121
2006 IEEE Radio and Wireless Symposium RK p 82 Nanostructures: RF Characterization, Modeling and Applications Sunday, 15 January 2006 2.00 pm - 6.00 pm . 0rganizeJ's: Steve Islamshah Amlani, Motorola Labs Keith A Jenkins, mM Chairs: Steve Islamshah Amlani, Motorola Labs Keith A Jenkins, mM 2008 IEEE RADIO AND WIRELESS SYMPOSIUM San Diego Convention Center San Diego, California, USA 17 - 19 January, 2006 http://www.radiowireless.org . General Chair: Fred Schindler, RF Micro Devices Technical Pro,gram Chair: Mohammad Madihian, NEC Laboratories America, Inc. Technical Pro,gram Vice-Chair: Xiaodong Wang, Columbia University Proceedings' Editor: George Heiter, Heiter Microwave Consulting Radio & Wireless 2006 Sponsors: IEEE Microwave Theory and Techniques Society (MIT-S) IEEE Communications Society (ComSoc) """1'

2006 IEEE Radio and Wireless Symposium

  • Upload
    others

  • View
    9

  • Download
    0

Embed Size (px)

Citation preview

Page 1: 2006 IEEE Radio and Wireless Symposium

••• • ••••••• • •••• • • •• • •• • •• •• • •

2006 IEEE Radio and Wireless Symposium

RK p 82 Nanostructures: RF Characterization, Modeling and Applications

Sunday, 15 January 2006 2.00 pm - 6.00 pm .

0rganizeJ's: Steve Islamshah Amlani, Motorola Labs Keith A Jenkins, mM

• Chairs: Steve Islamshah Amlani, Motorola Labs Keith A Jenkins, mM

2008 IEEE RADIO AND WIRELESS SYMPOSIUM San Diego Convention Center San Diego, California, USA 17 - 19 January, 2006

http://www.radiowireless.org

. General Chair: Fred Schindler, RF Micro Devices

Technical Pro,gram Chair: Mohammad Madihian, NEC Laboratories America, Inc.

Technical Pro,gram Vice-Chair: Xiaodong Wang, Columbia University

Proceedings' Editor: George Heiter, Heiter Microwave Consulting

Radio & Wireless 2006 Sponsors: IEEE Microwave Theory and Techniques Society (MIT-S) IEEE Communications Society (ComSoc)

•"""1'

Page 2: 2006 IEEE Radio and Wireless Symposium

!

I Workshop WS2

I

I "Nanostructures: RF Characterization,

Modeling And Applications"I

I Sunday, 15 January 2006,2:00 PM - 6:00 PM

I

I Organizers:

I Steve Islarnshah Amlani, Motorola Labs

I Keith A Jenkins, IBM

I Speakers:

I "Carbon Nanotubes as Carbon Nanotubes as Microwave Devices" I Peter Burke, University of California at Irvine

I "High Frequency Measurement of CNFETs: Experiences and Outlook" Dinkar V. Singh and Keith A. Jenkins, IBM TJ Watson Research Center I "Single Single-Walled Carbon Nanotube Walled Carbon I Nanotube High-Frequency Mixers" Sami Rosenblatt, Cornell University I "HF characterization of CNTFET'I Henri Happy, CEA Saclay - Commissariat a l'Energie Atomique

I "Design Considerations for High Frequency High Frequency "on on-wafer wafer" measurements"I

Glenn H. Martin, Universal Filmworks I "Challenges in Measuring S-parameters of High Impedance States in

Carbon Nanotube Structures"I Loren Betts and Hassan Tanbakuchi, Agilent Technologies

I "High Impedance S-parameter Measurements" Jon Martens, Anritsu CompanyI "Testing Frequency Response of High Impedance CNT Transistors"I Dan Woodward, Tektronix .

I

I

I

I

I

I

I

~

Page 3: 2006 IEEE Radio and Wireless Symposium

••••• •• •• • • •• ••• • • • •••

"Carbon Nanotubes as Microwave Devices"

Peter Burke University of California at Irvine

I

I

I

I

I

I

I

I

I ,S

Page 4: 2006 IEEE Radio and Wireless Symposium

•••• •• • •••••

• • •

I

I

I

I

I .carbon.Nanotubes as

••••

I Microwave Devices

Peter I. Burke '. Integrated NanosystemsResearch Facility

Department of Electrical~ngineeringand Computer Science University o~ California, Irvine

Funding (in alphabetical order) ARO, DARPA, NSF (ECS, DMR, CCR), ONR

• c 200S Peter J. Burke. AU·Rights Reserved .IEEE Slide 1

Outline· .

-Nanotubesynihe~is ' .

-DC electroniC properties .' . . -Resistance vs.length .

-AC Applications: Passives '. .. ' -AC Applications: Actives

.. -Nanotube":Aiitenna~ ... ",.. '.'

••

I

•I 1

Page 5: 2006 IEEE Radio and Wireless Symposium

Synthesis

1k"Ii",~gJ,,,_·_J,a... .... f'!41"jJ~.t~r".it;.(~'!tP.Ji'i!:t)~

*"'l"i~r0M«jtit,::f,t.1%XQ RfP!I'~_~ Af.mv",ryt['{ R,-(~h'Xi .}~'1\· JA ;t)().-I

In thl\ U<..d ..,,_~_~)lq(~rMfI_t_~ ~~l~#. <In';t~'5- ilf ,1,c. 111m _ ~"'lJ.1~ ~k~~l'l1Inoo .no~~.. (~f·

"(~,tj:";iDJt;'l ~~ (j:'~ "'Jl:n ~~1:ftil'tl*-.3nl! Hl-:t'l, Hlt" r~d;Ull:t<OtJ~ l,j,~I'Jri.'I-h.~ U~, ~t'~ ill' .t~,

.. ­1 Nowr~~?JCOA"p('IJ~P

;<"'lQ:z;t~

<:> 2005 Peter 1. Burke, All RighI> ReseIVed .,EEESllde3

-= E

E

E

II:

II:

II:

_~~I.t~

-­~_~_"""""""~w;""'""'_~ ,..::t'.<R<~""'_"*"""_ ...' ~ .....-~.~l'A!ll~~I'I_~~,\'l' ~....,....,..~ ___ ....~f!"<,~~~'""OWIr~~...t~ ~~__"'~.. .==.~==~..~WO::::~~~-~~,

IF

•••• -=

Longest contacted SWNT

II A

E1ectrJg1 Properties ofOAml l.o11gSlngIe-Walled. C8ri:IoIl NanoIubes .. ·~..-,'~~..,,.i~ :~~'_~;~~-~,IiJ ....,----..:..:.._-"".•... __._-'------.. "\-~l<~i..~_-f~z",,~,H>'l*~~~ .

•••••• I

I

2 •

Page 6: 2006 IEEE Radio and Wireless Symposium

•• • •

•••

• • •••

•• •

••

••••• I

••

•••••

•••

Electrically contacted SWNTs

10000 ,

i II

1000 4

• ~

100 J !

II

1 i......

10 ~ !

1~ !

0.1 ~ Y

II

II II

II .. 116

II

ill i

0.01 q; c: .!!J

'" '" lij-, '"~

~

0 <;> c: ~

C? ~

'" <;> c: .!!J

9 fo{ a

q ~ 0

~ a1 0

Date

Peter J. Bwb. Zben Yu, S. U. C. Rutherglen ·NanoIube, for ltF and Microwaves· (in¥ired plenary talk) Proc..ofEu,opean MicrolW1"e Week 2005. 1·5 (2005).

