PROCEEDINGS OF SPIE · and technical status of the critical mask industry. ... Taiwan Semiconductor...
Transcript of PROCEEDINGS OF SPIE · and technical status of the critical mask industry. ... Taiwan Semiconductor...
PROCEEDINGS OF SPIE
SPIEDigitalLibrary.org/conference-proceedings-of-spie
Mask industry assessment trendanalysis: 2012
Chan, Y. David
Y. David Chan, "Mask industry assessment trend analysis: 2012," Proc. SPIE8352, 28th European Mask and Lithography Conference, 835203 (16 April2012); doi: 10.1117/12.923746
Event: 28th European Mask and Lithography Conference (EMLC 2012), 2012,Dresden, Germany
Downloaded From: https://www.spiedigitallibrary.org/conference-proceedings-of-spie on 31 Jul 2020 Terms of Use: https://www.spiedigitallibrary.org/terms-of-use
Mask Industry Assessment Trend Analysis: 2012 Y. David Chan
SEMATECH, 257 Fuller Road, Albany, USA 12203
ABSTRACT
Microelectronics industry leaders consistently cite the cost and cycle time of mask technology and mask supply among the top critical issues for lithography. A survey was designed by SEMATECH with input from semiconductor company mask technologists and merchant mask suppliers to objectively assess the overall conditions of the mask industry. With the continued support of the industry, this year’s assessment was the tenth in the current series of annual reports. This year’s survey is basically the same as the 2005 through 2011 surveys. Questions are grouped into six categories: General Business Profile Information, Data Processing, Yields and Yield Loss Mechanisms, Delivery Times, Returns, and Services. Within each category is a multitude of questions that ultimately produce a detailed profile of both the business and technical status of the critical mask industry. We received data from 11 companies this year, which was a record high since the beginning of the series. The responding companies represented more than 96% of the volume shipped and about 90% of the 2011 revenue for the photomask industry. These survey reports are often used as a baseline to gain perspective on the technical and business status of the mask and microelectronics industries. They will continue to serve as a valuable reference to identify strengths and opportunities. Results can also be used to guide future investments in critical path issues.
Keywords: Mask industry, photomask, industry, mask yield, photomask yield, mask quality, photomask quality
1. INTRODUCTION
SEMATECH reinitiated a survey of the mask industry in 2002, intending to enhance the level of understanding of the unique and critical issues associated with the photomask industry.1 It planned to conduct the survey annually to provide a substantial and valuable reference database. This database has been built over time to help identify past trends and validate future projections. A similar survey had been done annually for seven years but was suspended after 1999 for because of lack of financial support to conduct it. The results from those past surveys were published as SPIE papers and presented at BACUS, where they were found to provide valuable insights for both mask makers and mask users.2, 3 From 2002 to 2011, the survey questions remained essentially constant4-15. Questions are grouped into six categories: General Business Profile Information, Data Processing, Yields and Yield Loss Mechanisms, Delivery Time, Returns, and Services. Within each category are multiple questions that provide a detailed profile of both the business and technical status of the mask industry.
2. ASSESSMENT APPROACH David Powell Consulting was the focal point for data collection. Participant identification markings were removed by them to make the raw data anonymous before the data were forwarded to the author. Data were requested from survey participants for the previous 12 months, from July 1 through June 30. The data were loaded into a spreadsheet and summarized for lowest, highest, and average values. Respondents received an Excel file with complete summary data for further analysis as an incentive for participation.
Invited Paper
28th European Mask and Lithography Conference, edited by Uwe F.W. Behringer,Wilhelm Maurer, Proc. of SPIE Vol. 8352, 835203 · © 2012 SPIE
CCC code: 0277-786X/12/$18 · doi: 10.1117/12.923746
Proc. of SPIE Vol. 8352 835203-1
Downloaded From: https://www.spiedigitallibrary.org/conference-proceedings-of-spie on 31 Jul 2020Terms of Use: https://www.spiedigitallibrary.org/terms-of-use
The survey participants by years can be seen in the table below.
