circuits in emerging nanotechnologies
play

Circuits in Emerging Nanotechnologies Rinaldo Castello University - PowerPoint PPT Presentation

Circuits in Emerging Nanotechnologies Rinaldo Castello University of Pavia Thomas Ernst CEA-LETI Seok-Hee Lee SK-Hynix Subhasish Mitra Stanford University Ion OConnor Ecole Centrale de Lyon Chair: Adrian Ionescu, EPFL Rinaldo Castello


  1. Circuits in Emerging Nanotechnologies Rinaldo Castello University of Pavia Thomas Ernst CEA-LETI Seok-Hee Lee SK-Hynix Subhasish Mitra Stanford University Ion O’Connor Ecole Centrale de Lyon Chair: Adrian Ionescu, EPFL

  2. Rinaldo Castello University of Pavia

  3. Thomas Ernst CEA-LETI

  4. FROM INTEGRATED CIRCUITS TO INTEGRATED SYSTEMS THOMAS ERNST Montreux Symposium on emerging trends in electronics 1/12/2014

  5. 19(:&%;."<"3=($&>&?$%;((@(,>(:;,>(:;&(&.A(B(( +"('(>&/&$?<(.&F(D?$-&:(A$,/&$>(( • @5*.'.1.54!.0!5%*0$!%.J!2.J,0!".>K • @5*.1.PB,! "&L!!37.MN!O<!$'-!0,%$*,-!4,0B,04!! 0##1.') 234*5)6789)*333):;)#<)*#4) • Q,$%*+!R!8&.#+&24! 3=;2)>))234*)&#..?@A) 6789)*333)B?<#) 9$C.0!4,1&#.'-5#*.0!4522%&,04D! ?A/?.%&A(95C!(<"3,%(D?.#E?%:#$,.3( ! *,#+'.%.(E!#$2$F&%&*E!&'!GHIIA! !"#$%&'()*+"),"$--.,/ !! =!>?@A!@%%!0&(+*4!0,4,0B,- !

  6. 19($&>&?$%;(:&%;."<"3=(A$,/&$>(( aKIH!E,$04!! • >9X"!"#$%&'(! • U^9!R!,1F,--,-!1,1.0E! T,B&#,!>,'*0&#!?B.%5P.'! • YW_$'$%.(_+&(+!B.%*$(,!O! 34#$%&'(<! @-B$'#,-!ST!!! • Ia!E,$04!O! U,J!1$*,0&$%4!/.0!$-B$'#,-!.2P.'4! • V.J!2.J,0!3WT"X7N!4*,,2!4%.2,O<!! • ?1FA!YY@9N!9Y@9N!U?9"!YW!R!"J&*#+,4! • "E4*,1!#,'*0&#!! ?',0(E!4.50#,4!$'-!1$'$(,1,'*! • Q,*,0.(,',.54!4E4*,14! ?1F,--,-!.2P#4! • ?1F,--,-!4,'4.04!3F&.N!,*#A!<! • "$/,!,%,#*0.'&#4!3$5*.1.PB,N!$&0N!42$#,<! • Ga!E,$04!O! 7',Z$#*!#.125P'(! • U,50.1.02+&#! • [5$'*51!#.125P'(!\! • 80,$`*+0.5(+!&'!-$*$!*0,$*1,'*4! "?MN!9.]!N!"2&'N!O! • 8&.!#.125P'(\! • @-&$F$P#!\!!AA.*+,04!! • G(8H( =!>?@A!@%%!0&(+*4!0,4,0B,- !

  7. I)19(&/"<#J".0(E"$(<"F(%".>#DKJ".0(;,3;( K&$E"$D?.%&>( ! Novel computing paradigms 5- 15 years ! Beyond CMOS hybridization Advanced concepts ! Bio-inspired 3D process ! Logic-on-Logic : ( Monolithic and 3D-Stacked) ! Monolithic 3D memories 0 -10 years Pitch <0,1 ! m ! PHOTONICS The alternative to “More ! RF/MEMS – ANALOG Moore” scaling ! Biochips ! Novel substrates Pitch ~1 ! m ! Si/Smart Interposers ! Memory on Logic Today ! Logic on analog The packaging evolution ! MEMS on logic Heterogeneous Era ! RF/ANALOG ! 3D imagers ! Biochips =!>?@A!@%%!0&(+*4!0,4,0B,- !

  8. !"D&(M$&?-:;$"#3;>':"F?$A(,.:&3$?:&A(>=>:&D>( !Q.J!*.!#.'',#*!_2$],0'!+,*,0.(,',.54!-,B&#,4!$*!%.J!#.4*!J&*+!4,B,0$%!*E2,4!./! • &'*,0#.'',#P.'4! 3777K^!.'!"&<! Wb"7XU!6!?1F,-!V.(&#!c!1,1.0E!c!'$'.K4E4*,14! • ".50#,6!M"9>! ".50#,6!V?M7! G(8L( =!>?@A!@%%!0&(+*4!0,4,0B,- !