o 2005 Peter J. Burke. All Rights Reserved SlideS+IEEE

Resistance Vs. Length

BedricaJ Ptopertles of G.4C/Il LongSiQgl&.WaIled Carbon NanoIubes . :......4i..t_w.: ....._ ...t~

~~~~~~~.~ ~~~.~f.~~-.~"!'t"-

"'~~:$ IllIlilQhm

~

~"""""~llO'J"l<_*"""'"_"""",~~_ .._b~,""'_'-.iJIlf

~-;.~~~~==::'~::~==i:-==~= Illl<Ol>rn ........ -i ::::=""~:.*:~=t,.~~~J;;:=:c.~-

3

Page 7: 2006 IEEE Radio and Wireless Symposium

An RF Circuit Model for Carbon Nanotubes p, I. Blld«:

~4IMmrtt-\\'4t ~p _.- .1 dt'C'.alt .~ f«Jr .,... v.a;W ~ ......M far&Mh 1lk.1lHl Q1pad.th·~~~tW lIt!'tMilt-> erin. B~'~ the- ••ItIJlQbI lh OIl UDdtndtiDtWiioo U- with db~ klbt'Ik ... ~t'tic"ducu.ae:e-ali: 'm'I.ll~ dl.'ltrara.t.d q-'i1J1tIl1ll ",.d -dffltonad<' ~lIIplkJf_«._~,tK<~... ~~ckpHHllst~t'C".rot.'"1IlNtyofllK"Mgmata' iJ"" _enn. 'E:S,fiUa:e ~ ".UQ'Il"'~ DC' k

,. l· (~t1I~~l.O*F*!Mim.'Ji).'o~~roa.s.~~·'l8\l! ....""dM, to ttindly ~_Ibt )'dot ~.... OM ...... ddWedp«oB#it idtI.Jh._Ipb._,Ito ........"YY _.lIft or. L!lftIII#<t 1lqoW.

c> 2005 Pe"" J. Burke. All Rights Reserved +IEEESlide7

:~ .,\1, ""

~

2:; .~;r;­-

"'.----'--------,

6 I

:=::~~~~~~~$t.=---_.....,

SWNT: 10GHz Interconnects 8 rr=====;,..------,

4

Page 8: 2006 IEEE Radio and Wireless Symposium

•••••••••••••••••••••••• I

I

I

I

I I

I

I

fa- ;;- ,p-lJt? G~ ~- uJi:~ trJ &tJ L :> K

~ J ~~ cJ:.e uJ([J.fl.-fc i cJ"\~ V-~ f4C-{) P~e1L ~. L

~~*'--""~~~ _t__~jll.~~~

AC l'Crformance or n;l1wc1e'c\rO""",; toward< a ballislic . 1'1:lU>:lD()(U!>e IransistQr

P&r"J•. 8lIik(' ~

soo .~~'~~~~p'"200

N 100

8 50 ..r

20

10

Si

'".'-....

GaAlO

0,05 OJ 0.2 0.5 I 2 5 10 Gale Length ((.1111)

'io ......._~_~("",~_.....,.,-_.-~._~("t_"­

""-.-....­~4;p(--"""""".."'" •

Lk .:-- 4'1 ++,0f1Vt

rrJ P~8 ~ 1fttV'

~ '" (Rs + Rtl)Cp .• + -}- (e" "" (~.p'" Cpl.)-rzlr ,.,m g~i ~ . .."..", ." "1 .,. ;;•. -Rs .,. RollC", .,. C,d,> ."' (til.e. .':)-Illf

., 2005 Peter J. Burke. AU Rights ReseIVed

+IEEE

,,""'-_..:?:JII--....._

-:l

f .. i -6

.e

5

Page 9: 2006 IEEE Radio and Wireless Symposium

First Microwave 300K eNT FET Mk1'Q:\\'aw NnRoIobt: ·h:;Ul...~tOf (~tit'U m"lHg-h Hin~

ZV••CR""""",PJ.a."", ~_~"~""-'-Abw."'J>f""'(~.~'~4¥k.:W/<I....1E'!p#-p:{......:t,('~&->'''''

{~«,(".Il't~w~4~-!i';i)t~·);,.9.1

w~ *M\IR:.~·~ 1 Gfli~M~«.'Dd~t;.""'C:ofOllUd-p~ li."~ l:PIW~~fid(te<<e'';~M-001b~·_tlW1l,d;:·'lHlb:\''OI~.Man~,,w~dk ~~~~~ilt_1G}J~-~~m-l&t~'~r~~> -{~ 1l;lo.dl:co p'• .tkm of;:l: ~'~'Y~w.1i ~ .. , ('dk ~ wan Np~.i/: ~fk~ ~U'f((l~¥d~.a Mt'l:'b<lft,"~I~~l;UIUI:(lmbc~W~ ~a.r ...hni';M'I'.~t!ll'lflfiGer::..-

;?~)J.h'.~!JHLl,-\1~f!.;)'J.J.2(2005).

'~::._c.,~~~~lOt·~ .•T,-5-"'·tI.'lts..srfEJ. ~~'l<~~~~.

hi «dtt {:) lnI1~ me dy~C'i1 iml-~ at 1l~*,W ~'""¥' ~e ~ rtwo NOX" It\:"tJXiir.lc: ~I 'R'C! 1~11?v lLd ~ PW~ ~:lIJIic: S:"'}'o.'·"N:S:1 '"

1>"-_-'-_-1.__'-_-'-_-' 0.0 ..0.5 .L{l -15 .):J~ .2.

v,:tJVt

<c 2005 Peter J. Burke, All RigblS Reserved Slide 11.IEEE

_____ctIT ·Ante~DDl...Ula_S:....-..~__..,-.. r--~-----:~'----l

Quantitative TbeoryQf Nanowire aOd NanotUbe AntemtaPerformance

Page 10: 2006 IEEE Radio and Wireless Symposium

•Ii

•• ! •

•••• •I

••••••••••••••• I

I

I

I

I

I

"High Frequency Measurement of CNFETs: Experiences and Outlook"

Dinkar V. Singh and Keith A. Jenkins

IBM TJ Watson Research Center

Page 11: 2006 IEEE Radio and Wireless Symposium

••• •• • •••••••••• ••• •• •• • • • •• ••

•I

. .: '. . . .

. High Frequency MeasurementofCNFETs: .. Experiences and Outlook . .'

D. V. Singh K. A. Jenkins

IBM T. J. Watson Research Center Yorktown Heights, New York

January, 2006 +IEEEs,icle1 .

Outline'· .

>Challenges' ,

»,~!,~e oomai~,~~a~Urerttentl; '. . . •. >F'requenc}"Dornall"tlVleasurements . "dir~ct .. '.' .

1

Page 12: 2006 IEEE Radio and Wireless Symposium

Challenge 1 :.Signal Level Multigated CNFET

....-,.....",.....,'"'"""T""""'T....................... 121-1 ,--,,---.-,.----,-----..---. 40.01-1 ..10.8 USj'\

.' 101-1

30.01J / :~ ,I \

,I ~ ~ « 20 0 I -E 41-1 ,I \...." .1-1.--:_ ~J

. '"""-~~ OJ 21J /

10.01-1 0 -, J Vd=1.1 V -21-1 \ Vd=1.1

0.0 -41-1 L-..J~--l._--'---'--:---'-5 -4 -3 -2 -1 0 1 2 3 -4 -2 0 2

Vgs(V) v (V) };>Single tube CNFETs typically drive only a feJi IJ.Amps ~Extremely low signal power

.- 4pW in 50 ohms or -84dBm for this example ~ Sensitivity of a network analyzer - -100 dBm ~ Spectrum Analyzer - -110 dBm

IV\. .IEEE Slide 3 MTT$'~ ~

IQ

&~~,"(A·

Challenge 2: Parasitic capacitance . R

~ Conventional RF techniques fail for such small devices

• Small capacitive signature precludes S,:,parameter ,meas• • parasitic probe pad capacitance >,> than the intrinsic device capacitance

'.... ' .

..... '0: '·'0

, .0..

Yp3

II:: II: 11= II:

II:

•11=

• •..II:

• II

II

IJ

I

I 2 I

I

Page 13: 2006 IEEE Radio and Wireless Symposium

-----------------

I

I

I

I

I

I

I

I

I

I " I

I

I

I

I

I

I

~;~Achieving Low Parasitic capacitance\o."""""A _

", Typical CNFETs with substrate used as gate has large Cparasitic

~ Cparasi6c in Top-gated FETs considerably lower

- Reduced gate-source/drain overlap

- Enhancement mode FETs can further decrease gate-sorce/drain overlap

- Insulating substrate e.g. quartz significantly lowers pad parasitics

S-parameter measurements

'on= 0'" 1,OpA

Top Gate FET

.32 •

INll l~nco -36

---" -40 N en

-44 g"oe i'lt..­-48 1x10. 1x10'

Frequency (Hz) +IEEE Slide 5

~s=I ~~~.= ~r-~CA ..·.

•I

A

I

•I

•I

• •I

Top-gated CNFETs

B source/drain tube dispersal patterning

I

I

I 3

I .

Page 14: 2006 IEEE Radio and Wireless Symposium

IEEE,Radio

5~~_" l~O_p_, _G_a_ted_'_C_N_F_ET_,.....5_:_ID_-_~_GS _ L

II =300nm

,After SID metallization Complete fabrication

Tox = 100nm , " To" = 10nm

I

b10n

100n

1n~__~~_~-,-~..... ~30 -15 0 15 -1 0 1 2

Vas (V) Vas (V)

»Devices transform for ~ toambipolar during processing ,

>- Attributed to change in SB height at the CN-metal interface'

IV\. +IEEE SOde7 M'rT$'*

Tiansf~rChar '" ',' ','

Top,GatedCNFETs :,.,DC Characteristics '

Page 15: 2006 IEEE Radio and Wireless Symposium

I

I

I

I

I ~ Time Domain Measurements I

I

I

I

I

I

I

I +IEEE SIl~9.·I

I

I

Low Id,sat makes time domain measurements challenging » I"n - ·1 J.tA/tubein 50 Ohms produces only 50flV, below the detection·

limits of conventional oscilloscopes

I

I

I

I

I

I

I

I

I

I

I

I

I 5I

I

Page 16: 2006 IEEE Radio and Wireless Symposium

VGS =av, VDS =1V, AVGS =1.5V, Input pulse tr =10n5 10r----......._-~__...--___.

>E··8 "-'

~ 6 ""5 0.. 4 .....'::l .

"5a.

2 o

, E -CNFET , ~ - - nMOS , ,. --..',;- _.

". ~,/..:: .. :1 " " ~

3 l l,

0~~---r-_...::.::~~=-1 o 100 200 300

Time (Jls)

~Risetimeof the CNFEToutput pulse is -10 IJS

~ First demonstrated switching of a CNFET at 100 KHz

IV\. ... .+IEEESllde11....... ,,~ ,.

10r-·--_---­ ~.,

.. '~"-."

-.;"\

!;!r$~t~~~;~ 'f?1" ·ic~I(~p()P.

;~1~Ec~:~':.;;;':;-~~~~;:tl,~J

E:nme Domain'Measurements m: 11=

a: II:

It

IJ:

It

Eli:

a: I:r

a a

6 a g;

Page 17: 2006 IEEE Radio and Wireless Symposium

I

I

I

I

I

I

I

I

I

I

I

I

I

I

I

I

I

I

I

I

I

I

I

I

VG

200K

Spectrum vG

,.....---,l------Hl-4----------1r W n

~ Gate of CNFET excited with a large sinusoidal signal.

>- Spectrum analyzer measures power spectrum at output

, ;.. High bandwidth,high sensitivity measurement '

19\ ,IF­ "IEEE Slide13 Mn'$~ ~~~ .

Device layout and impad:and cross-talk

7'

Page 18: 2006 IEEE Radio and Wireless Symposium

Frequency Domain Measurements

-60

-120

I-BO ~

~-100 :~

<> <>

~o ••

~ .

.. 5>65 . ,~

.. ~\ . . <> PcT,meas

~- CrossTalk Power - ­ PCT, fit

100k 1M 10M 100M 1G Frequency (Hz)

). CS mode: From cross-talk ,CgJparasitic) N 15fF

.IEEE Slide 15

m:Frequency Domain Measurements

• •CSmode •

Total Power ~""1 - \r--- .....

iF db

a::-60 • 1:1::

It:

a: It:

a: I:

I:

E:

I:

I:

-= 8 I:

I:

Page 19: 2006 IEEE Radio and Wireless Symposium

••

•g

••••••• •I

•I

•I

I

I

I

I

I

I

I

I

I

I

I

I

I

I

I

CSand CGconfigurations

..·..···········r..

R[ Yo R[ Yo

................1.......... ................1.........

(a) (b)

CGmode CSmode 'YCS mode: Coupling capacitance given by gate-drain overlap, Cgd

~ CG mode: Coupling capacitance given by S-O capacitance, CSd'

__ ~..;..•..'.' Expected to be lower than Cgd

#'Y' ~ "'IEEE. S1lde17 MTTS ~ t:U":~1~

Frequency Domain Measuremenb .

9

Page 20: 2006 IEEE Radio and Wireless Symposium

CI

~

a

~

Simple Power Subtraction

1)J 11 CG, me"s • CS, meas -Fits

100k 1M 10M 100M Frequency (Hz)

>- vdiff extracted from measured scalar powers using Vdfff"= (RLPT)l/2 - (RLPCTY/2 :>- Model prediction in good agreement with measured Vditl" ..• Coupling capacitance used in model is extracted from frequency dependence of PCT.

- CNFETt;urrentconsistent with DC measurements and assumed to be frequency independent

>- f-3db(VdirJ indicative of frequency when cross-talk begins to dominate

II\. . +IEEE Slide 19 -"

Measured CNFET frequency respOnse

CI

a ~

CI

a CI

t'J

a a t:I

tI (:I

r:I

E:I (:J

I:J g

a c: c: c a c c c: t:t

10 c a

Page 21: 2006 IEEE Radio and Wireless Symposium

iiJ­

I

I

I requencyI

I

I

I

I

I

I

I )

I

I

I

I

I

I

I

I

I

I

I

I

~ The -3db frequency of Vdifl indicates the frequency at which the cross talk

signal begins to contribute significantly to the total frequency

»BedUCin~e parasitic capacitancelincreasing the current increase.s f-3db

IV\. Ul +IEEE Slide 21 M'fT$.oI' ~~

I

I

I

I

I

I ParaJlelFETsI - Increased drive current - decrease Cparasitic penalty?

I

"." " ,:".' .

Arrays of eNs -Increaseddrive current - decrease CparaSitic penalty?

11

Page 22: 2006 IEEE Radio and Wireless Symposium

Radio Summary .

);>Time Domain measurements limited by system BW to -100KHz . . .

~ Developed novel frequency domain approach for direct measurement of the CNFET response at high frequencies .. - Technique uses large signal and higher sensitivity of spectrum analyzers

compared to network analyzers. - Using this approach CNFETs frequency response measured up-to 200 MHz - Ultimately limited by ability to extract signal from cross-talk

);> Method can be extended well into GHz frequency regime using - CNFETs with higher drive currents - Reduced gate-drain (for CS) and source-drain (for CG) capacitance

"IEEE Slide ~

Ie:

IC:

IC:

II:::

I::

IC:

I:: 12 a::::

Ie:

Page 23: 2006 IEEE Radio and Wireless Symposium

•••••"Single-Walled Carbon Nanotube High-Frequency Mixers"••• Sami Rosenblatt• Cornell University ••••••••••••••••••••••

II

Page 24: 2006 IEEE Radio and Wireless Symposium

••••••••••••••••••••••••••••••••

. . iii Carbonnanotube FET fabricationlcharacterization

iii MixerjRectifierexperimental technique·

iii Modeling nanotube mixing

iii High-frequency mixing

. iii Conclusions

+IEEE .. Slid. 3

Fabrication - Patterned ·Growth CVD with methane 10 min 9300 Ccatalyst Pad·

~

2

Page 25: 2006 IEEE Radio and Wireless Symposium

/

Standard Characterization

P-type FET

P FETe: (. I)I=-L- VbgO-Vbg+2"V,d V'd

G= IIV'd

BG L PFET="" BV Z

bg g

I ~

...~ ..

Local gate needed for high-frequency probing

·Outline·

.IEEE SlideS

·.Carbon nanotube FET . .fabrication/characterization

., :."", ',_: '"'..,. . ",",.":" ... - ,". -".";

~ .. ' ". '. . .'

.lmMixer/Rectifierexperimental technique

1 I!

,IIi 3II II

til

Page 26: 2006 IEEE Radio and Wireless Symposium

•a

• I

I

I

••• I

I

•• I

••••••••••••••• I

I

I

·.High..Frequency Techniques, . • Microwave Impedance Measurements

.,.......,---..---.--,--,

. .

S. D. Uet aI., Nano Letters, 4, 753 (2004) ... . X.Huoetal., IEEE IEDM Tech. Digest, 691 {2004) Z, Yuetal.;·NanoLetters, 5, 1403 (2005) ..

• Mixing

o ..4 JOt'!t '.V.\VJ

J.Appenzeller et: aI., V. Sazonovaet al., Appl. Phys. Lett., 84, 1771 (2004) Nature, 431, 284(2004) -----------. • IEEE Slide 7

S. Rosenblatt et al., APL, 87,153111 (2005) • Top-gate geometry/local gate .Insulatings~bstrate··. ... ..

. 810 2

. .. . Gate···· ·NT

.y..i-..

Mixing·at 'SO GHz with·a NanotUbe·.

4

Page 27: 2006 IEEE Radio and Wireless Symposium

~ :l

H'

i)l.t the peak

Experimental Results

V 'I

1000 .......­ __...............,-....:,

~ '00HHZ 10HHz

__ 100 ~~:G::: :a-l! 10 ~:~::: H" 1

o1 . Refemnce

Id:

c: c: c: c: c::f=10MHz

Vs...nns =400 mV p• Zmlx versus Yg agrees qualitatively with aGdJa~

. • Over 60 devices tested • Peaks within factor 1.5 below model ~ • Zm;)( agrees qualitatively with (~a')2

(Reference curve) • Exponent in range 1.9-2.2

r-"--------------., • Zmlx qualitatively explained by model • Amplitude is frequency-dependent

"IEEE Slide 9

.... ..,1,·... · , . ",,10<...

"::B:Il:!qUeI1cy,;t;;H z)

SimPlestimOdel:lo;'-p~~filter ',... . ' .......

'... \"t..ti(al)/2jyj~_,::,i .." ......•: , " 1;,,;(~-:t.()n~/':,;l+{aJr)2: .'.> ,';"

, "~ ",

'0' .~~~~~r1J:r~~~~i:··.··"

Frequency Dependence

. . .. .

,V.~nns.;;400mV . s. . ..... ,... ,._..;.: .. ",: ;- ".:. .'.... j>'~:- .. "1' i ~\' ;.~,,:;., .

" '<. )~<~>

o -3.0

50 N'40:::I:

~.30 >0' .

. U .. ,;.20 '::3

ito... u.

I:

II:

It: II:

I:

I:

I:

I::

I: 5 I:

1:,

Page 28: 2006 IEEE Radio and Wireless Symposium

••

•:I

• ~

• ~

•I

• •I

••I

I

I

I

I

I

I

I

I

I

I

I

I

Experimental Cutoff Frequency Mixing peak. several devices

30.......,1""""""_---.----..........,4

:~ 3 :/~.R_C. . 40 ao t.···

(i)'" - •• /i~ 2a! .J. - (!) 1 .',.. 20 'T 'r -.:; 10 ~ • / •

10·-----ilIIt 1 :l ... o........2L... ---....;.O:lll.S- 0 ...... 200=-----'~400:'-:--"""'"--~600::----'---.""""1.-S-.1..L.O o~' ----'--.......O

V (V)g

~peak in 5-30 ps range while Rpeak in 5kn - 1 Mil range

"IEEE Slidell

. . .

• Carbon nanotubeFET fabricationlcharacterizaticm. <.." .... "

lIII Mixer/Rectifier experimental technique '.

."

••..•.• ,High..freq~.~nc¥m.r)(fnb

t;~ti_c'"''''''¢'''''

•. Modeling nanotube'h1ixing

•.•.

I

I 6I

I

Page 29: 2006 IEEE Radio and Wireless Symposium

i

I

I

a IEEe, Radio

,liS Mixing with a ,Lossy Transmission Line .I(z, t) pIiz 1(z+Az, t)

~~+' ••• V(z, t)w w~...LC'AzV(z+Az,t) ...

V(O) V(L). _. T ',._ '" I I

it • I I

• Distributed nature of NT Az

~1~- Very lossy cable I IT

Rs:lin :RdI. I

: Zoo, r :• AC voltage profile of Zin~lI;n :Transmission Line:

I I

I I

• • •z o L

"" .. "IEEE Slide13 ~

a I

I

II

C

C

C

C

C

C

C

C

C

'Mixing withaNon~inear Transmission Line C

~

C

C

C

C

C

~

~

Ii:l

~

Ie

Ie 7 E

I:

Page 30: 2006 IEEE Radio and Wireless Symposium

•••

• I

I

••••••

••• I I

I Experiment versus Model I

boundI

I I

I

I

I

I

I

I

I

I 8I

I

+IEEEslide 15

Depends on combination of R, R$f Rd

Lower bound given by Rs

Cutoff analysis . complicated

Distributed Model -Results .

10·'

10·· I---r-. J 10··L~-

10·· •

10·" JR!. 10 -20 dBld~cad',I 10·' '.° 10' ~ ct 10' 10' 10' _

~-.:.,~- .

Include contacts Rsourr:e Rdrain RNT =R

.. .

Since Rs <RtDtal

. Either Rs or setup could be .. responsible fOr.cutoff

·~JVotdJ;'c1usjve• .. '..

•...... .. . 20

Page 31: 2006 IEEE Radio and Wireless Symposium

Outline

:.'. Carbon nanotube FET fabrication/characterization

II Mixer/Rectifier experimental technique.

.• Modeling nanotubemixing

II High-frequency mixing

• Conclusions

1V\ .IEEE.. SUde17· oms" ~

High~Frequency Behavior

C

E:

I:

IC

I::

E:

I:

I::

I:

I:: 9 E:

C

Page 32: 2006 IEEE Radio and Wireless Symposium

••

::I

l:I

::I ::I

:I

:I

:I

~

~

I

I

• I

I

I

I

I

• I

I

I

• I

I

I

I

I

I

I

I

Schottky contacts mix too, specially in ON state v;n = Vsd Vout

Conductive NT, Schottky diodes:

II mix < 0 at high frequency!!! See also S. Heinze et aI., PRL 89, 106801 (2002)

11":. ~

<.o-;:;;~~~~~:"04"

.i'·

f'!1.=;,·~';;:;!h":Y \:JH::":.J

---------.-<- .IEEE Slide 19

Mixing with carbon nanotube: Measured mixing up to 50 GHz Cutoff frequency 1-10 GHz Possible sources of cutoff: • Setup • Device + contacts • Defects 'min ~ [2nRsCgP ? Found Schottky mixing in on-state

.IEEE Slide20

10

Page 33: 2006 IEEE Radio and Wireless Symposium

tt Ct

c: 1:+

c:looking Ahead-----_.

0.5 1.0 zIL Drain

-D.8

-_._----­, eJ: Mixing model predicts DC build-up g: Use Scanning Probe Microscopy to verifx,A g:(EFM) ~ ~

de .; .... :; zy 3'? FaJw)= dz V"pVa,,(w) > g

i See A. Bachtold et aI., PRL 84, 6082 (2000) a

Investigate LClimit in ballistic I\JT c ---+ Plasmons c Use Schottky mixing as a tool c

I V plasmol! = V F / g , 0 < g < 1 C IV\. mi .IEEE Slide 21

M'f't$ -t,,: <••""~~>"""" 1:1

C

1:1

C

I:

I:

r: II:

E:

I:

I:

I:

I:

It

~ Paul McEuen (PI) I!I!l Vera sazonova

II sandip Tiwari (EE) II Hao Lin

II Jon Shu

·---------~..•IEEESlide22

C

C

11 Ie

C

Page 34: 2006 IEEE Radio and Wireless Symposium

02 L... ­ ~

10

1

I

I

I

I

I

I

I

I

I

I

I

I

I

I

I

~

"

H'

Dol 1 10

Frequency GHz)

Sami Rosenblatt

------ .IEEE Slide 1

I

I

I

I

I

I

I

I

I

I

I

I

I

I

I

I

I

III Carbon nanotube has distributed R, L, C C'f;::X:::;:V,

R L

III At high frequencies, inductance L becomes important

II For negligible resistivity ---> LClimit Plasmons - Collective low-energy excitations

111 First AC conductance resonance at ! = vp""~. /41 For a review, P. J. Burke, IEEE Trans. NanotechnoJ. 3, 331 (2004)

If I '" 10 11m,!'" 100 GHz

Goal is to study RClimit first: -+ lowerIt room temperature

.IEEE Slide 2, ,

1

Page 35: 2006 IEEE Radio and Wireless Symposium

;I

I

I

~

~

m II

II

II

:I

II

"HF characterization of CNTFET"

Henri Happy CEA Saday - Commissariat it !'Energie Atomique

m II

II

I

II

II

II

II

:I

II

II

II

• I

I

I

I

I

I

I

I

Page 36: 2006 IEEE Radio and Wireless Symposium

I

I

I

I

I

I

I

I

I

I

I

I

I

I

I

~

IHE RadiO ar'ld\Nire4ess

