Woodhead Publishing Series in Electronic and Optical Materials:
`Number 42
`
` Nanolithography
` The art of fabricating
`nanoelectronic and nanophotonic
`devices and systems
`
` Edited by
` Martin Feldman
`
`Oxford Cambridge Philadelphia New Delhi
`
`© Woodhead Publishing Limited, 2014
`
`NXP Ex. 2004
`Impinj, Inc. v. NXP B.V. - IPR2020-01630
`
`

`

` Published by Woodhead Publishing Limited,
` 80 High Street, Sawston, Cambridge CB22 3HJ, UK
` www.woodheadpublishing.com
` www.woodheadpublishingonline.com
`
` Woodhead Publishing, 1518 Walnut Street, Suite 1100, Philadelphia,
`PA 19102-3406, USA
`
` Woodhead Publishing India Private Limited, 303 Vardaan House, 7/28 Ansari Road,
`Daryaganj, New Delhi – 110002, India
` www.woodheadpublishingindia.com
`
` First published 2014, Woodhead Publishing Limited
` © Woodhead Publishing Limited, 2014, except Chapter 6 which is © H. Smith, 2014. The
`publisher has made every effort to ensure that permission for copyright material has
`been obtained by authors wishing to use such material. The authors and the publisher
`will be glad to hear from any copyright holder it has not been possible to contact.
` The authors have asserted their moral rights.
`
` This book contains information obtained from authentic and highly regarded
`sources. Reprinted material is quoted with permission, and sources are indicated.
`Reasonable efforts have been made to publish reliable data and information, but
`the authors and the publisher cannot assume responsibility for the validity of all
`materials. Neither the authors nor the publisher, nor anyone else associated with
`this publication, shall be liable for any loss, damage or liability directly or indirectly
`caused or alleged to be caused by this book.
` Neither this book nor any part may be reproduced or transmitted in any form or
`by any means, electronic or mechanical, including photocopying, microfi lming and
`recording, or by any information storage or retrieval system, without permission in
`writing from Woodhead Publishing Limited.
` The consent of Woodhead Publishing Limited does not extend to copying for general
`distribution, for promotion, for creating new works, or for resale. Specifi c permission
`must be obtained in writing from Woodhead Publishing Limited for such copying.
`
` Trademark notice: Product or corporate names may be trademarks or registered
`trademarks, and are used only for identifi cation and explanation, without intent to
`infringe.
`
` British Library Cataloguing in Publication Data
` A catalogue record for this book is available from the British Library.
`
` Library of Congress Control Number: 20131948203
`
` ISBN 978-0-85709-500-8 (print)
` ISBN 978-0-85709-875-7 (online)
` ISSN 2050-1501Woodhead Publishing Series in Electronic and Optical Materials (print)
` ISSN 2050-151X Woodhead Publishing Series in Electronic and Optical Materials (online)
`
` The publisher’s policy is to use permanent paper from mills that operate a
`sustainable forestry policy, and which has been manufactured from pulp
`which is processed using acid-free and elemental chlorine-free practices.
`Furthermore, the publisher ensures that the text paper and cover board used
`have met acceptable environmental accreditation standards.
`
` Typeset by Newgen Knowledge Works Pvt Ltd, India
` Printed by Lightning Source
`
`© Woodhead Publishing Limited, 2014
`
`NXP Ex. 2004
`Impinj, Inc. v. NXP B.V. - IPR2020-01630
`
`