  9. 4OP&?<:;( >.%%$F.0$P.'46!L+&%&24N!7'i',.'N!"&,1,'4N!".0&'N! "4C,&%#'.'&D%#<,&E) "M9&#0.,%,#*0.'&#4N!72-&$!N!8&.1,0&,5Z!O! ! W%,Z&F%,!45F4*0$*,!&'*,(0$P.'! 7'*,0#.!R!YTV!d,Z&F%,4! ! 7'*,(0$*,-!$'*,''$!.'!/.&%! ! ! b!V.J!2.J,06!! ea'1! " !Gf'1!WT"X7! ! YW7T!N!!YW!4522%E! ! ! 8&.#.12$PF&%&*E!R!+,01,P#!2$#`$(&'(! 7'*g(0$P.'!R!2$44&B$P.'!J$/,0!%,B,%! ! b%*0$K41$%%!+&(+!2,0/.01$'#,!2$44&B,! #$2$ST6!IhW_11G! ! G(8N( =!>?@A!@%%!0&(+*4!0,4,0B,- !

  10. Seok-Hee Lee SK-Hynix

  11. !=DK">,#D(".(4D&$3,.3(Q$&.A>(,.(4<&%:$".,%>( ",44&.'!I!6!?1,0(&'(!M,#+'.%.(&,4!!K!>&0#5&*4!$'-!T,B&#,4!&'!*+,!U$'.!?0$! L$',%!!!!!!!6!>&0#5&*4!&'!?1,0(&'(!U$'.*,#+'.%.(&,4! )&>,3.(E"$( R"F(K"F&$(S(P,3;(K&$E"$D?.%&()TU5( ) 2014. 12. 1 Seok-Hee Lee R&D Division, SK hynix

  12. Memory demand split across multiple sub-segments @!B$0&,*E!./!-,B&#,!J&%%!$22,$0!$'-!+&(+!2,0/.01$'#,!R!%.J!2.J,0!! TY@9!J&%%!F,!',,-,-!$4!.',!./!*+,!1.4*!&12.0*$'*!0.%,4!&'!*+,!4E4*,1! Low Cost R"F&$(Q$*((R&?-?3&(VR"F(W"F&$X Automotive Smart Devices PC / Server DRAM Wearable Low Power High Speed Devices Higher Tr. Performance (High Speed)

  13. Transistor for Higher performance & Lower Power U,,-4!+&(+,0!2,0/.01$'#,!2,0&2+,0$%!*0$'4&4*.0!/.0!TY@9!$'-!&*!&4!$j,#*,-!FE!*+,!4%.J!4#$%,K -.J'!*0,'-!./!M.Z! Jox Limit for Logic Tr [Source : ITRS 2003~2007] DRAM Tr Nitrided SiO2 Jox Limit for DRAM Tr 25nm node Jox [A/cm2] Tox [ Å ] HKMG Logic Tr HKMG Tox [ Å ] Gate Length of Peripheral Tr. [nm]

  14. Low Power DRAM design Technique example ^$0&.54!#&0#5&*!*,#+'&r5,!&4!F,&'(!4*5-&,-!/.0!%.J!2.J,0!TY@9 Power Gating Multi-VTH Dynamic Body Bias Reduce leakage currents by inserting a Reduce leakage power where speed Controlling body bias, high speed & lo Objective switch transistor into the logic stack is not needed by high Vt transistor w leakage current can be achieved Q&(+!"2,,-N!Q&(+!V,$`$(,! >%.#`!4&('$% L Vdd T,%$E!%&', Logic L Vdd Cell L L >.12$0$*.0 L L L L T,#.-,0 Logic Virtual Scheme Cell Ground 8&$4!o,'A Y,-5#,-!"2,,-N!V.J!V,$`$(,! Switch @12 sleep Cell H ^ TT ^ "" H L L L L V"7 L L k".50#,!6!V.J!L.J,0!T,4&('!9,*+.-.%.(E!$'-!T,4&('!W%.JK!l@U!9A!Y@8@?m! !!R!9A!9&E$n$`&!,!$%N!@!IAGo7L"_p!5L0.#!54&'(!42,,-K$-$2PB,!^*!>9X"q

  15. Solutions for Higher Performance & Higher Density Y,r5&0,!+&(+,0!42,,-N!%,44!7_X4!4#+,1,!-5,!*.!20.#,44!-&s#5%*E!$'-!+&(+!#.4*!./!M"^ � TSV-Stacking Density # High Density � Conventional Stacking High performance # High Density, Low power Low Performance # but, High Cost & Process Difficulty Wire Bonding [16Gb TSV, SK Hynix : 40nm 2Gb x 8] � Memory Requirements . Lower Power � Non-Stacking . Higher Performance # High Performance, Low Density . Higher Density Performance