~~~~~i,~~ HF characterization of CNTFET:c;,,!~ ~'); ",~,

Henri HAPPY'

J.-M. Bethoux', A. Siligaris", G. Dambrine', 1. Borghettib, V. Deryckeb, J.P. Bourgoinb

"IEMN - Institut d'Electronique, de Microelectronique et de Nanotechnologie

Avenue Poincare B.P. 69 - 59652 Villeneuve d'Ascq - France

W'Nw.iemn.univ-lillel.fi·

bLaboratoire d'Electronique Moleculaire, CEA Saclay- SPEC

9119 I Gif-sur-Yvette Cedex, France

U.M.R. C.N.R.S. SS20 f'ofltflltion Trnns/ert

Slide 1 .IEEE

IEEE Radio and Wire4e$s .

~,v~~I,~.~ Biography\..>:,.".~:t"J<>.{.'~ -=;"".._;;.",_:- __

Henri Happy received the Ph.D. degree in Electrical Engineering from the University of Lille, France, in 1992.

In 1988 he joined the Institut d'Electronique, de Microelectronique et de Nanotechnologie (IEMN), University of Lille, He is currently Full Professor of Electronics with the University of Lille. His first research interests are concerned high electron-mobility transistor (HEMT) modeling, using a quasi-two-dimensional approach. He is the main co-author of the software HELENA (Hemt ELEctrical properties and Noise Analysis - ISBN 0-89006-772-4) published by Artech House since 1995. From 1998 to 2003, his research areas involved with the design, fabrication and characterization (up to 220 GHz) of monolithic microwave integrated circuits (MMICs) for optical communications systems using either planar or three-dimensional circuit topologies. His current research area is concerned with fabrication. and HF characterization of nanometer

devices.

.IEEE Slide 2

1

Page 37: 2006 IEEE Radio and Wireless Symposium

IEEE RadiO aJ'id'Wh1Sess Sympo'!iium".,". ,,~' :,1-: ;/;;!¢

~"~ D;W;,~ "'..<-. Abstract A recent study of carbon nanotubes field effect transistors (CNTFETs) predicts that they

may be faster than conventional FET devices. Their physical properties make them promising candidate for high frequency (HF) operation, up to THz range. Besides, their nanometer scale is very attractive for future electronics. So, HF measurements are necessary not only to validate these performance, but also to deduced small signal equivalent circuit, which will be useful to circuit designers.

HF measurements of nano-devices is a challenge, because their small size results in the high contact resistance values, and they only generate a few microamperes of current. Its results in a very low AC level and a poor accuracy of scattering parameters. To overcome these problems, the parallel CNTFET structure seems to be a good approach. In this presentation, we report measurements on parallel CNTFET obtained with conventional S parameters network analyser, up tothe GHz range. From these measurements, we focus on the intrinsic HF potentialities ofCNTFETs by extracting an electrical modeling.

------ .IEEE Slide 3

II General context I1fiI HF measurements requirements

II Scattering parameters of CNTFET

mExtraction of small signal equivalent circuit II Conclusion

+IEEEs,ickl4

K;:

I:l=

IIC:

I:t

r.t

It

Itt

It Ie:

It:: II:

I: It;

It:

It

I:

c: I:

It

Ie

E

I: 2 ;

;

I

Page 38: 2006 IEEE Radio and Wireless Symposium

;a

I

I

I

I

I

I

I

I

:I

I

I

I

I

I

I

I

IEEE Il.~dto and\Vil~ Symposium

," ~ 1;- ;,'" ~",,,~,

. :':."r';·;;;:~" c~ General context----_.--_._-----,-,-- --_._._-­

Challenge

III Carbon nanotube: a promise candidate for high frequency operation?

Objective: CNT / CNTFET

!I!IJ Promise candidate for HF

Wl Experimental investigations in the RF/HF range

IiIl Approach: Small signal equivalent circuit

+-IEEE Slide 5

I

I

I

I

I

I

I

I

I

I

I

I

I

I ,

IEEE R~d;o .ndWirel.,.Symposium ~;.- ,.; : ';'~ ~!.;,.,/.

: ~'>'''' !~:~";(-'. {'/' General context .Small signal equivalent circuit interest

liB Useful: • to understand and study the influence of

physical/technological parameters • To understand HF behaviour

III Will be necessary for nano circuits design in HF range

High frequency measurements are necessary

.IEEE Slide 6

3

Page 39: 2006 IEEE Radio and Wireless Symposium

IHERadio and Wireless

~r,i~~~~;:',,,,,_, HF measure!f.en~~ire~nen~ ~ Carbon nanotude FET - Problematic of HF measurements

Ii Low driving current III High value of contact resistance I!!I HF facilities : Zc = 50 n -+ Sensitivity / accuracy of HF measurements?

Overview of existing HF measurements methods

Indirect measurements Direct measurements Extrapolation of HF Vectorial Network Analysis characteristics of Spectrum analyzer carbon nanotube Temporal pulse response ...

Slide 7 • IEEE

IEEER"dio and Wire1esi Symposium ~ : .. ::) ; ~I': ~~::>i':.

,,';....,~ ~"'*,;;('.. fA HF measurements - Overview CNTFET in transmission Carbon nanotube as resonator

12GHzCryogenics temperature - 2.6 GHz

X.Huoandal. IEDM2004

.IEEE 'Sllde8·.'

.. S. Liand at Nano Letters 2004

4

Page 40: 2006 IEEE Radio and Wireless Symposium

I

I

I

I

I

I

I

I

I

I

I

I

I

I

I

I

I

Time domain response of CNTFET - HF operation up to 100 MHz

~ ~ n_-__~:~~'~,~:;'~'~

;::" ~ 0 V~ ... ,1V.1

}~ ! ~ "'" ,. ~" j ti '''' '. 1o "r .. .,'. '

10010: 1M HlM 100M

F""l"<ncy (Hz)

D.V. Singh and aJ. IEEE Trans. On NanD.

Vol. 3 - Sept. 2004

.IEEE Slide 91!:l ~~~;...

I

I

I

I

I

I

I

I

I

I

I

I

I

I

I

IHERadlO and Wirdess _ '"

:{;~,~iC;~_ H~easurem~n~requlreme~__ Proposed solution

mt! Fabrication of a multi-channel (parallel) device structure

• Increase driving current

• Reduction of access resistance Carbon nanotubes self assembly

deposition S. Auvray and aI., NanoLetters, Vol. 5 -March 2005

.IEEE stide10

5

Page 41: 2006 IEEE Radio and Wireless Symposium

Device cross-section

SourceDrain

Ion implantation for gate control

n++ Si Gate

High resistivity Si

Reduction of transmission line loss access

.:.::F:. .IEEE Slide 11

u~:ri'~~,,;......,

lilt:

Itt: IEEE l\ddio ~nd Wireless Sym~ill,::!

E

I:

I:

Et

I:t

m Influence of

..... Metallic carbone m nanotube

m Itt

+IEEE Slide 12

11:!

Multichannel FET

P-type transistor 130 ,.....,.......,..T'""-.-..............,......,-.-........,

120

110

80

70 .......-.-............--..-,r-r"..........,.--.......:I -8 -4 0 4 8

Vas(Y)

I:C

c: 6 It

C

C

Page 42: 2006 IEEE Radio and Wireless Symposium

• •

I

I

I

I

I

I

I

I

I

I

I

I

I

I

I

I

I

IEEE Radio and\<Vrl'ef€:ss.

~t~~'\IJ;~:J:Jf C~,~ra$riza~ion Q.roced4re , "__~__'" Multichannel FET

LRM calibration procedure

Scattering parameters of CNTFET structure - Agilent 8753 Network Analyser

Low frequency analysis

Lower AC signal level -o.08+-......,...._.--...-.............,_-l ·120+-......,......................._ ....._-4

o 10 0 2 10

frequency (MHz) frequency (MH%)

VGS = -8V VDS = 1.5V .IEEE Slide 13

I

I

I

I

I

I

I

I

I

I

I

I

I

I

I

Active properties of CNT FET

[yactive ]= [ymeas]_ [yopen]Extraction procedure

Structure with Identical passive structure nanotubes without nanotubes

100

=

150-r-......"'T'"_-r-.............,..-........... Cii V =-BV ~.=;, 10 , . " "S'¥~d.:..t;:,~,,·.i<{-:."Pt..~~ilJ.~ " = ~-i';.co"1o • ~.. , ...JI;',,'t ... ~.-..-.y;."tIJI'i'o. ~ : =

ACTIVE DEVICE VGO'" 0 V ' .