`

` Related titles:
` Carbon nanotubes and graphene for photonic applications
`(ISBN 978-0-85709-417-9)
`
` Laser growth and processing of photonic devices
`(ISBN 978-1-84569-936-9)
`
` Handbook of solid-state lasers
`(ISBN 978-0-85709-272-4)
`
` Details of these books and a complete list of titles from Woodhead Publishing can
`be obtained by:
`
`•
`•
`
`•
`
` visiting our web site at www.woodheadpublishing.com
` contacting Customer Services (e-mail: sales@woodheadpublishing.com; fax:
`+44 (0) 1223 832819; tel.: +44 (0) 1223 499140 ext. 130; address: Woodhead
`Publishing Limited, 80, High Street, Sawston, Cambridge CB22 3HJ, UK)
` in North America, contacting our US offi ce (e-mail: usmarketing@
`woodheadpublishing.com; tel.: (215) 928 9112; address: Woodhead Publishing,
`1518 Walnut Street, Suite 1100, Philadelphia, PA 19102-3406, USA)
`
` If you would like e-versions of our content, please visit our online platform: www.
`woodheadpublishingonline.com. Please recommend it to your librarian so that
`everyone in your institution can benefi t from the wealth of content on the site.
` We are always happy to receive suggestions for new books from potential editors.
`To enquire about contributing to our Electronic and Optical Materials series, please
`send your name, contact address and details of the topic/s you are interested in to
`laura.pugh@woodheadpublishing.com. We look forward to hearing from you.
`
` The team responsible for publishing this book:
` Commissioning Editor: Laura Pugh
` Publications Coordinator: Lucy Beg
` Project Editor: Elizabeth Moss
` Editorial and Production Manager: Mary Campbell
` Production Editor: Richard Fairclough
` Project Manager: Newgen Knowledge Works Pvt Ltd
` Copyeditor: Newgen Knowledge Works Pvt Ltd
` Proofreader: Newgen Knowledge Works Pvt Ltd
` Cover Designer: Terry Callanan
`
`© Woodhead Publishing Limited, 2014
`
`NXP Ex. 2004
`Impinj, Inc. v. NXP B.V. - IPR2020-01630
`
`

`

`   
`
`

ÿÿÿÿÿÿ ÿ !"
`ÿ#$"%"
`&''(ÿ*ÿ+,-.
`/010ÿ23
`4 ÿ5106 
`< =;1ÿ2 >ÿ: 1
`?2 @ÿ1ÿ2 >
`A1 :ÿÿ;ÿ2 >
`A1 :ÿÿ;ÿ2 >
`1ÿ 9ÿ: 1;
`CDEFGÿIJGGKDLÿMENODPQRGLÿMEFGDPSTGEDÿUEFGJUGÿVKGJPWXLÿYEEVRKJVÿZTSWPXRPFQÿ[KDPKXÿPF
`5@ÿ1Bÿ ::1;;
`\WKUGDEFPUÿJFVÿ]NGPUJWÿ^JGKDPJWXLÿZDK_JUK
``ÿaÿ]NGPUJWÿNDEbKUGPEFÿWPGREQDJNRO
`< >ÿ:
`1 ?2; :ÿ 
`<cdcÿA6
`e 1;ÿ-f.-e@ : ;1g1=ÿ 2; :
`hÿaÿ\iGDKIKÿTWGDJjPEWKGÿk\lmnÿWPGREQDJNRO
`< >ÿ:
`1 ?2; :ÿ 
`<cocÿp:1
`
`ÿ8 2
`
`NXP Ex. 2004
`Impinj, Inc. v. NXP B.V. - IPR2020-01630
`
`

