattern-mask is first formed on the base substrate and
`afterwards, a film material is deposited over the mask and onto uncovered
`substrate areas,
`thus creating the desired pattern. When deposition is
`complete, the resist is swelled by a strong solvent and dissolves away, lifting
`off unwanted deposits from the masked areas of the pattern.
`The subtractive process dominates production pattern transfer. The addi-
`tive method is mainly used for masking laboratory prototype devices.
`Plasma etching can be applied to steps in both processes.
`
`E.
`
`ISOTROPIC AND ANISOTROPIC ETCHING
`
`Differences in etching mechanisms have an immediate effect on the profiles
`of features in both “wet” chemical and “dry” plasma chemical etching.
`Pzzrely chemical etching usually has no preferential direction. This leads to
`isotropic circular profiles, which undercut a mask [14] (see Fig. 8). As
`
`tlves un-
`
`derlying
`zpattern
`it (acids
`such as
`at
`‘Li in a
`UQ
`
`i-
`
`.
`
`-LAMi3xhiriiapeg
`
`LAM Exh 1015-pg 3
`
`

`
`12
`
`Daniel L. Flamm and G. Kenneth Herb
`
`\\\\\\\\\V s\\\\V MASK
`
` SUBSTRATE
`
`/
`
`(‘H
`
`g
`
`\\\\\\\\\i
`
`‘
`
`.\\\\\
`
`ANISOTROPIC
`
`H II
`
`‘
`
`: dv
`
`ISOTROPIC
`l
`A\\\\\\\\‘ &\\\V
`.-
`I
`
`OVERETCH
`
`FIGURE 8. Replicating the pattern from a mask stencil into a film depends on the ability to
`control horizontal etching. Left: Isotropic etching; the horizontal and vertical etch rates are
`equal, creating an undercut, d“. in the film beneath the niask, equal to the film thickness, [IV .
`If etching is continued, this lateral attack will continue and the etched film edge profile will
`become almost vertical. Right: Anisotropic etching;
`the horizontal etch component is very
`small, resulting in faithful pattern transfer.
`
`shown, the thickness etched (dv) equals the undercut (JH) until etching
`reaches the film-substrate boundary. Overetching past the point where the
`underlying substrate is just exposed increases the undercut and radius of
`the undercut cross section. At large degrees of overetching the walls are
`essentially vertical. In practice, overetching is necessary because of wafer
`surface topography, nonuniform film thicknesses and Variations in etch rate
`in the reactor (see Section IV.C.2). Moreover, in isotropic plasma etching it
`is difficult to control the degree of overetching beyond the endpoint because
`the rate of undercut may accelerate at the end point where {unfortunately)
`control is needed most. This loading effect is discussed further in Chapter 2.
`Ideal
`ion enhanced plasma etching, on the other hand, produces an
`anisotropic profile (see Fig. 8). When the mean free path of ions in a plasma
`is long compared to feature depth, electrical fields (the sheath field) make
`ions strike horizontal surfaces almost exclusively at normal incidence. This
`ion bombardment preferentially accelerates the chemical reaction in the
`
`
`
`Plasma Etching T4
`
`vertical directior
`masked features
`
`can also be prod
`and superimpost
`It is importai
`features does nc
`profiles are forrr
`(as noted befor
`
`etching along 3}
`common in wet
`instance when
`
`halogen radicals
`or tapered walls
`cutting is usual
`practice the edg
`erosion (e.g., bf
`profile. If the II
`ting, it is possil
`with the walls C
`
`F. THE TRAl
`PLASMA l
`
`Plasma etching
`stripping [16] ft
`19705. By the e
`patterning silict
`underlying met
`and indirect.
`/‘
`
`ashing in oxygt
`lines. Productic
`and a host of
`processes were
`seemed to lie is
`
`gases such as
`products and s
`recognized tha:
`that greatly 6X4
`Because of t’
`
`a 1 um mask <
`
`
`
`LAM Exh 1015-pg 4
`
`

`
`
`
`Plasma Etching Teehnolegy—An Overview
`
`13
`
`vertical direction so that vertical sidewalls are formed along the edges of
`masked features at right angles to the substrate. In practice, tapered profiles
`can also he produced (see Sections IV.C.l, IV.C.2) because of ion scattering
`and superimposed isotropic chemical etching elfects.
`It is important to add that the formation of vertical walls on etched
`features does not necessarily mean that etching was ion enhanced. Vertical
`profiles are formed in isotropic chemical etching during severe undercutting
`(as noted before), and can also be produced when there is preferential
`etching along specific crystal planes. Such crystallographic attack is more
`common in wet chemical etching {I5}, but it also occurs in plasmas, for
`instance when IIl—V compound semiconductors are etched by gaseous
`halogen radicals. Dry crystallographic etching can be used to form vertical
`or tapered walls with only a small undercut (see Chapter 2). While under-
`cutting is usually obvious when the mask is compared to the feature, in
`practice the edges of a mask may be tapered rather than vertical, and mask
`erosion (e.g., by sputtering, or chemical attack) leads to a more complex
`profile. If the mask erodes at an appropriate rate relative to the undercut-
`ting, it is possible to attain features in which the edge of the mask aligns
`with the walls of etched features, even though the etching is isotropic.
`
`F. THE TRANSITION FROM WET TO
`PLASMA ETCHING
`
`Plasma etching was explored as a cheaper alternative to wet solvent resist
`stripping [16} for integrated circuit manufacture in the late 1960s and early
`1970s. By the early 1970s, CF4/O2 plasma etching was widely adopted for
`patterning silicon nitride passivation of its selectivity over resist masks and
`underlying metalization. The wet chemical alternatives were complicated
`and indirect. Around the same time, oxygen plasma resist stripping, or
`ashing in oxygen plasmas, was finally integrated into many manufacturing
`lines. Production plasma etch processes were next developed for polysilicon
`and a host of other materials. These initial dry plas1na—assisted etching
`processes were purely chemical and isotropic. At first,
`their advantages
`seemed to lie in unique processing sequences, substitution of safe nontoxic
`gases such as 02 and CF4 for corrosive liquids, easily discharged waste
`products and simple automation. By the late 1970s, however, it was widely
`recognized that plasma etching offered the possibility of a vertical etch rate
`that greatly exceeded the horizontal rate (e.g., anisotropic etching).
`Because of the isotropic undercut, wet etching a 1 pm thick film through
`a 1 am mask opening at best yields a cross—sectional profile that measures
`
`ability‘ to
`rates are
`
`3655. (IV.
`rofile will
`
`it
`
`very
`
`etching
`iere the
`tdius of
`alls are
`at” wafer
`tch rate
`
`ching it
`because
`
`inately)
`apter 2.
`aces an
`
`plasma
`i) make
`Ce. This
`L
`the
`
`LAM Exh Loi5§p.g
`
`LAM Exh 1015-pg 5