Table 1 – Participants by year
Description of survey sections:
1. General Mask Profile Information This section seeks to establish a general description of the basic elements of the mask business: technology groundrule categories; application types (logic, memory, microprocessors, etc.); mask types by glass size (PSM, binary, OPC); magnification distribution; and distribution of fabrication methods used for patterning. The type of pellicle is also included.
2. Data Processing This section requests data on file size and data preparation time average, maximum, and 95th percentile of each respondent’s distribution. For the fifth time, write time data were gathered. The 95th percentile is requested to help discern what file sizes and write times are truly representative of the norm without having extraordinarily large, anomalous files over weighted in the analysis.
3. Yields and Yield Loss Mechanisms Overall yield by glass size and mask type (binary, phase-shifting method) is given. Fifteen yield loss mechanisms are offered to generate an industry Pareto chart. Binary and phase shift mask losses were separated since the 2007 survey.
4. Delivery Time Data on average delivery time and time for the first three layers of a new mask set are collected by four different mask type categories: binary, binary with moderate to aggressive optical proximity correction (OPC), attenuated, and alternating phase shift masks Note: Attenuated phase shift masks are also known as embedded phase shift masks or half-tone phase shift masks. Alternating phase shift masks are also known as strong phase shift masks or hard shifted masks.
5. Mask Returns This section collects data on mask returns according to nine categories.
6. Mask Service “Mask maintenance” services are divided into eight service categories.
Participants 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011Advanced Mask Technology Center YES W/Toppan W/Toppan YES YES YESCompugraphics International Ltd YES YES YES YES YES YES YES YES YES YESDai Nippon Printing YES YES YES YES YES YES YES YES YES YESToppan Photomasks Inc/DuPont Photomasks Inc YES YES YES W/Toppan W/Toppan W/Toppan W/Toppan W/Toppan W/Toppan W/Toppan
Hoya YES YES YES YES YES YES YES YESHynix YES YESIBM YES YES YESIntel Mask Operation YES YES YES YES YES YES YES YES YES YESMicron Technology Mask Shop YESPhotronics Inc YES YES YES YES YESSamsung YESSemiconductor Manufacturing International Company (SMIC) YES YES YES YES YES YESTaiwan Mask Shop (TMC) YES YES YES YES YES YES YES YES YESToppan Printing YES YES YES YES YES YES YES YES YESTaiwan Semiconductor Manufacturing Company, Ltd. (TSMC) YES YES YES YES YES YES YES YES YES YES
Total Respondents 7 10 10 8 9 8 9 9 10 11
Proc. of SPIE Vol. 8352 835203-2
Downloaded From: https://www.spiedigitallibrary.org/conference-proceedings-of-spie on 31 Jul 2020Terms of Use: https://www.spiedigitallibrary.org/terms-of-use
NOTE: Comof the overallappropriate, respondents. averages by r
1. General
MasDesadvaindienco
Logto odeclchipsust
mparisons amonl industry (as sthe results froGraphs with
response for al
Mask Profile I
sk volume for pite the economancement. Theicating a numbouraged many
gic designs are our endless appline while the mp (SOC) produtained.
ng the surveys seen in Table 1om the 2003 to
2002* have dll years.
Information
the < 45 nm mic turmoil in
ere were also sber of relevanproducts and a
still driving thpetite for mobimicroprocessoct percentages
3may not be tot
1). In addition,o 2011 surveysdata that are v
nodes increasn 2008 and 200significant incrnt products at application inn
he majority of tile application
or suffered a sigremained stab
3. DATA tally accurate snot all particip
s have been vovolume-weight
sed at an appr09, the high-endreases in mask
the mature novations with t
the mask volums. The memorygnificant declin
ble. More data
since the survepants answeredolume-weighteted except for
roximate growd device makevolume for th
nodes. The ecothe mature tech
me. The devicey percentage ine in 2011. Thpoints are nee
eys represent a d every questioed. All 2002 re2002. Charts
wth rate of 70%rs continue the
he ≥ 250 nm nonomic problehnology.