•.~... ~,t

P:\SSIVE ~OPEN" STRUCTURE= o~....fI'i'll~~liIlfIi~~~ O+-.......,..---.._-.-....--r---I

o 2 4 6 8 10 0 2 ·4 6 8 10 frequency (MHz) frequency (MHz)= 1.5V

Extrinsic Transconductance

VDS

No frequency ·Extrinsic Conductance . dependance

.IEEE slide14

7

Page 43: 2006 IEEE Radio and Wireless Symposium

C It:

\

E::

I:::

IEEE Radio andWil'd= Sympo<;um ""'!¢-.'.".:;.-;;;f HF C~aracterilation ~r~£~l.!~._~_.P._'_

Active properties of CNT FET

Positive maximum stable gain Frequency limitation close to I GHz of extrinsic device

10

20

ACTIVE DEVICE •~_ 10 :<...~.~;.:.:;:./-.:::,.\~:::.~.:/:,.,:~.;-" .....

.' .....

~ 5 ,

II " = ~ o+-~.........~..........................................m.t

0,2 0,4 0,6 0,8 1,010' 10' 10' 10' 10' frequency (Hz)

=-8V =1.5VVGS VDsE "IEEE Slide 15 <.;.:.._ ......_<, ... :"')O.~.

frequency (GHz)

E::

a:c a:: a:: I:

E:

c: c: Ie:

Et

It

Proposed small signal equivalent circuit of CNTFET

Related to the device structure

Gate

CNT

""I High resistivity $1 ~I ~~--

..iV\ .IEEE Slide 16

IlIt

m: E:

1=

Et

E:

C

m II:

m I:

m Itt

m m

8 E:

c

R g

_..

Page 44: 2006 IEEE Radio and Wireless Symposium

l

I

I

I

I

I

I

I

I

I

I

I

I

I

I

I

I

Rg is deduced from DC measurement Gale f\ Cw;! I ~......o-4-'MAl-t---! . =;U" E::~~:~C

c.,.

r ............ - ......... - .... _-_ .. __ ...., Expressions of small signal circuit VG*! -: Cd:

,..-----,r-I '

C~ R5L • ~e"-ic..e_J

Assumption made:

Drain elements are deduced from extrinsic [Y] I

parametersi,, ~: ,,

J - Source and drain contact are symmetric - They are considered as fitting parameters

Extracted parameters (Vgs= -8 V; Vd= 1.5 V)

R.~442n R,=Rd= 6.8 kn R;g,= 15 kO Cs = 10 IF ~gd= 300kO Cd = 10 IF Cpg,= 1.9 pF gm= -110 ~S

Cpgd = 50 IF &1= 800 I-lS

"IEEE Slide 17

I

I

I

I

I

I

I

I

IEEE Radio .and Wireless Sym;>osiurn ~:-. ';.:::,~ ~.':i.>:,_ Extraction procedure, :>".:'~'""':i!~·,C:-.

Intrinsic extracted parameters

0-r---r-....-.,...........,...........,.-......., 1500

1000 Iii . ;... 'C;:- .. II" • Iii.=, -100 .=,~ ..y:.~~:cf or.. ". ..' ,._. >II ':' 500

-150

.-200+---r-........,...........,..--"T".....-t 0

o 2 4 6 8 10 0 2 4 6 8 10

frequency (MHz) frequency (MHz)

.IEEE .Slide 18

9

Page 45: 2006 IEEE Radio and Wireless Symposium

Comparison between calculated [Y] parameters and measured [yactive]

i ·10

>N ~ -11

~ -12 ....L. -"'_o...-..L­ ...L_.....I

~ -0,5

parameters 0,0 ".

~ lil (\,1·1,0 >~

~ -1,5

0,0 __..- _...,O,O.",..~"""T-.~.---r....,.--,r- .......,

2

'$ 2; -0,5

i ~1,O

~'1,5 I! ... -2,0

o

~-o,5

4 6 8 10

frequency (Mtz)

St,;})~~\(~;~\~;b:? Slide 19 <4)IEEE

AC properties ofCNTFET arc investigated through direct measurement

III

B

III

The active properties ofdevices are demonstrated up to the GHZ range

The small signal equivalent circuit ofCNTFET is proposed

From this small signal equivalent circuit, device structure will be optimized

III The parallel device structure is an efficient method to probe HF properties of CNTFET via standard VNA measurement

.• IEEE SlIde20

10

Page 46: 2006 IEEE Radio and Wireless Symposium

• I

I

I

I

I "Design Considerations for High Frequency

I· "on-wafer" Measurements"

I

I

I Glenn H. Martin

Universal Filmworks I

I

I

I

•• I

I

I

I

I

I

I

I

I

I

I

I

I

I

I

I

Page 47: 2006 IEEE Radio and Wireless Symposium

I

I

I

I

I

I

I

I

I

I

Ii

I;

It

I

I

I

I

I

I

I

I

I

I

I

I

I. I

I I,

••I

DesignCon,~iderations for High Frequency "on-wafer" '

,measurements

Dr. Glenn Martin Universal Flimworks

CEO@universalfilmworks;com

__ R. rw, ~ +IEEESllde1

M't'1$~ ~

... Agenda

• On-Wafer Measurements . _ What... Why... How Tricks... ? '_Implementation... . .

• EXamples-'" . _ 2 dimensional devices

.' II fET's, Planar Diodes, etc...

.,"," ,"." .".

<1ft:;""""'".;: h'"

-: ~:.._.......:,","

.

. ..... ·~~':""'~H;;io-tyPes. .•. ..•. . ill Embedcfed:Impedance:Nismatches' ".

'..'.• Techniques} •...·•· ':~~?n1Ple~' .j.? '.' .

'/;.\' '". " ~ ;·;i~;·.~-.,' " ~.<:':::". ; . :

1

Page 48: 2006 IEEE Radio and Wireless Symposium

tl: c: c;:

c:

What... On-Wafer Measurements

.Accurate incorporation . of "adapter:like" . . .

interfaces into an error model for direct representation of

device characteristics

------- .IEEE Slide 3

g:

R: g;

a;

~

~

c: c:

I:

r:

Why...· On-Wafer ~easurements

I:

It

It

m:

m:

I:

E

E

E

J:

I:

I:

E

t £

2 r t

C

Page 49: 2006 IEEE Radio and Wireless Symposium

••

~

• I

I

:­I

I

I

I '

I

I

I

I

Accurate.' and repeatable .. fabrication .and characterization of

interfaces to the modem macroscopic device

environment

I

I

I

I

I

I

I

I

I

I

E$,..~ • .' . . ~;~~. TI1ck... On-Wafer Measurements

Innovativemini111iz~tionof . .the11hysical reflection aiM .. .. transmissions losses for the .'

···~!:~:~~rt~t~!~evlcec ' . ~- "x - - ""':

,.

.~: '.. -. ,; " . ':" -- ..,. - f; _• ";"'--L ~ -- - ~, '.

.',., .... :" ;ft'!); . ,..,.' .. ,,:" " ~)i;:~/"Yi::,j~'fj>

'~r'" .','.:, ~. ':.:/·,:;·J):·~l;~\;,·~:;:

Jfi,:,.1.;:u>·....,.:,<~~':...-.-:·.;~. ~~;.. -~.-,......,.,: .......,.c., ..

3

Page 50: 2006 IEEE Radio and Wireless Symposium

I

c­c c: c: c: c C

II::

II::

II::

I:

C

4

~;~_~:!e:j~U2 ..... .,'C ":<'/:'~:-:.Lj::",,;:, "":':'.,:./'

": .. ',.'"..

,~,

Implementation

II Enabling On-Wafer Technologies • Thin-film .fine~lineprocessing

• Micro~Electro~Mechanical. Systems ~ Mathematical Modeling • hmovation . • Future Technology...

II Current Technologies have allowed our existing micro-electronics industry .

fi Direct application to Nano-Devices needs refinement

+IEEE Slide 7

IEEER4dio . W~ .

'9~~ On-Wafer RF-Measurement Basics~r.~;t!'1¢,Q. .". ,.

• System Calibrations II Set ·ofStandards to remove all repeatable re1~ectiqn!?/transmissions from adapters' .

IIFrom2~portfheory (error matrix requiring "aserj~sofrepeatable measurements' .

Page 51: 2006 IEEE Radio and Wireless Symposium

!

iii

Ii

11 Ii

1 I;

II

11 Ii

II

I

I

I

I

I

I

I

I

I

I

I I

Typical2-port Device layout

cn;-Wafer On-Wafer Probe Probe

.IEEE Slide 9

Typical Calibration Layouts Open _ .. _< ... . Line

5

Page 52: 2006 IEEE Radio and Wireless Symposium

C

t:t

R R

IEEERa<Jio

~a Basic 2-terminal On-wafer RF layout~~,tt;:~;< ... __.-;.., .-;.., _

.IEEE Slide 11

~

R R

R E:I

E:I

E:I

lC

C

C

C

t:l

On-wafer RFcaUbrationStandards

Rf' Calibration

C

J:

I:

I:

E:

C

C

Il:

C

Ii:::

II:

I::

I:

6 Ie

I:

C

Page 53: 2006 IEEE Radio and Wireless Symposium

• I I

I

I

I

•••• I

•• I

I

I

I

I

I

I

I

I

IEEE Radio . .

~i~ Matherrtatical Modeling of RF Currents

+IEEESlide 13

.. ....

. .

.. ... . . .." ,~~: '::.'; . . . . ....

7

Page 54: 2006 IEEE Radio and Wireless Symposium

I

iC:

c: c: r::

Self-Assembly

AClUALsur·ASSEMDl£OMONOlAYER

SOUJTIOH

--. .I!.~C . "'" , .. Mn"$" ~.~<ot"'"

a::: 1=

1==

~

r.:: m:::: c:: c:: c:: rc:: I::

Il:::

It::

·Molecular Memory It:

m:: g:.:

II:

c: c: c: c:1 c: c: c: II::

8 II::

C

C

Page 55: 2006 IEEE Radio and Wireless Symposium

Ii

I

I

I

I

I

I

I

I

I

I

I

I

I

I

I

I

I

I

I

I

I

I

I I

I

I

I

I

I

I

I

DCCharacterisitics (RTD)

100

80

60

40

1 20

1: 0CD to u" -20

40

-80

·100

+IEEESlide 17

Molecular Device (Organic)

9

Page 56: 2006 IEEE Radio and Wireless Symposium

__

... Organic,Molecule Evaporated Metal (shadow Mask)

+IEEE Slide19

IEEElWlio

~._._S_...:..,.pa_ra_m_e_te_rs_o_f_M_o_n_o_n_itrodith_o_l_

10

Page 57: 2006 IEEE Radio and Wireless Symposium

I

I

I

I

I

I

I

I

I

I

I

Example topology that gives similar

characteristics

,"""""'.­ ·-..S{1,21 *' srj."1t< ~

---­

"IEEE Slide 21

Problems with Metal-Qrganic-Metal Devies

<> Problem: • Lack of (reproducible) data ~

t Gold pu~ Gold lIlIs gaps at holes in SAM grsln boundaries

I Gold tills gaps .

<> at edge

~~\Wjl)jf#hApproach: • Develop,a reproducible starting pointformoleculardevice .

fabrication . . , • Ideally allowing for high sample throughput .

• Confirm the presence of molecular components • Characterize the formation of self-assembled monolayers in the

testiQg'environments. . . '~(

..~: ~ ,.)"';f

-.

" • Preveh~'shorting when applying 'thetop'cOntaet .•.

.,< ,,~:>~

'.- ..

:' ",: >", '.~' ~.-.

.. ".-- ~ '-~---':>,.\- ,.. '

11

Page 58: 2006 IEEE Radio and Wireless Symposium

Who hasthe··magic key 1.

...~ ....•.... ,

.IEEE Slide 23

12

Page 59: 2006 IEEE Radio and Wireless Symposium

i

I

I

I

I

I

I

I

"Challenges in Measuring S-parameters of High Impedance States in Carbon Nanotube Structures"

I

I Loren Betts and Hassan Tanbakuchi

Agilent Technologies

Page 60: 2006 IEEE Radio and Wireless Symposium

I

I

I

I

I

I

I

I

I

I

I

I

I

I

I

I

I

I

I

I

I

I

I

I

II

II

II

II

II

II

Ii

I]

Challenges in Measuring S-parameters of High Impedance

States in carbon Nanotube Structures

Author: Loren Betts, Hassan Tanbakuchi - Agilent Technologies Author: Terry Burcham, Junko Nakaya - cascade Microtech

+IEEE Slide 1

Outline

•FormaI Definition of Impedance .Impedance Measurement Techniques

• Using voltage and current ratios(RF I-V) • Using ratios of power waves(S-parameters and network analysis) • Synthetic versus physical baluns for impedance transforms

.•Future High Impedance Measurement Research ...•CNTFET Measurement Systems

• Probing techniques • De-embedding parasitics • calibration considerations

1

Page 61: 2006 IEEE Radio and Wireless Symposium

IEEE Ra<lio andWill!les;;

,t"~:l~~~ D_e_fi_ln_it_io_n_o_f_I_m_ped::.....--.....a_n_ce _

Impedance is defined as the opposition a device places on an alternating current and is represented by a complex number associated with the real (resistive) and imaginary (reactive) . components of the system. .

Reactance can,be two forms: Capacitive and/or Inductive

.IEEE Slide 3

RectaUnear Coordinates.

Complex impedance can be represented using rectangular . coordinates where the horizontal axis corresponds to the real

portionand the vertical axis corresponds tothe imaginary .. ... .,portion. ..

.' '-. ~ '. ' .' ". : . . : .. ..