`

`  ÿ
`

   ÿ 


`ÿÿ !"#$%&ÿ'!()ÿ *#+%,$(-+.
`ÿ  



ÿ
`/010ÿ2

3 
`  ÿ45667
`

   ÿ 


`8ÿÿ9%":;!<ÿ*%&ÿ'!();ÿ=%$ÿ&(&%)("+*&*&,ÿ(&<ÿ*)(,*&,
`ÿ  



ÿ
`>0ÿ? 
`  ÿ66@67
`

   ÿ 


`AÿÿB(;C;ÿ=%$ÿ)*"$%ÿ(&<ÿ&(&% *#+%,$(-+.
`ÿ  



ÿ
`D0ÿ2  

ÿ Eÿ>0ÿF
`  ÿ6746 4
`

   ÿ 


`GÿÿB(;C !;;ÿ-+%#% *#+%,$(-+.
`ÿ  



ÿ
`>0D0ÿH IÿJ0ÿK Iÿ000ÿF0L0ÿMN
`  ÿ6
`6
`O
`

   ÿ 


`PÿÿQ+!)*;#$.ÿ(&<ÿ-$%"!;;*&,ÿ%=ÿ$!;*;#;ÿ=%$ÿ&(&% *#+%,$(-+.
`ÿ  



ÿ
`0ÿR3 N

ÿ EÿM0ÿS
`  ÿ6
`4@
`

   ÿ 


`TÿÿU*$!"#!<ÿ(;;!)' .ÿ&(&% *#+%,$(-+.
`ÿ  



ÿ
`M0ÿ> Iÿ>0ÿ/    ÿ EÿF0ÿVE
`  ÿ4 O6
`

   ÿ 


`
`NXP Ex. 2004
`Impinj, Inc. v. NXP B.V. - IPR2020-01630
`
`

`

`ÿÿÿ
` ÿ
`


`ÿ 
`!"ÿ# $
`%& ÿ'()*'+,
`%-
` .$ /ÿ


`
`01ÿÿ2343526ÿ784ÿ9ÿ42
` ÿ
`


`ÿ 
`:"ÿ; &
`%& ÿ'+<*',)
`%-
` .$ /ÿ


`
`00ÿÿ426ÿ95=452ÿ7ÿ3ÿÿÿ5ÿ245
` ÿ
`


`ÿ 
`%"ÿ! >$?ÿ"ÿ@AAA -
`ÿAÿB"ÿ.
` $
`%& ÿ',C*'D<
`%-
` .$ /ÿ


`
`0Eÿÿ95=4526ÿ784Fÿ2542ÿ7ÿGÿ952ÿ95=452ÿ9ÿ42
` ÿ
`


`ÿ 
`H"ÿ!$  *#
`%& ÿ'DD*+I'
`%-
` .$ /ÿ


`
`0JÿÿK4L39426ÿ5452ÿ9ÿ42
` ÿ
`


`ÿ 
`M"ÿN$&ÿAÿO"P"ÿQ
`$R$
`%& ÿ+I+*++'
`%-
` .$ /ÿ


`
`0SÿÿK95ÿ7ÿ ÿ452252
` ÿ
`


`ÿ 
`
`"ÿT ÿAÿM"P"ÿU R$
`%& ÿ+++*+C)
`%-
` .$ /ÿ


`
`0VÿÿK2W23825ÿ5ÿ=ÿ57554ÿXÿ752
` ÿ
`


`ÿ 
`
`NXP Ex. 2004
`Impinj, Inc. v. NXP B.V. - IPR2020-01630
`
`

`

`ÿ
` ÿ !
`  "

ÿ 
`%&ÿ(ÿ)*+,-.//ÿ012345,66ÿ*5ÿ5.51/*74130.849:ÿ10*3*56;ÿ<,701/139ÿ.5+ÿ+,=*>,ÿ,?,>76
`#ÿ   ÿ $
`"ÿ@  ABÿÿC$A ÿ AÿCÿ
` ÿ DDE
`  "

ÿ 
`%Fÿ(ÿG,-ÿ.88/*>.7*156ÿ.5+ÿ,<,03*53ÿ7,>451/13*,6ÿ*5ÿ5.51/*74130.849
`#ÿ   ÿ $
`ÿH $Bÿ@ÿIBÿÿÿ  
` ÿDJ
`  "

ÿ 
`L5+,M ÿNED
`#ÿ  $$ÿ Kÿ 
`ÿ ÿ
`

$Aÿ
`  
`O   Aÿ  Bÿ AÿA P ÿQ  Aÿ ÿ ÿ  R ÿA P $ AÿQÿ   Aÿ  BÿP
` A$ÿ ÿ$$ Bÿ ÿ $ KBÿ Aÿ ÿ

Q$ÿS ÿ ÿQÿ  ÿÿ  ÿTÿ
`  ÿ ÿ 

ÿUÿÿQ

ÿAV ÿÿ

A ÿSÿÿ ÿÿ 

ÿ ÿR  BÿWX
`
`

ÿ
` ÿ  
`Y$ ÿ ÿ  ÿ ÿQÿ ÿÿ ÿ$  P ÿ  $ÿ  $ ÿ ÿA ÿÿ  ÿ


` ÿQ ÿ A  ÿ ÿV ÿQÿ   A ÿ ÿÿ   $ ÿA  
`@P ÿ$ ÿ  R Bÿ $A ÿ $ÿU  Bÿ K  ÿ$P$ ÿZ["\Bÿ   B
`
`