e landscape is s now trending
he RF/mixed sided to see whe
different crosson. Where possesults are averwith 2002 ar
% from 2007 eir investmentsode in 2010 anem in 2008 an
changing, perhg up after somignal and systeether the trend
s-section sible and raged by e simple
to 2011. s in node nd 2011, nd 2009
haps due me recent em-on-a-s will be
Proc. of SPIE Vol. 8352 835203-3
Downloaded From: https://www.spiedigitallibrary.org/conference-proceedings-of-spie on 31 Jul 2020Terms of Use: https://www.spiedigitallibrary.org/terms-of-use
Morhas avaicoul
Thevolucert
re than 84% ofsteadily increailable.) The mald be indicative
e production phume has remaiain amount of
f photomasks aased. (Note: Thask volume fore that some old
hase-shifting mined very low.year-to-year fl
are currently mhe question wasr the 5X magnider generation
masks (PSMs). Attenuated mluctuations.
ade on the 602s asked differeification is abouwafer steppers
have been excmask type is st
25 format. Oveently in 2002; thut the same as s are beginning
clusively madetable ~10-12%
r the last five yherefore, no co4X for the firs
g to be retired.
e on 6-inch ma%. The particip
years, this percomparative datst time in 2011
asks. Alternatiant mix could
centage ta are . This
ing PSM d cause a
Proc. of SPIE Vol. 8352 835203-4
Downloaded From: https://www.spiedigitallibrary.org/conference-proceedings-of-spie on 31 Jul 2020Terms of Use: https://www.spiedigitallibrary.org/terms-of-use
AggTheare toptimanand aggr
Pellcateseemnod
Thedeclrelat
gressive opticae no OPC categtwo possible eical exposure hnufactured usin
wavelength cressive OPC. M
licle usage seemegory. This coums to be upwae volume incre
e combined vollining in favorted to the incre
al proximity cogory also had axplanations: 1)has reached itng double patteconstant. Tran
More data poin
ms to remain fuld be an artif
ard trends for bease.
lumes for the rr of the high vease in ≥ 250 n
orrection (OPCa significant jum) year-to-year pts limit at 40 nerning as a waynsition from snts are needed t
flat over the yefact of the chanboth DUV and
raster spot e-bevoltage shapednm products.
C) was growingmp in 2011 aftparticipants vanm half-pitch y of extending single patternito see whether
ears with a twonge in the mixI-line pellicle
eam tools and td e-beam tools
g at a steady pter stabilizing aaried, and 2) ac
(hp)16. Todaythe half-pitch wing to DP mathe trends can
o-year higher thx of mask shop usage, which
the ≤ 50kV sha. Increase in t
pace until a sigat ~70% in fromccording to ITRy, 32 nm hp flwhile keeping ay temporarilybe sustained.
han normal usaps represented could directly
aped e-beam tothe usage of la
gnificant drop m 2007 to 201RS 2011 editiolash devices athe numerical
y provide reli
age in the “no by the survey
y result from ≥
ools have beenaser patterning
in 2011. 0. There n, single
are being aperture
ief from
pellicle” y. There 250 nm
n steadily g tools is
Proc. of SPIE Vol. 8352 835203-5
Downloaded From: https://www.spiedigitallibrary.org/conference-proceedings-of-spie on 31 Jul 2020Terms of Use: https://www.spiedigitallibrary.org/terms-of-use
The volu2010. Threplace Ssubseque
The perc
increased
ume percentagehe reason for thSEM measuremently need to h
centages of died dramatically
es for optical mhe significant iments. More dhave a better br
:database and in 2011, perha
metrology toolsincrease in 201data points are nreakdown in th
die:die were esaps due to the i
s using < 300 n1 is not clear. needed to see wis category.
ssentially the sincrease in mul
nm wavelengthInnovative optwhether we hav
same from 2006lti devices/shu
h were stable frtical means couve a real trend.