··How'Clo you represent anopenorashortimp~danCe·in rectalinear coordinates? It can be difficult.. .

2

Page 62: 2006 IEEE Radio and Wireless Symposium

I

I

I

I

I

I

I

I

I

I

I

I

I

I

I

I

I

PolarCoordinatesc(Smith Chart)

· To represent a vast range of possible impedance values can be challenging using rectangular coordinates. They are much better suited ·for polar coordinates and hence the birth of the Smith chart. .

The center of the Smith chart corresponds to the normalized system impedance of the measurement system.

A short is represented on the far left and an open on the far right. Circles of constant resistance and reactance are illustrated.

.IEEE SlideS

I

I

I ·

I · I

I

I

I

I

I

I

I

I

I

f

Outline

•Formal Definition of Impedance .Impedance Measurement Techniques

• Using voltage and current ratios(RF IN) • Using ratios of power waves(S-parameters and network analysis) • Synthetic versus physical impedance transformers

.Future High Impedance Measurement Research .' .CNTFET Measurement Systems

• Probing techniques • De-embedding parasitics • calibration considerations

3

Page 63: 2006 IEEE Radio and Wireless Symposium

IEEE Radlej

~~. Impedance Measurement Techniques

There are a number of different techniques:.

1; Bridge Method 2. Resonant Method 3. I-V Method 4. RF I-V Method 5. Network Analysis Method 6: Auto Balancing Bridge Method

For measurements from 100 MHzto 3 GHz. the RF I-V method .currently has the best measurement capability, and from 3 GHz·and

up the network analysis is the recommended technique.

+-IEEE Slide 7

3f1 Impedance Measurement Techniques' '.

.'. "" i..... ,~F ": . .Studied In this

presentation

II:

Ie:

E:

II::

I:

C

r: IC:

I::

E:

I::

I::

E: . I::

Ii:

I:

C

C

II:

t C

C 4 C

C

Page 64: 2006 IEEE Radio and Wireless Symposium

;e

I I'

I

I

I

I

I

I

I

I

I

I

I

I

I

I

I

I

I

I

I

I

I

I

I

I

I

I

I

I

I

RF I~VMethod RF I-V method Zx

The RF I-V measurement method uses an ­impedance matched measurement circuit (50 Q) and a precision coaxial test port for operation ;:It higher frequencies. There are _ two types of configurations which are suited to low impedance and high impedance measurements.

V 2fl z.~-~--

I ..:!L _,Impedance of the device under test (OUT) v, is derived from measured voltage and _ current values. The current that flows through the DUT is calculated from the voltage measurement across a known low value resistor, R. In practice, a low loss transformer is used in place of the low value resistor, R. The transformer limits the low end of the applicable frequency range.

High Impedance type Zx V 1lf - 1'1 (v,__ --1 ) t 2 V,

.IEEE Slide 9

RF I..;VMethod The RF I-V method can be done in either a! series or shunt configuration. The series configuration is better suited -for measuring high impedances wher~astheshuntisbetterfor low impedances.

.. t..v~

I =-~ x R­

- (-) _ 2,

--v, .c,y• .;.;vz ;'0 _­:.~ :

I =. -V2 + VI , R R

,' .... : ': •~> .•' • • •

, -//" ..":',1-:-,"."':'.:.'.",-~':"'"7'~~~J

.!~3~\--

5

Page 65: 2006 IEEE Radio and Wireless Symposium

lIl::l Ie

m: m:

HighZlvpe

V2 R 25 25 nVI ~ ~2Z;-;·Ft·,:"Z~'~~25"

Now: Zr. is tht) rolt..fJft" Iml~lln("(lo

'whero tile voltage mUo for tt\e rosie­

·20H~~h~

V2N,~J.? f-!"~niL-+~+~F'9 V,N2 ~ (dB} -40 ~.-.",j'--+~+-+M,-"i

·EO 71::t=::1tIlu.ooL= D.'

The RF I~V method can be configured to measure changes in high impedance states using the High Z type of measurement.

10 100 1000 10000_""ca)

RF I-V Method

Network Analysis Method The .network analysis measurement method uses a single source with directional couplers andcornparesratios oUraveling waves froml1)e measuredS-parameters to calculate the device IlTlpedan~.. TherEl·are.two mainmethodsofmeasuring. ". impedancein this mode: ' .... ' . . .. . . '. . . .

. RefIectior":~o~e" .Throughmode

Series through ShunHhrough ..

.,·'f'·-.">'.:

Uve DUT deere...... b}' 6 dB and UIe ",ll8IIe ratio ro-r \be reacUve DUT de<:rcase.... by 3 dB {rum U1(~ refereu{:'(.~ Iewl (0 dB).

+IEEE Slide 11

1=

1=

1=

a: E:

E:

E:

I::

Ie:

C

C

E:

E:

I:

I:

I:

I:

r: C

C

C

C

C

C

C

C 6 C

C

Page 66: 2006 IEEE Radio and Wireless Symposium

I

I

I

I

I

I

20 11_

I a1 ---.: Ix

V11 b1 ..

I Reftection r

I

I

I

I

I

I

Network Analysis Method

I ~..!:i. I z,

.. s ~z,-z, ~ Z,-50 • II ZX+Zj> Z;I+SO

.IEEE Slide 13I

I

I

I

I

I

I

I

I

I

I

I

I

I

I

I

I

I

_12

tb2

$-param_ 811,821 Serles·through mode

Zo 11_

a1---.. f :v. -I,Z, I b1 V1!.,......__. !

IRe(Z,)I\ " ---:..__.-:...iV:..::2__

I, ~-/2' V, ~-I,Z., 11; ~/,(Z, +Z,)

~~= .~~~--," __N_.e_tw_o_rk_"_A_na_I.:.,.VS_is_"_M_e_th_od _

"Series-Through Method

".. Y.+IZ V.-IZ " bydefimllon :a, =~R~Z:)I' ~ =~~z:~ ,b,

7

Page 67: 2006 IEEE Radio and Wireless Symposium

I

I

I

I

Shunt-Through Method Zo

S =!2=V1 -/2Z2 Z =z =Z =500 11 a, V, + I,Z," 1.

I, =_,,_,_~ V,(Z, + Z.) Z,Z. Z,Z.

Z, +Z.

Z, -Z, -Z. Z,,+Z:x.+ZIJ

___ _.._ 'j . -Z -25

:.SII =-_.-=-­[ 22,+Z. Z,+25.~•••••••w~ ••• ••••• _ ••••_ ••••_ ••_ ••••••_ ••••••, ••••__._•••••••••_~

Network Analysis Method 11 __ 12

al--4· Zob1..-!"

Soperameter 811.821 Shunt4llrough mode

+IEEESIlde 15

I

I

I

I

I

I

I

I

I

I

I

$-parameter $11,821 Serie.·th""'llb mode

Network Analysis Method r, ~~ .~:;: I

8

I

Page 68: 2006 IEEE Radio and Wireless Symposium

. ;Q15.4xlQ-<i,:<··,· >~::~-". ; ~-- "

<~~\::.-'>~';: . :;; .,::.....

I

I

I

I

I

I

I

I

I

I

I

I

I

I

I

I

I

Network Analysis Method The instrument must be able to accurately distinguish even small differences in the impedance characteristics which can btl challenging using a network analyzer in a reflection measurement configuration over a broad impedance measurement range such as required for the current nano structures. . Z -Z

0 •P =_'__ 2 0

is typicallySO ohms in a network analyzer(defmed by hardware/software)Z,+Zo .

. Z-50 p=-'-­

Z,+50

If the system impeda within a specific rang desired impedance, t analyzer must be ab differentiate accurate small changes in a la amplitude (reflection) difficult and so limits impedance measure accuracy.

19\11"'"" ., --m-n-

1

0.8 ...'. , ..... ... , ... ': .'. , '/" •..,.. oj

'"' 0.6 ,;. i; "

nce is not ~ 0.4.. -'I ,_., :',;, ;;,.:" e of the .~ 0.2 hen the

.. : ',I8 0 : ...

" .Ie to " Iy between

iJ ·0.2 ..... ,., :1, ....;.;.,"" ....: .I!! -0.4 I ..;

rge '" / .' ;

'" -0.6 I ..... : ..... , :'+ . oj.... , '., . This can be -0.8 .. , . I"~

the ;/

-1 ment range 0.1 1 10 100 1000 10000

l.."edance (!l)

y .._-- Slide 17

I

I

I

I

I

I

I

I

I

I

I

I

I

I

I

'1:"' Network Analysis Reflection Method Most modern network analyzers digitize the received signal. The digitizers have a specified bit resolution that determines the minimum amount of signal change detectable. Forexample,'a typical digitizer used today may provide 14 tits of resolution (max signal to min signal).

. , . . . ..

". '1'·· n(max-min)';W-OV! ·O·'~·Vltlb·Example: .' resoutwn=u- .. ,... 14 =15.4xl ..0 It . . # of bIts,. 2 ," .' .14 bitADC 50 X 103-50' .Min signal level :: 0 V [I = , .998001998002...

MaxsignaJ level:: 1 V . 50xlO' +50. .

.' 'Impe~~nce::1: ~. -.59;kohm . . [,lQxlO: -50=;990049751244•. ~ .;'. • . .lmpedance:'2;=10kphm; . . lOx 10 +50 " ..

20==50ohm';";':: ;:; .,Ll=~II-Jr2t=7.95224f7SSxiO·r'

.~itd~fe~ce~i ~7.952:246758XiO·':·~S16~i~'·

9

Page 69: 2006 IEEE Radio and Wireless Symposium

I

.

Network Analysis Reflection Method . If the system impedance was now closer to the intrinsic device impedance then the bit difference improves and we are able to more clearly see changes between impedance states.

. .. . (max-min) IV -OV .Example: resolution =0 -1-'-'" 15.4 KIO-<; Vo1t1blt

Hof bits 214 bitADC 5Ox10' -30xlO'Min signal level =0 V . r, '" , , .25

Max signal level = 1 V 50x10 +30xlO

Impedance 1 =50 kohm r =10xlO'-30xlO' -.5 Impedance 2 =10 kohm 2 lOx 10' +30x 10' ..

Zo= 30 kohm A'" Ilr,l-lf,li '" .25

bitdifference",~", .25 -6 -16xlO'bits Q 15.4xlO

. When measuring small changes in a large amplitude signal we . are relying on the inherent resolution of the digitizers.

IV\.. .. "IEEE Slick! 19 M'I'I'$i!; ~~

.Network Analysis Reflection Method

This graphJliustrates the effective impedance measurementranges over . ···different methods. As can be seen the network analyzer reflection method has

a narrowiimpedancemeasurement window but can go much higher in . .. frequency than the current RF I~V method•..

1G 100 Hz

I·V (Probe) RFI-V

/ Network Analysis

r--fo-~~I--'\/

Auto Balancing Bridge

10 100 lK 10K

t: c: I::i

I::

E:

I::: I:

1= c: E::

a: I:

I:

C

C

I:

I:

C

E:

I:

I:

I:

I:

I:

£

I:

I:

I: I:

I: 10 I:

E

Page 70: 2006 IEEE Radio and Wireless Symposium

••••

••••••

••;Ii

••••

•I

• •I

••I

I

I,

I

I

1

I:

NetworkAnalysis Reflection Method - Physical . Transformers

Physical impedance transformers could be used at the devicelevelto change the effective system impedance but there are some issues:' .. . .

There Is no such thing as an ideal physical impedance transfo~. Should the calibration be performed before or after the transform:

. • If done before(Zo= 50 ohm) then the measuremems will contain. characteristics of the device and transformer.. .ideally we want just' the device. .

The transformer characteristics can be de-embedded from the measurement results so that the data contains only the device characteristics. This is only valid ifthe transformer .. ' characteristiescan be measured accurately.·. . .. '

Calibration after the transformer(Zo = xx kohm) requires high impedance traceable calibration standards which are difficult to find.

crrSD.·LlJ

..zo .. 5C ohm ZO =2C kOlHr

+IEEE Slide21

Network. Analysis Reflection Method .... SYnthetic Transformers . .

Mathematical impedance transformers could be embedded into the . measuremenlbut: •. , . . . . .....

The system hardware chara~teristicimpEldancestiUisthElfi~rici~erital . ..limit in the accuracy of the impedance measurement as was-preViously

discussed.· .' '.. ..

11

Page 71: 2006 IEEE Radio and Wireless Symposium

I:t

c: C;

E:;

Transistor Measurements·Using NA Through Method

If the device can be placed between the analyzer's test ports such that the intrinsic impedance parameter requested is configured appropriately in a through method a more accurate measurement can be made compared to the reflection method: . .

If the intrinsic impedance of the device can be connected in a series-through connection between ports 1 and 2 of the analyzer then 821 can be

The challenge: Can the intrinsic measured and therefore the . impedance of the device be · impedance can be measured.. isolated such that these If the intrinsic impedance of the techniques can be used.

device can be connected in a shunt-through connection between ports 1 and 2 of the analyzer then 811 can be measured and therefore the impedance can be measured.

1ft. ~ "IEEE Slide 23 t.ftT$.* ~~

~

c:: 1;:%

~

c: t:

I::::

E:

C

C

C

."...Future High Impedance Measurement Research .•CNTFET Measurement Systems

• Probing techniques

• De-embedding parasitics

• calibratiqn considerations

Outline

..Formal Definition of Impedance

..Impedance Measurement Techniques • Using voltage and current ratios(RF I-V) • Using ratios of power waves(S-parameters and network analysis)

• Synthetic versus physical impedance transformers

c: I:

c: E:

I:

c: I:

C

I:

C

I:

I:

C

I:

C 12 C

C

Page 72: 2006 IEEE Radio and Wireless Symposium

I

I

I I

I

I

I

I

I

I

I

I

I

I

I

I

I

..Formal Definition of Impedance

..Impedance Measurement Techniques • Using voltage and current ratios(RF I-V) • Using ratios of power waves(S-parameters and network analysis) • Synthetic versus physicalimpec!ance transformers

..Future High Impedance Measurement Research IliCNTFET Measurement Systems

• Probing techniques • ~mbedding parasitics • Calibration considerations

.IEEE S1ide25

I CNTFET Characterization System

I

I ..