ÿ
`
`NXP Ex. 2004
`Impinj, Inc. v. NXP B.V. - IPR2020-01630
`
`

`

` 3
` Electron beam lithography
`
` T. R. GROVES , University at Albany (SUNY), USA
`
` DOI : 10.1533/9780857098757.80
`
` Abstract : A focused electron beam (e-beam) represents the smallest,
`fi nest practical writing pencil known, with the capability of producing
`pattern features down to a few nanometers in size. Electron beam
`lithography does not rely on a pre-existing patterned mask, but can
`write the pattern directly from stored data. Because of its inherent high
`resolution and pattern fl exibility, e-beam lithography remains the method
`of choice for fabricating nanometer-scale structures in low volume.
`The historical Achilles heel of e-beam lithography has been its low
`throughput. This can be mitigated by exposing many pixels in parallel. A
`survey of present-day e-beam lithography is presented.
`
` Key words : electron beam, lithography, maskless, throughput, multiple
`beams.
`
` 3.1
`
` Introduction
`
` Electron beam lithography, also known as e-beam lithography, is the
`process of tracing out a pattern in a suitable recording medium using a
`focused e-beam. The underlying physical mechanism relies on the fact that
`the recording medium, typically a thin organic polymer fi lm, is altered by
`the passage of fast electrons. The recording medium is generally called
` resist . In a subsequent development step, the exposed material is removed
`(positive-tone process). Alternatively, the unexposed material is removed
`(negative-tone process). In either case, the result is a patterned fi lm, which
`acts as a binary mask for further processing. This processing might include
`reactive ion etching, selective ion implantation, electroplating, or physical
`vapor deposition, to name a few. The patterned binary mask is a versatile
`and inexpensive enabler for a variety of subsequent processes. By super-
`imposing multiple pattern layers, an enormous variety of useful devices
`can be fabricated. A typical positive-tone process is shown schematically
`in Fig. 3.1.
` A focused e-beam represents the smallest, fi nest practical writing pencil
`known (Pease and Chou, 2008; Pease, 2010). The ultimate electron optical
`resolution is the same as an electron microscope, in the range of 0.06–0.15 nm,
`
`80
`
`© Woodhead Publishing Limited, 2014
`
`NXP Ex. 2004
`Impinj, Inc. v. NXP B.V. - IPR2020-01630
`
`

`

`Electron beam lithography
`
`
`
` 81
`
`(1)
`
`(2)
`
`(3)
`
`(4)
`
` 3.1 Typical positive-tone resist process. From left to right: electron
`beam scans (1) and leaves behind a latent image in the resist layer (2).
`Exposed resist (3) dissolves in the development process. Subsequent
`reactive ion etch (RIE) selectively removes substrate material (4), using
`the developed resist layer as a binary mask.
`
`depending on the energy of the incident electrons. Ultimate lithographic
`resolution is not limited by the electron optics, but by the range of interac-
`tion of the beam electrons with the resist layer by scattering and second-
`ary processes. Ultimate lithographic resolution is typically in the range of a
`few nanometers, depending on the energy of the electrons and the specifi c
`nature of the resist. This is about an order of magnitude smaller than the
`lithographic resolution obtainable with conventional optical lithography.
` The pattern data are typically created using commercially available soft-
`ware for computer-aided design. These data must then be converted to a
`format usable by the e-beam writer. A digital electronic data path auto-
`matically converts and sends the data to the e-beam writer. The e-beam is
`then scanned over the writing surface using electric or magnetic fi elds, and
`turned on and off while it scans. Practically any arbitrary binary pattern can
`be written in this way.
` In conventional optical lithography, one forms a demagnifi ed image of
`a pre-existing patterned mask onto a resist-coated wafer. The mask can be
`used repeatedly to make many copies of the same pattern. In high-volume
`manufacturing of semiconductor chips, the patterns are highly complex.
`For example, 30 nm minimum-sized features might be distributed over a
`30 mm square area. This represents an upper limit of 10 12 pattern features.
`Consequently, the mask can be expensive and time-consuming to fabricate.
`It is only cost-effective if many wafers are exposed with a single mask, since
`the cost of the mask is amortized over all of the wafers exposed.
` E-beam lithography does not require a pre-existing mask, since the pat-
`tern is created and transmitted electronically. This permits great fl exibility
`in trying out a large number of different patterns in a short time. This is
`ideal for low-volume applications, in which few copies of a given pattern are
`needed. An e-beam writer is a pattern generator , whereas a conventional
`
`© Woodhead Publishing Limited, 2014
`
`NXP Ex. 2004
`Impinj, Inc. v. NXP B.V. - IPR2020-01630
`
`