6 through 2010uttle in wafer lo
rom 2006 throuuld be deploye. We may
0. The die:dataots.
ugh d to
abase
Proc. of SPIE Vol. 8352 835203-6
Downloaded From: https://www.spiedigitallibrary.org/conference-proceedings-of-spie on 31 Jul 2020Terms of Use: https://www.spiedigitallibrary.org/terms-of-use
Theof resteadispbett
Desyear
e no repair percecord yield (se
adily increased proportionate iner see the trend
ign data file sirs of the survey
centage had beeee below) and t
from 2007 to 2ncrease in matud.
ze averages any. This represe
en stable from the mature tech2010 and then ure technology
nd 95th percentints a growth o
~35% to ~40%hnology volumdeclined slight
y volume than i
iles are increasf over 3X over
% with a big jume increase in 2
tly in 2011. Thin high-end vo
sing at a steadyr the last four y
mp in 2011. Th2011. The use ohis is more likelume. More da
y pace of 1.4X years.
his could be thof e-beam repaely due to the ata points are n
per year over t
he result air
needed to
the ten
Proc. of SPIE Vol. 8352 835203-7
Downloaded From: https://www.spiedigitallibrary.org/conference-proceedings-of-spie on 31 Jul 2020Terms of Use: https://www.spiedigitallibrary.org/terms-of-use
Pattthe incrsigndata201
Datayear
Theover
terning write tim95th percentile
reasing consistnificantly (44%a volume in 201 did not incre
a preparation Crs.
e average yield r the last six ye
me has been ge represents thtently since 20
%). This might 10. Such a cha
ease.
CPU processin
for binary maears.
athered since 2he e-beam pat002, 2010 waimply that e-b
allenge may ha
ng times are sta
asks was at a re
2007. The avetterning write s the first yeabeam mask wrave subsequent
arting to increa
ecord high of 9
rage is dominatime. Even th
ar when the 9riters were nottly been resolve
ase again, but m
97.6% in 2011
ated by the lasehough the dat
95th percentilest able to keep ed as the 95th p
more data poin
. Yields have b
er patterning tota volumes has write time inup with the inpercentile write
nts are needed
been slowly in
ool while ave been ncreased
ncreasing e time in
in future
ncreasing
Proc. of SPIE Vol. 8352 835203-8
Downloaded From: https://www.spiedigitallibrary.org/conference-proceedings-of-spie on 31 Jul 2020Terms of Use: https://www.spiedigitallibrary.org/terms-of-use
*Represenpaper by B
Thethre
Theaccoat abeneand
ts survey years (JuBrian Grenon.
e 6-inch PSM yee mask categor
e major procesount for more tabout 15%. Crefit from the fPSM losses.
uly to June). The 1
yields have geries in 2011.
ss-related yieldthan half of thritical dimensiofeed backward/
1995 to 1999 data
enerally increas
d loss mechanhe yield loss. Hon (CD) mean/forward proto
are any data avail
sed over the la
nisms are esseHuman errors (an-to-target is aocols often use
lable in the calend
ast four years.
entially constaadministrative always higher
ed in wafer fab
dar year reported i
Record yields
ant from year and manufactuthan uniform
bs. The chart b
in the 1999 BACU
s were achieve
to year. Harduring errors) a
mity and could below combine
US survey
ed for all
d defects are stable
perhaps es binary
Proc. of SPIE Vol. 8352 835203-9
Downloaded From: https://www.spiedigitallibrary.org/conference-proceedings-of-spie on 31 Jul 2020Terms of Use: https://www.spiedigitallibrary.org/terms-of-use
Delidrivtimeas ccom
As e
ivery times varver for deliveryes. Both attenucompared to b
mplexity are inc
expected, the d
ry significantlyy times. OPC muated and alterbinary masks. creasing.