Cables

Probes

" ,0. ,".' ; • .' ,. "

'»<;;'" - .'-.'C'_'_'

. . M'mie.'.':"

Vector Network Analyzer

Probe positioners

Probe"station·· .

Parameter Analyzer

;':CafitjtktI~nSubstrate{1SS) ~.,::-~:,;;~~j-~::.~.:~:~>n····' ... .~ "-, .

"'M~~;:"~'<">·",.·..,......;..;;,,;;..~~~~

13

Page 73: 2006 IEEE Radio and Wireless Symposium

Key Requirements for CNTFET Electrical .. Characterization

.IEEE S1lde27

P.;.TypeCNTFET

• Plot shows the electrical characteristics of a single-wall CNT P-FET. • Note that its characteristic curve obeys the square law jUst like a traditional silicon FET. • The transistor action is due to the electrical field around the gate modulating the SChottky barrier at the points where the source and drain terminals contact the nanotube.

14

Page 74: 2006 IEEE Radio and Wireless Symposium

•••

II

I

I

I

I

I

I

I

I

I

I

I

I

I

S<atl\'$l¢~c..o,. ~~....'~ :;a~ Agilent B1500A Nanotech Application Note

:..;::.:..,v;::;:g.:.:.g=;:.;...:;:..... _·A"j~ ·11~blt.

. • Publication number is 5989­.. 2842EN (also downloadable

from the Agilent external web site).

• Characterization of single-wall CNT FETs using the 81500A is explained.

III Agilent created this application note on CNT-FET evaluation using the 81500A in cooperation with Professor Matsumoto of Osaka University.

+-IEEE SlIde29

I

I

I

I

I

IEEE Radio .. . . andW"'" .

t~~'1Agilent B1500A Application Note Data

1

.:l. 8t 9K : 4 . 2! . ·a 06=", = 'f§o _

- 1__. ........-''T-'--k-'"'--;;-'-:l.1

15

Page 75: 2006 IEEE Radio and Wireless Symposium

c c: I:

I:

CNTFET Evaluation with Agilent B1500A

Shielding box

Slide 31.IEEE

TrlulalCollle

E:

r:: c: J::

.::

.:: c: t:

E:

I::

E:

It:

I:

• Dryf Frost Free environment

• AuxiliaryChucks To -Hat

Cascade MicroChamber· 1=

I:

E:

I:

I:

I:

I:

I:

I:

I:

I:

E 16 I:

I::

C

Page 76: 2006 IEEE Radio and Wireless Symposium

••••• ••

£lSM

:I

I

I

I

• I

I

I

I

PN1 07~088 Brackefrigidly mounted. to positioner.

External Bias Networks

---------- ­ . • IEEE slide33

Cables compatible with probes and top hat

I

I

I

I

I

I

I

I

I

,<ft

: .....: ..

• There are many proposed structures for CNT FETs. Do not be surprised to hear the terms "back-gate," "top-gate," "side-gate," etc.

• None are optimized for 5­parameters

. Proposed ·SChemes for Gate COntact

17

Page 77: 2006 IEEE Radio and Wireless Symposium

i!-~ft;'2..A~:;; 10 GHz S-par~!"1eters Demonstrated .

• Back-gated devices • GSG probes • No de-embedding

Figure Credit: Zhang et aI, High Frequency S-parameters of Back-gate cntFETs, IEEE, 2004

+IEEE S1lde35

I::: I;

c: c: c: £:

£:

.lnflnityTM Probe for Characterization E:

I:

Il:::

II::

~

Ie

Ie

Ie

c: t

I:

t 18 t

C­!

C

Page 78: 2006 IEEE Radio and Wireless Symposium

i

I

I

I

I

I

I

I

I

I

I

I

I

I

I

I

I

I

I

I

I

I

I

I

I

I

I

_Wi_IEEE Radio

~~~~_" __P_ro_be_5.....0:--m_ic_ro_n~pa:::d:::s::o:::r:::l:::ess::::::::::::~

II Probe tips are 12 x 12 microns

II Skate 20 to 30 microns

II Probe multiple times on a single pad in fresh metal

l1li Minimizes pad parasitics

.............tNilo..... ·.150' Sllp"'"",

.IEEE Slide 37

Crosstalk vs. Probe separation

• II Recommend >5Odb for high impedance S-parameters

• II Probes should be separated at least 200 microns

::

19

J 11

Page 79: 2006 IEEE Radio and Wireless Symposium

C

I:

c: c:

.. Rules-ot-Thumb for Reliable S"'parameters

.. Whenever possible use GSG • Use GSG above 26GHz

.. Probe pitch affects S-parameters • Use smallest practical pitch

• 1/50th Aof highest frequency for GS , 1/20th Aof highest frequency for GSG Df(""

+IEEE Sllde39·.

c: c: I:

I:

I:

c: c: t:'

c:

t: ~

t:

.Manual·vs. Automatic calibration

. WorstCase Accuracy to 40GHz .

.~ ~~~., "f+-­ .. ) ..~_ ~ , ll.l .~.

·.. ,Four Manual Calibrations

.. "'; ~ .. 115% Error - . . !10% spread . f

"t . 0:

. . . .' ".

Ten Semi-Auto Calibrations .

~!!-~

I1 5% Error - 0.3% spread ~

." ~

I:

t

C

C

C

C

C

C

C

C

~

C

C 20 t

C

C

Page 80: 2006 IEEE Radio and Wireless Symposium

I

I

I

I

I

I

I

I

I

I

I

I

I

I

I

I

I

I

I

I

I

I

I

I

I

Random Errors

• Noise - KTB, Phase noise, etc. • Optimize VNA for greatest Signal-to-Naise

• Highest practical power level • Minimum IF Bandwidth

• Crosstalk • Increase RF probe spacing • Minimize common lead inductance

• Use GSG layouts in lieu of G5/SG III Use Infinity Probes in lieu of ACPs

• Unwanted modes • System Drift

• Time & temperature

+IEEE Slide 41

Measurements vs. Input Power

21

Page 81: 2006 IEEE Radio and Wireless Symposium

C

t:i:

I:;:

t:;

Measurements vs. IF Bandwidth

.IEEE Slide 43 .

Measurements vs. IF Bandwidth

.. • IF BW will affect Statistical Data

t:;

t:;

~

~

.t::;:

~

t:t r:: c: c:: t::

c:: r:: t::

1:= E::;:

a J:j:

Ct

Ct

Ct

c:: t::

-c: c::

22 c::

C

C

Page 82: 2006 IEEE Radio and Wireless Symposium

I

I

I

I

I

I

I

I

I

I

I

I

I

I

I

I

I

General PNA.Recommendations ·

• Signal-to-noise ratio degrades calibration • Use < 300Hz IF bandwidth with no averaging

• Can be increased after calibration • Use high RF Power

• Stay within OUT linear output limits • Use the least amount of necessary attention

• Port 1 normally ~ 10dB for 55 gain devices III Port 2 normally Zero for SS gain devices

• Use shortest possible low loss cables • Gore is best

• Maintain constant temperature

"IEEE Slide 45

I

I

I

I

I

I

I

I

I

L

I

I

I

I

I

General 8510 Recommendations

• Signal...to-noise ratio degrades calibration

".' :+~. :"

> •• ~,'.~

• Use ~ 256 averages (Fixed IF BW) • can be decreased after calibration .

• Use high RF Power • Stay within OUT linear output limits

• Use the least amount of necessary attention • Port 1 normally < 20dB for SS gain devices .. Port 2 normally Zero for 55 gain devices

• Use shortest possible low loss cables • Gore is best

:?",-_ • Maintain constant temperature ~~:t~::: :,M't"I'lJ"' ....

23

Page 83: 2006 IEEE Radio and Wireless Symposium

ao

--. IIISTlAl~'_1ly1-I:r--A LAAIA_'Ol""""" ..__

G--EI """ G--O SOI.T

O.l!O

0.05

Froqu,,"cy (GHz)

I:

:: :: ;: ;: g:

;: ;:

m:

Ie

C

C

C

C

C

24 C C

._ Which calibration Technique is Best?

• SOlT • All Hi-Q measurements <20GHz • Most measurements requiring attenuation .• Most measurement <-30dBm input power

• SOlR • All probe card applications • All dual signal probe applications • Right angle probe applications

§ 0.15

i e0.10

~ ?;:

NIST Verification

• System drift baseline • LRRM compares with system

drift limit • best fixed probe position

calibration

• SOLT /LRM • growing error wjfreq • possible cal Kit error • possible ref plane error

+IEEE

d

:a :I ~

::I ::I

~

~

~

~

t: t:

~

C

C

Page 84: 2006 IEEE Radio and Wireless Symposium

•••••••

iii

I

I I

I

I

I

I I

•I

•••• I;

Ii

I

Ii

I :,.

I

Ii

I:

I: I:

EERadio Wlrdess

;1~~~ Which calibration Technique is Best?

II LRM (with auto load inductance) • Most accurate attenuation measurements • Some on-wafer standards

II LRRM (with auto load inductance) • Best for broadband mmW transistors • On-wafer standards with a single load

II TRL III Microstrip mmW device characterization • Waveguide banded measurements • III-Van-wafer mmW miaostrip standards

.IEEE Slide 49

Independent calibration Verification Standards

Open stub

25

Page 85: 2006 IEEE Radio and Wireless Symposium

IEEE Radio _WIldes:>

~S~~~ Measurements "and De-Embedding

. - After calibration, the measurement reference plane is at

. the probe tip

- What is measured is the response of the device and the parasitics associated with the pads

+IEEE Slide 51

De..;Embedding and Verification Test Structures .

26

Page 86: 2006 IEEE Radio and Wireless Symposium

t

I

I

I

I

I

I

I

I

I

I

I

I

I

I

I

I

I

I

I

I

I

I , I

I

I

I

I

I

I

I

I

~R.= ~~~~ De-Embedding from OPEN and SHORT

•-­

The parasitics of the OPEN consists only ·of parallel elements to the OUT

- More importance for high impedance devices

. The parasitics of the SHORT consists only of series elements to the OUT

- More importance for high impedance devices

Use of Z and Y correction also helps eliminate residual cal errors'

.IEEE S1ide53

.OPEN dummy

.~:

" .. -, ....

Sxx

·.• Itt·· ........" ....

":. .";

27

Page 87: 2006 IEEE Radio and Wireless Symposium

11.""'0 ,r S;t'lr;~(A

.. }~ SHORT dummy, de-embedding from OPEN .

Sxy R1 L1

L3

+IEEE Slide 55

. Z1, TO Ii 'ZFl'o .. ~R

Sxx

ZI

Sxy>,·····

28

Page 88: 2006 IEEE Radio and Wireless Symposium

IEEE Radio andWI.-.

~~~~~~_. R_eq ..................u_i_re_m_e_n_t_S_u_m_m_a_rv _

.. Microvolt &femtoamp measurement resolution • Parameter Analyzer • Triaxial Chuck for Back..gate

.. low noise floor • MicroChamber for best shielding

III Reliable probing tools • semi-automatic probers • Automatic calibration & verification

III High impedance S-parameters • High frequency device layouts • De-embedding devices

.IEEE Slide 57

There are a number ofdifferent techniques for measuring high Impedance states utilizing either a Network Analyzer or Impedance Analyzer. .

Accurately measuring high impedance states at microwave frequencies is a challenging endeavor. .

Measurement systems are available from Agilentand Cascade to help you deal with the challenges of measuring CNTFET's.

29

Page 89: 2006 IEEE Radio and Wireless Symposium

•• • •

•~

•• I

I "High Impedance S-parameter Measurements" I

•I

Jon Martens Anritsu CompanyI

I

I

•I

••I

I

I

I

I

I

I

I

I

I

I

Page 90: 2006 IEEE Radio and Wireless Symposium

~

I

I 11/23/2005

I

I

I I I

I

I

I

I

I

I

I

I

I

I

I

I

I

I

I

I

I

I

I

I

I

I

I

I

I

I

I

IEl;ERadio andWil'\!lw 5vtnpo&iUm ·~t"·!p~lw(;u5 _5;\."""',.{~_'

High Impedance S-parameter Measurements

J. Martens

AnritslJ Company

---IInrltsu --- ­ +IEEE Slide 1

10 Introduction

'. .'

• S-parameter measurements of multi-kohm devices can be difficult due to the usual 50 ohm hardware:

- Mismatch ' ..

. - Signal'7to-noiseissues... ... '. '. .... ., . .- .High'fevel noise and repeatability swarTjprng~impedance

changes

.- Capacitance effects

1

Page 91: 2006 IEEE Radio and Wireless Symposium

---

11/23/2005

Ex: uncertainty in a 50 ohm measurement

Absolute Impedances of 10500 ohm load with transferred uncertainty (no transformer)

20000.,.----------------,

18000

16000

14000

-; 12000 • III ! 10000 --Lower limit

.. .. .. Upper limil § 6000

6000

4000

2000

o +-~-_-_-_~-~-_----'

o 0.02 0.04 006 006 0.1 0.12 0.14 0 16

Frequen~y (GHz)

Ignores DUT capacitance, etc.. Just the basic measurement uncertainty assuming a good calibration. .

____ ~II~n~ritsU!..!.:·== _ .IEEE Slide 3

Outline

• ··Basic uncertainties: why are there problems with conventional . SO ohm approaches .

• . Transformil1gstructures: dis9rete. tapered 'Jineand diplexed .