`

`82
`
`
`
` Nanolithography
`
`optical lithography tool is a pattern replicator . Incidentally, the method of
`choice for patterning masks for optical lithography is e-beam lithography.
` An e-beam lithography system is comprised of several subsystems,
`including
`
`•
`•
`•
`•
`
`•
`
`•
`
`•
`
` an electron optical column, to produce the focused e-beam;
` analog electronics to produce, focus, blank/unblank, and scan the beam;
` digital electronics to store and transmit the pattern data;
` a high-precision mechanical XY stage to position the writing substrate
`relative to the e-beam;
` a high-vacuum system, with provision to move the writing substrate in
`and out of the vacuum;
` high-speed computers and microprocessors, to automatically perform all
`of the necessary tasks;
` an extensive software system.
`
` A considerable engineering effort is needed to make all of these compo-
`nents work reliably together.
` E-beam lithography originally grew out of scanning electron micros-
`copy. It was fi rst proposed by Buck and Shoulders, (1958), and fi rst dem-
`onstrated by M ö llenstedt and Speidel (1960). An electron microscope
` captures a high-resolution image of a pre-existing object. An e-beam
`writer creates a structure with moderate spatial resolution (relative to
`an electron microscope), but high accuracy and an enormous number of
`pixels. A high-resolution electron micrograph typically has 10 6 pixels. A
`high-fi delity e-beam-written pattern can have 10 12 pixels, with each pixel
`precisely positioned to within a reasonably small fraction of the pixel size.
`Consequently, an e-beam writer is much more complicated and expensive
`than an electron microscope.
` The overriding goal in any lithographic patterning is to produce a pat-
`tern in resist that approximates the original design pattern with the greatest
`possible fi delity. Lithographic patterns as exposed in resist are binary. As
`such, they have no gray scales. (The exposure process can utilize gray scales,
`but the resist image is typically binary). In this context, pattern fi delity con-
`sists of two basic attributes. One is the quality of individual pattern fea-
`tures, as embodied in dimensional control of the feature size. This includes
`the smoothness of feature edges, and the sharpness of corners. The other is
`accurate placement of pattern features.
` The written and processed pattern generally does not perfectly match the
`ideal, desired pattern. Errors in pattern feature size and placement can arise
`from multiple sources. For example, unwanted fl uctuation of the exposure dose
`leads to non-uniformity in the printed feature size. Electromagnetic noise and
`mechanical jitter in the system lead to random errors in pattern placement.
`
`© Woodhead Publishing Limited, 2014
`
`NXP Ex. 2004
`Impinj, Inc. v. NXP B.V. - IPR2020-01630
`
`

`

`Electron beam lithography
`
`
`
` 83
`
`Charging of contamination layers in the electron column, together with
`uncontrolled thermal expansion due to local temperature fl uctuations, causes
`the beam to wander. Such errors are often not predictable or repeatable, and
`must be minimized by an iterative process of measurement, deduction (to
`determine the source of the error), and reduction to a tolerable level.
` Some errors are predictable and repeatable, such as average defl ection
`distortion, average position error of the mechanical stage, and average scat-
`tered dose variation arising from local variation in pattern density (so-called
`proximity effect). This class of errors is amenable to correction by auto-
`mated measurement, computation, and feedback. Pattern fi delity is deemed
`acceptable if the errors in printed feature size and placement are a reason-
`ably small fraction of the minimum printed feature size, also known as the
` critical dimension (CD).
` In order to measure these errors, sophisticated metrology is needed as
`an indispensable adjunct to the writing process. Much of this metrology
`capability is built into the e-beam writer, and automated within the writ-
`ing process. For example, the position of the e-beam relative to the writ-
`ing surface can be measured by using the beam to scan alignment marks
`placed on the writing surface in a prior processing step. The position of
`the beam can thus be corrected to compensate for distortion and drift. In
`addition, a laser interferometer can be used to measure the position of the
`XY stage relative to a stable mechanical datum built into the system. The
`laser interferometer forms the built-in reference standard for all measure-
`ments pertaining to pattern placement. A typical laser interferometer for
`e-beam lithography has a resolution of λ /1000, where λ is the wavelength
`(632.8 nm) of a He-Ne laser. Higher resolution is also available. In princi-
`ple, this resolution represents the smallest possible increment of pattern
`edge placement.
` The historical Achilles heel of e-beam lithography is its slow speed. This
`arises from two limitations. First, the writing process is essentially serial,
`with the pattern traced out sequentially using a probe beam. This is in con-
`trast to conventional optical lithography, in which an entire complex pat-
`tern is exposed in one fl ash or scan. Second, the useful writing current is
`limited. The beam electrons are randomly scattered by one another in the
`drift length of the electron column. This degrades the resolution as the cur-
`rent is increased. In addition, every electron optical system is limited in the
`amount of current it can supply at the writing surface for a given resolution.
`Useful writing current at any given resolution is always limited, either by
`Coulomb scattering in the beam path, or by limited ability of the electron
`optical system (especially the electron source) to supply the desired writing
`current.
` A useful estimate of writing speed is the pattern area swept out per unit time
`by the beam. Typical e-beam writers operate in the range 0.0001–1.0 cm 2 /s,
`
`© Woodhead Publishing Limited, 2014
`
`NXP Ex. 2004
`Impinj, Inc. v. NXP B.V. - IPR2020-01630
`
`