Deliv
delivery times f
y by the typesmasks take lon
rnating PSMs ta In general,
very Times
for the first thr
of masks. Expnger to manufaake longer becdelivery times
for First T
ree layers are b
perience anticifacture due to tcause of additios have been s
Three Laye
etter than the a
pates this as ththeir increasedonal lithographstable even th
rs
average deliver
he yields are thd complexity ahy and processihough the con
ry times.
he major and write ing steps
ntent and
Proc. of SPIE Vol. 8352 835203-10
Downloaded From: https://www.spiedigitallibrary.org/conference-proceedings-of-spie on 31 Jul 2020Terms of Use: https://www.spiedigitallibrary.org/terms-of-use
Mashandrefin
sk maintenancedling procedurned as new cat
e service was rres may be neegories in futu
Average
required most oeeded. The “oture year’s surve
Delivery T
often for replather” category eys.
Times
acing damaged is growing at
pellicles. Retrt a significant
raining of in-fapace and nee
ab reticle eds to be
Proc. of SPIE Vol. 8352 835203-11
Downloaded From: https://www.spiedigitallibrary.org/conference-proceedings-of-spie on 31 Jul 2020Terms of Use: https://www.spiedigitallibrary.org/terms-of-use
Whether248
We received responding cthe photomas
The economithe landscapeindicating a ntheir investmapproximate volume, the RF/mixed sigsustained.
Mask yields yield loss, co1.4X/year. Thand keep pacindustry segm
The authorsInternational ManufacturinManufacturinalso thank Dcontributions
Most importtechnical assi
en the 2011 phre is a strong d
nm and 100X
survey data frcompanies reprsk industry. Th
ic downturn froe of the photomnumber of rele
ments in shrinkgrowth rate omemory perc
gnal and SOC
for binary, EAost, and delivehe productivityce with growinments addresse
s heartily thaLtd., Dai N
ng Internationng Company, L
Dede Adams ats.
antly, we wouistance and gui
hoto-induced ddependence on
of 365 nm.
from 11 comparesented more his database pro
om 2008 to 20mask industry. evant and innoking feature sof 70% from 2centage is now
C products rem
APSM, and AAery time. Daty of e-beam wrng file sizes. Ded by each mas
ank the particNippon Printin
al Company (Ltd., for their st David Powell
uld like to thanidance provide
degradation sern exposure wav
4. COanies this year,than 96% of thovided a good
009 and our unMask volume
ovative productizes. The mas
2007 to 2011. w trending up
mains stable. M
APSM are at ata file sizes coriters must con
Data preparationsk maker.
ACKNOWcipating compng, Hoya, Hy(SMIC), Taiwspirit of cooperl Consulting, I
nk SEMATECed in the genera
rvice rates are velengths. The
ONCLUSION, which was a he mask volumunderstanding
nsatiated appetiincreased signts using the msk volume forWhile Logic
p and microprMore data poin
an all-time higontinue to incntinuously be imn times vary a
WLEDGEMEpanies, Advanynix, IBM, Inwan Mask Shoration in providInc. for her rol
CH and its memation of this pa
normalized wnormalized da
NS record high si
me shipped and of the state of
ite for mobile anificantly for th
mature nodes. Tr the < 45 nmdesigns are st
rocessors trendnts are needed
gh. Hard defectrease with themproved to ma
among respond
ENTS nced Mask Tntel Mask Opop, Toppan Prding this valuale in data colle
mber companiaper.
ith the percentamage rate for
ince the beginnd about 90% of the photomask
applications hahe ≥ 250 nm noThe high-end dm nodes contintill driving theding downward to see wheth
ts continue to e 95th percentiaintain the curr
dents, which m
Technology Ceeration, Photrrinting, and T
able data to theection and Gre
ies for their fu
tage of masks r 193 nm is 9X
ning of the serof the 2011 revk industry.