• .Calibration approaches in transformed systems . "":,Standards requirements

->Uncertainlyjmpacts .. .'- Measure/1lentexamples""

. " .:;,".:" ·vi i ,':,: ,".',;;:.., <:.' :~:- ~\., '.'.0 :.:~: '>'. ~ . " , ." . ~:<.~(: -: ·~:-\'.'.'::~t:'-~~< ,,/<' -,'­

.........• ';Fl,lllylrahsformed rii~~sllremerrt$:~ppr()~hhks~~dlimitations'.,· . ."' ...~ ,'- - ....... '~.'.., . ."

.. ,.~ . '--'.<"

2

Page 92: 2006 IEEE Radio and Wireless Symposium

---

•~

•••••••••••••••••••• !I

••••••••

Conventional measurementissues

• When measuring high impedances with a 50 ohm systems, the mismatch issues are obvious (affecting power delivery).

•. The resolution problems may be less obvious.

IrI=-O.17dB In=-O.09 dB @ Classical uncertainties in reflection may be on the order of o. 1 dB making this distinction challenging.

___.=-,II:.::.:n=-=rI~tsI:::=u _ +IEEE . Slide 5

~= Conventional measurement issues II ~6~~;~~

..,....--------....,.---------­• Dominant uncertainty components for this measurement type

. - High level ortrace noise (set largely by source phase noise and receivernoise); on the scale 010.01·0~02 dB pk; .

. .

- ResiduaLsource match (set largelY by t~ecalibratiOn); often on the . scale 010.02 dB. ...... . . . . .. .

"'-cCabledrift and C()hn~ct()rrepeakibnity:AffectedstrongIY by the .. ' setup.·Ganbeon t,hescaleof,0.02.0.05 dB (Of Worse):.. .. .

.:;'~ __ .:.',.,;. ':""'>:/;>'--<:""'_ '" " >~ -'-'.~<: .• " .~,,'<' .,.. ,," .-:,::' '.' ~Y':"_ :;;:':.:-~ <'I,.,,;

11/23/2005

3

I

Page 93: 2006 IEEE Radio and Wireless Symposium

11/23/2005

Conventional measurement issues III

• In transmission, the problem is more directly signal-t<rnoise related.

.Even ignoring capacitive effects, JlS-scale gm leads tolS211 < -80 dB.

With measurement noise floors in the -100 dB range «40 GHz), this can lead to issues.

___ ~/f~n~rI~tsI~u. _ .IEEE . SIlde7--_.

!= ConventionaHssues: capacitance t1"~~i4'}'t,J.~ ...o..!""' --~---_- _

.• Even withoutmeasurementinstrument complications, pad . capacitance can ;Cause·a very low frequency pole..

4

Page 94: 2006 IEEE Radio and Wireless Symposium

••

• I 11/23/2005

I,

I

I

I

I

•• I: Ii

I,.: I

I

I

I

Numerical impedance transformations?

• Numerical tools are available to generate S-parameters asjf they were calibrated in a different reference impedance.

• This does not escape the physical SIN and other limits. • Typically one is IimitedJo the':" 1-3000 ohm range witha50 '

ohm physical calibration;

Numerical transform behavior relative to uncertainties

0.8c 0

'li 0.6 ••Uncertainty limit ..i 0.4.l!! a; '0 0.2

0 0.1 10 100 1000 10000

l\11>Bdance environment (ohms)

____:1'1....::;'n=rItsu== _ Slide 9IV\' &. o.a-................. .IEEE

I

I

• I I

I

I

I

I

I I

I

I

- get to 500-1000 ohm ~nvironment will enable 10-50 kohm measurements

,", ~Structures. ""., " ~Calibrationlssues'

_. Uncertainties·'

IEEE Radio S t' ~= ome op Ions ~i".-:<,;l(.';.'.;:'~'~9~¢;(A.

• Local transformers:

t' t0 Improve measuremen s ' _

5

Page 95: 2006 IEEE Radio and Wireless Symposium

t:t

11/23/2005 a t:I:

Transforming structures

• Improvements from broadband transformers - Capacitive interaction problem reduced (if transformer

placed between pads and DUT) .. . - ·Measurement problems reduced (closer to·the sweet spot of

the measurement tools) .

• Options . - Active transformers (easier wI CMOS integration), multi-GHz - Magnetic structures, scale 1 GHz

- Tapered transmission line, potentially 100 GHz

.- Diplexed or switched combinations of the above

___.~/I~n~rl=tsu= _ .IEEE~WbM'$~

::; ::; ::;

ll=I=

R=

l::+

c:+ ~

:;

t:+ t::+

Slide 11 c:

~ERadiO·T. d '" t$ym= ..... apere ,me concep s . "'"O_<A-..,.'t'r--'9<4~:;a~ -..,. _

• As is well-know,tapered transmission lines can be used as broadband transformers covering decades. . .

• Lowtrequencylimit set bylength·oftaper(qu~rte;' wavelength· scale). ... .. .... ....•. .

·.HighfrequencY1ilTlit often set by granularttyoftaper,·:.•... dispersior( andmoding.· .. . ... ...

,''T-----,-------, - 7:1 transformation .. \ . ,:

I

ct

r:t

c: ;::+

a ;:;

~

;:;

;:; ;::

c C

t:t

=:t

c= 6 e:t

a

Page 96: 2006 IEEE Radio and Wireless Symposium

---

freq_~,:~,:..<},..

11/23/2005

~= Tapered transmission structures (higher Z) 5al'-'~(:A _v.... w'~~~ ,

To get to higher impedances, different transmission structures may be needed.

Reduced ground plane microstrip can generate Zo,of many hundred ohms. Skin effects (anomalous and normal) become more important at higher ZOo

lD ia' Nduced ground mlcrotbip, erdJ

"',---------------, "" . '"

• •

• " ,+---_--~-_--_---i, ',5 2,5

woo

C. E. Smith, at aI, MTT-33, pp. 835-839. Sept. 1985.

____=.II-=.n=f:.:.;itsl==u _ +IEEE Slide 13

IEEE Radio ~~ Tapered, high Zstructures >;"»N ~$' .;O!tt~tm~ _•. ~U~pffi ....;.

To illustrate, a single quarter-wave transformer was created using ground, plane cuts to perform the transformation. "

'. Return 1018 Of high ZIold wlIh quarter wIve ground

plsne 1rlnsformer

o -2

11 -4I ~:r---__

! -10 o It ·12

Alon~e~:~mobil7er . 'S i :~: cc • cCc .1ap~iivQ!JI(Jiif}fJlJlfi>;':.i'~;:' ·20 +-.--~--8--~7--~---1

.:~[~~t,~~~.~.:.,.;,1c~c'1';'~'-"':"';c,...,.E':.. ,.. '.,..

>·M+i.">: .. _.-.' ----,....,.....·~·;:'.;iel:~;~~~····

7

Page 97: 2006 IEEE Radio and Wireless Symposium

---

SWitched/dip/exed transformers

. . . . ..". . .

• Since tapered line constructs may not cover the low end, diplexing or sWitching with another technology can help.

• .This canexpandthea~ailable bandwidth from a :decade or so to much more.

___,::-I'I=.:n:=-=rI=t5u= _ +IEEE Slide 15

Ie.

11/23/2005 C

c:J

t:f

CI

1:1

~

~

1:1

E:1

E:1

~

c:I

C

C

tl

C

lEES' Calibration approaches with transformers l"r-:':'J:~§,"'t.~

"'l>~CA~~...",.~__~~-,!",,__~~_...,..~-.-~_-:- _

• Approach 1: Calibrate in 50 ohms and de-:el11bedthe transformer - Simpler calibrations '. '. • " .. . . . '. . -. Transformer charactertzationmaybe difficult. 'Back-to-back measurements

limit aCcUracy. May still need standards at high impedance. '.' ·t .' .' . . . '. .' t

'J" • • I " '.~: .. ~. . ~

".. ~I . I . ",'" .1 . I "I

1 I I I

. "' I .' .:'....... .... Calibration planes, > :

.~ :Appr~6ij~~~atibratedi~~ctiy~tf1i9H)mP~(j~nhe" ..'\, ....• : "'.'. Need~ndards:loads,willbemoStehallengJngf>ioadband: ' .... , ,,', "~; ::Defi~ed-~ta'}~ar~sm~thods (SQLT/SO~R)Wi!i:Wrir~(atreastover:l;rIiited ..•.... . .

:fr~u~pY,~n:ge~tburli~e-base~'.~p!q~C~~~.·q;~!::;;fllg,)a~''rn9!e:a!tr~gt!y~: .'

C

~

~

~

c r: c c r: r: t:

t: C

8 C

l:

Page 98: 2006 IEEE Radio and Wireless Symposium

---

I

Ii

I:

1

1

I'

I

I:

I

I

I

I

I

I

I

I

I

Ila io High Z calibrations: defined standards

- Techniques such as SOlT require one to know {characterize} . the reflection standards such as open, short and load.

• Classically, this is done by reference to airlines {the impedance standard} or some other impedance artifact. .

- Shorts and opens can be characterized by a similar fitting process or via EMsimulation.

.Must characterize Must characterize series fringing capacitance (+ inductance. electrical length if any).

___,:.../I=:n:.:..::rI=tsu= _ .IEEE Slide 17

11/23/2005

I

I

I

I

I

I

I

I

I

I

I

I

I

I

I

..

!$More on defined standards t""'~i&'1::~

~IaO~tA

-load m~o:-:d:r:e~li=-ng~is~m=-o:-:r=-e-:co~m~pT:le:':":x-:di':":u=-e~to::-t~h-=e-r.in:-:c~re:-:a:-:s~e":l'd-r.im~po:-:rtaT..~n-=ce=--=oTf-shunt capacitance {since the resistance is higher}.

,',' .<--~.'" '.-::"">,,' ': :-..~:.:::>, ~ .. ~ ,;;-':'." :'·:".'·::\/:f'<

Classically. just some series inductance is modeled.

Additiona/termsbecome mandatory . .forhigh Z.

9

Page 99: 2006 IEEE Radio and Wireless Symposium

---

t:

11/23/2005 C

I:

r:

High Z calibrations:Lines

• For TRL, LRM, etc., line quality is important and Zo will set the reference impedance. .. . .

·As discussed withthetransformers, some unique geometries maybe needed. . . .. . .... .

Line 1

Line 2

.. .

Since TRL cal BW scales with line length difference, low end coverage is problematic.

_____,;../I~n~ri~tsu= _ .IEEE Slide 19

I:

r: r: r: I:

1=

1=

I:

t: C

C

t

TRANS/REfL311 REFLECTION '"PEDAMCE ,MUP J

.~~ .Calibrations after transformers: example

.~~~ ;~O~<A,-. ~~--_

•. Transformers were used to set Zo=700 ohms and reflections from a 105000 8M resistor (as OUT) were measured.

t I:

t J:

~

C

C

C

C

C

t

C

C

10 = ~

"'-*

C

Page 100: 2006 IEEE Radio and Wireless Symposium

iii

• 11/23/2005

• I

I

I

I

I

I

I

I

I

I

I

I

I I

I

=High Z cals; more examples y W$.'R,w'J4

~V!\~':A

• Extracted impedance of several different OUTs are shown here. Cal @ 700

. ohms Me.IUred 111tp«18ftCe .r18500 ohll'l ellip N:lkeor (cal .ltBr

trWI'fGnMr)

/~ r:::: ~~l- ......... .. J---~~~~~~----4' o 0.02 aB< 0.06 ooa 0.1 012 014 016

F...........".fllMl.l '-------~~-~~~--

.

-;::::::==::::::::==:::::::::::::=~~__ IlualrI'Od iIIIp«IanC8 of 1000"'''' chip rellllor(C11lllfter

"-••1CM~

"'.--~~--~~---r' 7"0 11

,~ . "" .

• 110 ,.....,. 11 r-;RIJ:: ...­ ~o~ 680 6Jg

:L,"iIoor..~.~~~,~~!!I!!... .. J:~.~~~,~.!:~",.~,..~ . o16

'''.-CY(G$!

......* I!ft,pedWiIt. GI' 1000 ohlncll~ ,..lato, tllll'-'-' Inllifonner)

1100 •••• •••••••••••••••••••.•••.•.••••••h.......... .... ··.. ··· ·.. ,·.· .. 100

tOllll ~ :

.'02{l 00 r-;!i1.~'~"~I: : ~ -- "- . ~

-(l tun \l.Ot 0.06 M'lI IU 0.1'1 01' 0.18

, F_""fIGl4l

. Trace noise-tO for 1 kQ OUTs, -50n for 10kn OUTs

_WNrs_'~. ___·.~/I::.:.:n:.;.::rl;:=tst:::.::u. _ .IEEE SlIde 21

I ~£Uadlo . . .'=.= Uncertainties for transformed measurements .""'''''11'',(1\~1'1<~~~ . _I

.• , SIN noise effects reduced (noise floor now·lessof an issue .'I fortransmission,trace noise less so for reflection). '.. . .. . .

I .' •. Quality ofcalibrationstandards willaffecft/1eresiduals. :'

I .• Repeatability of the transforming structures is critical.

. .... . '-" .' . . .' , . .' .

,'I .... IrrtpedancerBSolution isimi>rc)v~d~inC('lthe ~Elnsitivjtyof;·· .... ,··.impedancel0.reflectioncoefficient;(~ss~:bytbe;VNA)has·· ..

,":C:'..§'·;decreased,}.\.. ,;....• ,,:""'" "'N':""';,)i;")\' .... ' , .

11

Page 101: 2006 IEEE Radio and Wireless Symposium

Uncertainty in transmission

The noise floor effects alone can be sLibstantial. Adding in other effects makes the difference much more extreme.