`

`84
`
`
`
` Nanolithography
`
`depending on the desired resolution. By comparison, conventional opti-
`cal exposure tools operate in the range 20–30 cm 2 /s. Conventional optical
`systems are thus several orders of magnitude faster than currently existing
`e-beam systems. For this reason, optical lithography remains the method
`of choice for high-volume manufacturing of integrated circuit chips, while
`e-beam lithography remains the method of choice for device fabrication at
`low volume, with superior lithographic resolution.
` In summary, e-beam lithography has the dual advantages of high spatial
`resolution and fl exibility of pattern generation. It has the drawback of low
`speed. The purpose of this chapter is to explore these factors in some detail,
`and to offer some analysis of the possible avenues of future improvement.
`
` 3.2
`
` Using pixel parallelism to address the
`throughput bottleneck
`
` As mentioned, e-beam lithography systems use electric and magnetic elec-
`tron lenses to form a sharply focused e-beam, which is scanned over the
`writing surface. The simplest possible confi guration is one in which an image
`of a point-like electron source is formed directly on the writing surface
`(Chang et al ., 1976; Herriott et al., 1975; Kelly et al., 1981; Alles et al., 1987).
`
`8
`
`2
`
`4
`
`3
`
`3
`
`1
`
`α
`
`5
`
`5
`
`6
`
`7
`
` 3.2 Typical Gaussian beam system confi guration.
`
`© Woodhead Publishing Limited, 2014
`
`NXP Ex. 2004
`Impinj, Inc. v. NXP B.V. - IPR2020-01630
`
`

`

`Electron beam lithography
`
`
`
` 85
`
`An example of this is shown schematically in Fig. 3.2. The compact electron
`source (1) is typically a thermally assisted fi eld-emitter, also known as a
`Schottky emitter. The virtual source, seen ‘looking back’ toward the source
`from the electron column, typically has a diameter in the range 15–20 nm.
`A magnetic lens (2) forms an intermediate image of the source (1) in a
`set of electrostatic defl ection plates (3). The defl ection plates (3) move the
`beam onto an edge (4) for blanking. A magnetic lens (6) forms a demag-
`nifi ed image of the source onto the writing surface (7). Electromagnetic
`or electrostatic defl ectors (5) move the writing spot laterally on the writ-
`ing surface (7). The writing surface (7) is mounted on a movable stage for
`increased range of motion. An aperture (8) limits the illumination cone
`semi-angle α of the beam measured at the writing surface. An optimum
`aperture size exists for a given resolution that balances the effects of spher-
`ical aberration, diffraction, and Coulomb scattering within the beam. These
`effects will be described in more detail later.
` The writing spot represents one pixel, and has a lateral intensity distri-
`bution that is roughly Gaussian. It is scanned over the pattern area using
`variable electric or magnetic fi elds, and turned off and on as it is scanned,
`thus generating the pattern. The beam can be scanned in a raster pattern,
`similar to a scanning electron microscope. Alternatively, the beam can be
`defl ected only to those places where pattern features are to be written. This
`latter approach goes by the term vector scanning. Writing one pixel at a time
`is the ultimate serial writing process. Its inherent simplicity comes with the
`penalty of limited speed.
` An alternative writing strategy is to project a rectangular writing spot of
`variable size and aspect ratio in a single fl ash (Pfeiffer and Loeffl er, 1970;
`Pfeiffer, 1978; Goto et al ., 1978; Trotel, 1978). It is shown schematically in
`Fig. 3.3. The beam from an electron source fl oods a square aperture. The
`source is typically an extended (as opposed to point-like) single-crystal lan-
`thanum hexaboride LaB 6 thermionic emitter. The extended source insures
`that the fi rst square aperture is illuminated uniformly. A condenser lens
`forms an electron optical image of the square aperture in the plane of a sec-
`ond square aperture. A spot-shaping defl ector moves the image of the fi rst
`square aperture on the second square aperture, thus forming a rectangular
`compound spot of variable size and aspect ratio. The spot-shaping defl ector
`is positioned at an intermediate image of the source (1). This insures that
`the illumination remains uniform as the beam is defl ected. The resulting
`rectangular spot is then demagnifi ed, and imaged onto the writing surface.
`In this way, many pixels can be exposed in a single fl ash, thus increasing
`the pixel parallelism. This is called a variable-shaped beam approach. Pixel
`parallelism varies with the size of the shape, but typically 64–256 pixels are
`written in a single ‘fl ash.’ Obviously, this is a signifi cant increase over the
`single-pixel Gaussian beam.
`
`© Woodhead Publishing Limited, 2014
`
`NXP Ex. 2004
`Impinj, Inc. v. NXP B.V. - IPR2020-01630
`
`