  16. High Reliability & Low Cost DRAM Y,r5&0,!#.4*!0,-5#P.'!*+0.5(+!-,#0,$4&'(!-,/,#*!0$*,!FE!X'K>+&2!?>>N!LLYN!,*#A Item Objective Effect, Status Issues Compensate Retention Time Degr Advanced Refresh Refresh Increase IDD adation Smart Refresh Refresh Improve LtRAS VBB Temp Modulation Write Write Recovery Time DFR (Design For cell Reliabilit y) POD (Post Over Drive) Refresh Compensate Sensing Margin Increase IDD On chip ECC Error Correction Area Overhead PPR (using ARE) Error Aging Fail

  17. Subhasish Mitra Stanford University

  18. The Next 1,000X Subhasish Mitra Collaborator: H.-S. Philip Wong Department of EE & Department of CS Stanford University

  19. Abundant-Data Applications 27

  20. N3XT 1,000X Energy Efficiency 3D Resistive RAM Massive storage Monolithic 3D Carbon nanotube FET compute elements Inter-layer vias: thermal STTRAM 1,000X TSV L2, L3, … density Carbon nanotube FET (or even silicon) thermal compute elements 1. All data active on-chip 2. Computation immersed in memory 3. Variability, yield, reliability 28

  21. Carbon Nanotube FET (CNFET) 1. First CNFET computer CNT: d = 1.2nm CNFET d 2. High-performance CNFETs I ON ( ! A/ ! m) Sub-litho CNFET Si FETs pitch (Stanford fab) (foundries) 29 [Shulaker Nature 13, ISSCC 13, IEDM 14]

  22. First CNT Monolithic 3D IC Conventional vias, Process temp. no TSVs < 250 o C Logic + memory Inter-layer digital circuits CNT 3 CNFETs V OUT (V) 2 Resistive RAM 1 Resistive RAM Silicon FETs V IN (V) 0 0 1 2 3 30 [Wei IEDM 13, Shulaker VLSI Tech. 14, IEDM 14]

  23. Ian O’Connor Ecole Centrale de Lyon

  24. ],/&(^$?.A(9;?<<&.3&>(E"$(9,$%#,:>(,.(4D&$3,.3( +?.":&%;."<"3,&> ( =%#FA)*?<)GHI#<<#% ) ,?<A#&#<<#%J'&K.L#<AF% ) *<EMD$D)N'E)B?<#D'&O<#.#1,'E)N')2L#< ) P<,Q'%E,DR)N')2L#<)S)3&#.')I'<D%?.')N')2L#< ) 1.>J:#:(A&>(+?.":&%;."<"3,&>(A&(R=".(Y5T(9+T!(Z6N7((((((( ! ! ( ((((((((((((((((((;[K'\\,.<*%.$>*E$(

  25. W&B,!(0$'-!#+$%%,'(,4! IA Y,-5#,!,',0(E!#.'4512P.'!$'-!&'#0,$4,!,',0(E! ,s#&,'#E!./!-$*$!20.#,44&'(! GA b4,!*,#+'.%.(E!*.!&*4!%&1&*!*+0.5(+.5*!&*4!%&/,! #E#%,N!$'-!,'450,!0,%&$F%,!4E4*,1!.2,0$P.'!&'!*+,! 20,4,'#,!./!F.*+!5'0,%&$F%,!-,B&#,4!$'-! &'#0,$4&'(!#.12%,Z&*E! SA T,B,%.2!+$0-J$0,!4.%5P.'4!/.0!:8!-$*$!4*.0$(,! 45&*,-!*.!*+,!F&(!-$*$!,0$! tA T,B,%.2!4E4*,14!$-$2*,-!*.!',J!4,0B&#,4!39G9N! 7.MN!#%.5-!#.125P'(!AAA<! aA L0.1.*,!$'-!,Z2%.&*!*+,!2.*,'P$%!./!ST! +,*,0.(,',.54!&'*,(0$P.'!

  26. ?',0(E! • )GG'1! – 4J&*#+&'(!I!F&*!&'!$!*0$'4&4*.0! #.4*46!! (_8O87(?`( – 9.B&'(!$!IKF&*!-$*$!.'!4&%&#.'! #.4*46!! (_8(K`\DD ! • uoQn!4J&*#+&'(! ".50#,!6!YA!T0,#+4%,0N!bA!80,1,'! • uo!*0$'4&4*.04_#1 G! • IH S !$F.B,!*+,.0,P#$%!%&1&*! `AMA%'3G<! 3^.'!U,51$''KV$'-$5,0! v I(a`(VIb87 O68 `X( F.5'-<! )SHHw!

Recommend


More recommend