ave a lasting imodes in 2010 andevice makers nued to increae majority of trd. The percenher such trends
be the major ile at a growthrent level of w
may reflect the
enter, Compuronics, SemicoTaiwan Semicoe industry. Theeg Hughes for
unding support
shipped, X that of
ries. The venue for
mpact on nd 2011, continue
ase at an the mask ntage of s can be
factor in h rate of
write time different
ugraphics onductor onductor e authors
his past
t and the
Proc. of SPIE Vol. 8352 835203-12
Downloaded From: https://www.spiedigitallibrary.org/conference-proceedings-of-spie on 31 Jul 2020Terms of Use: https://www.spiedigitallibrary.org/terms-of-use
REFERENCES 1. K.R. Kimmel, “A Mask Industry Assessment: 2002,” 22nd Annual Symposium on Photomask Technology and
Management BACUS, Proceedings of SPIE Vol. 4889 Part 1, pp. 1-14, 2002. 2. B.J. Grenon, “1999 Mask Industry Quality Assessment,” 15th Annual Symposium on Photomask Technology and
Management BACUS, Proceedings of SPIE Vol. 3873, pp. 162-174, 1999. 3. E. Gonzalez-La’O, “1998 Mask Industry Quality Assessment,” Proc. SPIE 14th Annual Symposium on Photomask
Technology and Management BACUS, Proceedings of SPIE Vol. 3546, pp. 10-29, 1998. 4. B.J. Grenon, Mask Makers Data Book, 2002 Edition, published by Grenon Consulting, Inc., 2002. 5. K.R. Kimmel, “A Mask Industry Assessment: 2003,” 23rd Annual Symposium on Photomask Technology and
Management BACUS, Proceedings of SPIE Vol. 5256, pp. 331-43, 2003. 6. G. Shelden and S. Hector, “Mask industry Assessment, 2004,” Photomask Technology Conference, Proceedings of
SPIE Vol. 5567, pp. 1-12, 2004 7. G. Shelden and S Hector, “Mask industry Assessment, 2005,” Photomask Technology Conference, Proceedings of
SPIE Vol. 5992, 2005 8. G. Shelden and P. Marmillion, “Mask industry Assessment, 2006,” Photomask Technology Conference,
Proceedings of SPIE 6349-2, 2006 9. G. Shelden, P. Marmillion and G. Hughes, “Mask industry Assessment, 2007,” Photomask Technology Conference,
Proceedings of SPIE 6730-2, 2007 10. G. Hughes, H. Yun, “Mask industry Assessment, 2008,” Photomask Technology Conference, Proceedings of SPIE
7122-02, 2008. 11. G. Hughes, H. Yun, “Mask industry Assessment, 2009,” Photomask Technology Conference, Proceedings of SPIE
7488-2, 2009. 12. G. Hughes, H. Yun, “Mask Industry Assessment Trend Analysis: 2010,” 26th European Mask and Lithography
Conference, Proceedings of SPIE 7545-2, 2010. 13. G. Hughes, D. Chan, “Mask industry Assessment, 2010,” Photomask Technology Conference, Proceedings of SPIE
7823-3, 2010. 14. G. Hughes, D. Chan, “The 2002 to 2010 Mask Survey Trend Analysis,” 27th European Mask and Lithography
Conference, Proceedings of SPIE 7985-1, 2011. 15. Y. D. Chan, “Mask Industry Assessment: 2011,” Photomask Technology Conference, Proceedings of SPIE 8166-2,
2011. 16. International Technology Roadmap for Semiconductors, 2011 Edition, Lithography Chapter.
Proc. of SPIE Vol. 8352 835203-13
Downloaded From: https://www.spiedigitallibrary.org/conference-proceedings-of-spie on 31 Jul 2020Terms of Use: https://www.spiedigitallibrary.org/terms-of-use