`

`86
`
`
`
` Nanolithography
`
`Electron source
`
`First
`square aperture
`
`Condenser lens
`
`First image of
`electron source
`
`+
`
`–
`
`Spot shaping
`deflector
`
`Second
`square aperture
`
`Shaped beam
`
` 3.3 Typical variable-shaped beam system confi guration.
`
`(a)
`
`(b)
`
`48
`
`2
`
` 3.4 Exposure fl ashes for a Gaussian beam system (a), and a variable-
`shaped beam system (b). The intensity distribution is plotted as a
`function of lateral position below. Incremental variation in pattern edge
`placement is also indicated.
`
` It is possible to place the edge of a pattern feature with a precision that is
`a small fraction of the pixel resolution. This is shown schematically in Fig. 3.4.
`The pattern of writing for individual fl ashes is shown for a Gaussian beam on
`
`© Woodhead Publishing Limited, 2014
`
`NXP Ex. 2004
`Impinj, Inc. v. NXP B.V. - IPR2020-01630
`
`

`

`Electron beam lithography
`
`
`
` 87
`
`Fig. 3.4a, and for a variable-shaped beam on Fig. 3.4b. The number of individ-
`ual fl ashes is indicated in both cases. We assume that the pixel resolution is
`the same for both cases. We now desire to move the leftmost pattern edge by
`an increment that is a small fraction of the pixel resolution. With a Gaussian
`beam, this can be done by exposing the leftmost column of pixels with reduced
`intensity. This will cause the edge of the resist image to move incrementally
`to the right, depending on the amount of exposure. This is referred to as gray-
`scale writing. This can also be used in conjunction with defl ecting the beam by a
`small increment (Abboud 1997). With a variable-shaped beam, the placement
`of pattern edges is determined by the shaping increment, which is decoupled
`from the pixel resolution. This ability to place pattern edges with an increment
`that is a small fraction of the pixel size adds complexity to the writing, but it
`enhances the ability to write an enormous variety of useful patterns.
` One can add further pixel parallelism by projecting a character or cell
`in a single fl ash. This is similar to the variable-shaped beam approach, with
`character apertures replacing the square apertures. The character is auto-
`matically selected from a library of shapes that reside in the electron column
`in the form of small stencil masks. Each stencil can be electron-optically
`imaged at will onto the writing surface. This approach, called cell or charac-
`ter projection (Pfeiffer 1979), is especially useful in applications for which a
`small pattern is repeated many times. In practice, this roughly doubles the
`pixel parallelism of the variable-shaped beam approach.
` One can add still further pixel parallelism by using multiple e-beams in
`a single electron column. A single electron source fl oods an array of aper-
`tures, forming an array of individual beamlets. Below each aperture is a pair
`of electrostatic defl ection plates, which is used to steer the beamlet onto a
`downstream aperture, thus blanking only that individual beamlet. It is nec-
`essary to address each pair of defl ection plates individually.
` In Fig. 3.5 (Platzgummer et al. , 2008) a single source of ions (or electrons)
`is collimated by the condenser optics. The beam fl oods an aperture array.
`Individual beamlets are either transmitted or blanked, consistent with the
`pattern pixels. The entire aperture array is imaged and demagnifi ed by a
`system of lenses onto the writing surface below. The spacing between beam-
`lets is demagnifi ed by the same factor as the beamlet spot size. All of the
`beamlets pass through two intermediate crossovers in the column. An array
`of Gaussian beamlets is formed at the writing surface, each of which can be
`turned on and off at will. This approach is referred to as projection maskless
`lithography patterning (PMLP).
` In Fig. 3.6 (Kruit, 1998; Wieland et al. , 2001; van den Berg et al. , 2011) each
`beamlet is individually imaged and demagnifi ed onto the writing surface
`below. The spacing between beamlets does not change, and the individual
`beamlets do not pass through common crossovers. An array of Gaussian
`beamlets is formed at the writing surface, each of which can be turned on
`
`© Woodhead Publishing Limited, 2014
`
`NXP Ex. 2004
`Impinj, Inc. v. NXP B.V. - IPR2020-01630
`
`

`

`88
`
`
`
` Nanolithography
`
`Programmable
`aperture plate
`system
`
`200 x Reduction
`particle beam
`projection optics
`
`Precursor gas
`injection system
`
` 3.5 PMLP system confi guration.
`
`Ion source
`
`Condenser optics
`
`Aperture plate
`Blanking plate
`Deflection electrodes
`
`1st lens
`
`Stopping plate at
`beam cross-over
`
`2nd lens
`Substrate / stage
`
`Electron source
`
`Collimator lens
`
`Aperture array
`
`Beam blanker array
`
`Beam deflector array
`Projection lens array
`
` 3.6 MAPPER system confi guration.
`
`and off at will. This approach is referred to as multiple aperture pixel-by-
`pixel enhancement of resolution (MAPPER).
` A variant of the multiple Gaussian beam approach is to project multi-
`ple shaped beams. In Fig. 3.7 (Slodowski et al. , 2011, Doering et al. , 2012),
`an array of 64 shaped beams in a single column replaces the single shaped
`beam depicted in Fig. 3.2b. Introducing multiple shaped beams represents
`
`© Woodhead Publishing Limited, 2014
`
`NXP Ex. 2004
`Impinj, Inc. v. NXP B.V. - IPR2020-01630
`
`

`

`Electron beam lithography
`
`
`
` 89
`
`MCC
`column
`
` 3.7 Multiple shaped beam (MSB) concept.
`
`Host
`
`Disk
`
`Gun HV power supply
`
`LENS power supply
`
`Pattern
`data
`
`Data
`correction
`
`Analog
`circuits
`
`Stage control unit
`
` 3.8 Multi-column cell (MCC) system concept, schematic, with character
`projection (CP).
`
`a relatively minor change to an electron column with a single shaped beam.
`This concept therefore builds on previously proven technology in an incre-
`mental way, thereby minimizing expense and risk.
` In Fig. 3.8 (Yamada et al. , 2008; Yamada et al. , 2010; Takizawa et al. ,
`2011), multiple columns are employed, with each column having multiple
`cell projection capability. This concept is referred to as multi-column cell
`projection (MCC). Additional columns can be added with modest effort
`and expense. Both systems depicted in Figs 3.7 and 3.8 have the advantage
`that each beamlet has many pixels. This permits one to use relatively few
`beamlets to achieve high pixel parallelism. In both cases, beams can be
`easily added to an existing platform. This permits one to add parallelism
`
`© Woodhead Publishing Limited, 2014
`
`NXP Ex. 2004
`Impinj, Inc. v. NXP B.V. - IPR2020-01630
`
`

`

`90
`
`
`
` Nanolithography
`
`DPG
`
`DPG lens
`
`Upper demag lens
`
`Wien filter
`
`Cathode
`
`Gun lens
`Condenser lens
`Field lens
`
`Lower demag lens
`
`Wafer
`
`Electrostatic
`bender
`
` 3.9 Refl ective electron beam lithography (REBL) system concept.
`
`in an incremental way, thus mitigating the risk associated with increasing
`system complexity.
` An alternative concept is shown schematically in Fig. 3.9 (McCord et al. ,
`2010). The beam from a thermionic electron source (cathode) fl oods a large-
`area digital pattern generator (DPG). This is an array of individual pixels,
`with each pixel independently addressable with a voltage. The beam is
`decelerated to a very low energy at the DPG, so that beam electrons are
`either refl ected or absorbed, depending on the pixel voltage. The refl ected
`beam containing the pattern information is then accelerated. The pixel array
`is demagnifi ed onto the writing surface. The electron column is compact,
`thus mitigating Coulomb scattering of beam electrons.
` An alternative approach is to project an electron optical image of a pat-
`terned membrane mask onto the writing surface. This approach, called pro-
`jection e-beam lithography, forms a demagnifi ed image of a pre-existing
`mask. It was fi rst conceived by Heritage (1975) and independently by Koops
`and Bernhard (1975), based on the operating principle of