19211 Uncertainty for example OUT (excludes launch effects and assumes good slandardsl

2.5 .--~~~--~~~-~-,

850 ohm cal

.700 ohm cal

N

i1 0.5

:.~ ~~ 26Hz

L . ---'

+IEEE S1lde23

M=1e-6S

R1=10000 ohm

R2=50000 ohm

~ ff

t

t11123/2005

:: :: .. !fill

~

;::

c: c: ~

~

~

I::

c: C

c:

.<"; ,,~~ ..~~:,., :~'~;'~';/<:;/~L,,::/':

;;.i~~~~~~1~2~

MeallUred Iliol 10500 ohm chIP tor and tnlnllfelftd w"certalnty boy"" (CIt Oft)

... ,.. O-l--~-~--~_-_--<

o

-,----~~---~----,

".. -::l~"""I ­ ......­EEI2l

-loWIrlillll

. .. .. ""'~

,

~ ~. i.. . ;

. ' .. ~l

AhsolU1e lrnpfldane.n ot 10500 Ohm lead wilh IrlnafelTed uncet1lllntr (no I\'l!lnaformer.

Uncertainties in reflection coefficient and.impedanceare shown here (calibration in 700 ohms).

J.!ia~~ Uncertainties in impedance Symposium ~r"'<~ l't"Ut't!~,~O:t\i¢;("A _

• ThelZJ uncertainty isnow10x smaller than with a straight 50 ohm calibration. . .. . . .

C t:::;

C

I:;

I:l: g

a c I:

It

It

It

It 12 lit

It

C

Page 102: 2006 IEEE Radio and Wireless Symposium

---

••••• I

I

I

I

I

I

I

I

I

I

I

11/23/2005

~~ Full 2 port measurements: example F"WMttZ.~xK>: . . 'S.Vj~<:A_· _

While direct comparisons to other calibrations is difficult, one can compare to simulated results using simple models. Measurement frequency < 1 GHz. .

IS211 01 series high Z structure (calibration In 700 ohm environment)

·5

iii' ·10 :!.

II

.measured -mmodeledN

~ -15

! -20

·25 1000 10000 100000

Seri6& R (Dhm,)

___,~/I=n:.:..:rltsl==u _ .IEEE S1lde25

I

I

I

I

I

IEl:E ftadio ~~~ Full 2 port measurement (cant.) ~~l¢'~~tl<~1Vi:m'~~~ _

IS111 of the same transmissive nUT class of varying impedance levels. .' The agreement became slightly worse at impedance extremes; likely due

to stretching the OUT model class outside its intended range..

1$11101series blgb Z SlnIclure (calibration In 700 ohm envlronmenl)

., t--+--+--HH-:I:-HlF----t--t-H-+-t+t1 ·2 t--+--+--HH-H+t--t--t-H-+-t+t1 -1t--+-.....+lrl-H+t--t--t--H-+-J-ffi

lii .. t--+-+-+lH-H+I--t--t--H-+-J-ffi .measured

~ -5 t--+-+-+lH-H+I--t--t--H-+-J-ffi .modele6

ii! ... t--+--+--HH-H+j---+--i-t-+-H-I-ti ·1 t---+--++t+t-H+--+-++lr+H-H .. t--+--+---t-IH-rt+t--t--t-H-+-t+t1 -9t--+--+--hH-rt+t--t--I-H-+-t+t1

.'0 -l-_.J.---l-..l.-'W-U.l.+--_J....,-.J-.l-..l..L.l..l.Lj 1000 100000

13

Page 103: 2006 IEEE Radio and Wireless Symposium

---

C

r::

11/23/2005 c:

~

Summary: transformed measurements

.... .. .. . . .. . .'. . .

• For measuring 10kil-classimpedances,transforming to the few hundred, ohm range can help - Modestuncertainties

Reasonable accuracy

- Relatively practical transformers and cal structures

• The usual VNA cal approaches still apply. Some care must be takenon load standards (if used) c:lndthe,line standards.

___,~/I~n~rI~tsu= _ .IEEE ' SlIde27

= Fully transformed structures ;'7- l t,J.o>ffl,1!1:3:'¥t"".,,0_<,,_,....;. --­

• ,The refieCtometers for measurement can be built entirely onto the

DUT",afer.,:' ,', '" ,', ••',' "'..," ,'" ,'.'. ,,' , • ,'This can improve measurementstability since the reflectometryis

:not1J1asked"byl~etral}sf~im(3r/Jaunchand itslpss. ..' '" ',>

,,' Same'requir~me~ r~maiiloncreating calibration ~tandards.

~

~

1=

J:

E:

~

~

I:

~

C

C

t

t:

:: ;:

c c C

t ;:

E

t

t t

t

t 14 t

~

Page 104: 2006 IEEE Radio and Wireless Symposium

---

I 11123/2005

I

I

.;.",:,·.,;.·f:·'>~~ ", " .>->J.'-:'::~.\, <,.' .;w'"

,-).CMi;J' ·.~.i;.·.]~t;j\._ .. :,i.,~.;.~<~---'; ---........,

I

I

I

I

I

I

I

I

I

I

I

i~$ Stability improvements ~rd:;:W¢..t:A _~{.'"''!-9'.~~~t)Q6.

·Loss (and mismatch)between VNA and DUT reduces raw directivity. .' .

-The transformers,in this case, can then reduce stability somewhat in the classical setup. .

. . .

-With on-chip refleetometers, this excess loss/mismatchis removed and raw directivity is theoretically 5-10 dB better (depending on freq range).

.New raw directivity=

1 . S12S21

. - 1+ S22XSU + ed)1

___,.:::-/1:.:..:O::-="=tsu= _ .IEEE Slide 29

!= .Stability improvements (cont.) ;S..9i<>l"(A _1"'f9"'f'·fi~

. ­-While an extreme case, a 6 dB transformer loss can lead to measurable deterioration in <1 hr (but still within uncertainties)~

. . .. .

-A thermally changing environment exacerbates the effect.

-Match ofattuu line sometime after cal is shown here. . .

Rell'ldual ttlrv ine 922 (0.5 IIrafter cal, therm81~ changing)Resldu811t1ru line S11 (0.5 hr after cal, therm8~ Changmg)

15

Page 105: 2006 IEEE Radio and Wireless Symposium

---

Receiver structures

(usually) off-chip converters

To IF system

Transformation may even not be needed in coupled arms due to lower criticcllityof mismatch~

Net noise performance usually better since high freq paths are minimize,d.

___...::....:/I:.:..:n~rl:==tsu:.=:. . _ .IEEE Slide 31

Q

11/23/2005 C

I:l

C.

E:

E:

E:

E:

E:

E:

E:

r: t: r: C

t: ~

IEEERadlo ry::,~~Fully transformed measurements (cont.) ~7':"~<:;m~'

,"",1';e;R<A_' _

.' Places weaker constraint on transformer repeatability. • . LikelybEitterdynamic range (depending on how detection is .

done). May need a complex receiver structure. . . .Mor~spa~needed, . " . .' ' " .

2.5 _.~_ __ _ __ __ _ M •••••••••••••_._••••__ •• _ ••_ ••

r=--,------,---------,I .... ··;Dyn~rnicra~ge;' ,';,·jmproVf3mentsfrom

==-====I··~ih§l~mbedde-dstructures· ·'::c?rt4SJpfJiUt.:j:t.l;··: ". .' ..~. - _. ','," ..

Example 15211 uncanalntles (DR-baaed) for hypotlletlcal constructs

i 0.'

Ii 2 +-------­~

tI U+-------­~ . ,+----­~

.' .Lowendb~ndwidtheven more chalienging to achieve. .

~

I:

t:

C

I=:

c: C

C

c: C

t:

c: t:

16 C

~

Page 106: 2006 IEEE Radio and Wireless Symposium

•••••

•••••••

11/23/2005

I

I

I

I

I

I

I

I

I

I Ii

I

• I

I

••

IEEE Rad,o ~= Summary ~('>W~~~(X> _~"'l~lOw~.C4.

• Higher impedance S-parameter measurements are possible with some modifications to the environment.

• Transforming to 500-1000 ohms can allow reasonable measurements in OUT impedance environments up to 20-50 kil.

• Traditional calibration techniques basically apply with some attention paid to standards.

• Embedded reflectometry can also help. There are potentially stability benefits but more space is needed, low end coverage is difficult. and the apparatus will be more complex.

___.::...;/I:.:;.:n::.,:rI=tsu= _ .........- .IEEE Slide 33 _

I

I

I

17

Page 107: 2006 IEEE Radio and Wireless Symposium

••••••

••••••

I

••I

• I

I

I

I

I

I

I

I

I

I

I

I

I

I

I

"Testing Frequency Response of High Impedance CNT Transistors"

Dan Woodward

Tektronix

Page 108: 2006 IEEE Radio and Wireless Symposium

IEeER~iO

""Oi_}';,~ Testing Frequency Response of High Impedance

.... CNT Transistors ...

Dan Woodward- Presenter

II

.Tektronix, Inc.- PhoeniX, AZ • Account Manager and Application Engineer

"IEEE Slide1

BriefBio • Dan Woodward- Tektronix, Inc.

. • Mobile phone. 602~359-8490 • EEgraduateo{Unl\I.·ofIIHnois, . Champaign/Urbana ••... . ..

, .' . ,

1

Page 109: 2006 IEEE Radio and Wireless Symposium

trllt'<l*'w-A

RH:Gain vsFrequencyof CNT Transistors ~1l'~~A_' .....;. _

The problem: .

CNT transistors have very high gate and drain impedance compared to PET's.

Why this is a problem ?

1. Traditional VNA's are designed for 50 ohms. They basically short out the CNT transistor.

2. High impedance probes ( 1 M ohm) are not readily available for wafer probers.

3. Off the shelf solutions for non-50 ohm RF characterization are rare.

+IEEE Slide 3

~= SVl\'lposlUIlI Abstract _,,,"~,,,,,,.(A_'~i'~·"'!?·u~~

.. ' . .

• Testing Frequency Response ofHigh Impedance CNT Transistors' '. .... .

D Carbon Nanotube-baSedtransi~ors have avery high input and '.' output impedance, which rules ourS-parameter testing with traditional Vector Network Analyzers. New approaches to measuringthe gain at multi Gigahertz frequencies were needed at Motorola.' '.

ill .A method was ;propgSedthatwouldtal<e advantctgeof new. RF~band· .;.wi.dth.•·.o.~c.iII.oscopes.,. ".h.i9 ... ... ce pro.res(anddigital.....S;9.".. . hinP.u.t i.n1pedan .. .. a.I . processlnSJaI9orithms;llle'm~sured -results would be the., .... ......•. .'

Frequency R~sponse·Functioti{FRF} (jftl1e.·CNTtransistors;ra~er ..'. .' than the traditional S~parameter:S2a~crhls 'rnethodoflTleasurJl1g.· . " theFRFallowsa<choice ofstimulus; 'broadbandnoise,impt;l1sej .\

;,chii'p~j()rstepf~nction. StepJundionswere chosen ;forfhis,;pro]ect " 'because~Aire.readiJ)'avallab!, ·ttiJa~,~dges ~at g:nerat~" .. \" frequendesfrom J:)C,to ~1;5 GH~.", , ,"" , ,~, '~ ,0

\;~;(: t ;~c:ff~~:c'!:;t}?C'e~'?~~~;;Y'·:.i~k~(

-= E: II::

-= I: 1:,

a: c I:

I:

I:

a::: c I::

C

C

C

C

c: t:

t

:: w:

2 :: t

Page 110: 2006 IEEE Radio and Wireless Symposium

I

I

I

I

Abstract

• The tests are conducted by stimulating the gate of the eNT device . with a ·fast edge and c;:apt~ring the averaged inputand the output

.simul~neouslywith .hlghIlT)pedance(1 Mo~m). probes..These stepfunctions are then differentiated to getthe Impulse responses at the . input and output, and then the averaged FFTs can be computedwith the 'scope's math functions. A ratio of the magnitude of the FFT of the output over theFFT of the input yields the gain of the transistor. A similar~process. is uSeQ to get the ph.ase sflift between .the output and the Input using a difference function.' An even more

.' robust math solution Involves the use of The Math Work's MatLab DSP Toolbox, which uses the'cross spectrum average approach to compute the FRF (gain and phase) and the Coherence function. The Coh~rence function adds th~abili~ to determine the degree of quality of the FRF as afunction of frequency.

• It is necessary tode-embed the cables and probes from the entire network. Aprocedure for doing thatwill be described and the results documented. . .

-------------.. • IEEE Slides

I

I

I

I

I

I

I

I

I

I

I

I

I

I

I

I

I

I

I

I

I

I

I

I

I

I

I

I

Radio

~~ Gain vsFrequency ofCNT Transistors

Experiments all coriductedat Motorola Labs, Tempe,AZ by: . . >.,' ',' .

. "." . ..."... .

• ·tshunslIahA.mlarti~'MotorolaLabs: Tempe,AZ

II DanWood~ard·· -Tektronix, Inc.: Phoenix, Ai -Tektronix, Inc.: llYllIC.,. "--.t:\.

3

Page 111: 2006 IEEE Radio and Wireless Symposium

c C

I:

I:

I:~$ Available Micro ProbeslIi.."''''I',(A ..............;.-.;.

~

-= -= r:

-= 1=

1= I:

I: 1ft .'~ ------- .IEEE Slide?

MnS>~ __#,­

r:

<k=~ 1 Megohl1l SCope Probe ' I:

I:

C

I:

C

E:

I:

I;

I::

I::

It 'I::

E: 4 I:

I:

C

C

Page 112: 2006 IEEE Radio and Wireless Symposium

I

I

I

I

I

I

I

I

I

ScQpe Probe Interfaced to Wafer Probe

High BW Scope ,

5

Page 113: 2006 IEEE Radio and Wireless Symposium

1

I

a Cl: 1:1:

1:1: (EEERadiO

~= M hod useel taln FRF~¢,~~:&'><I et to Ob'

. 1. Stimulate DUT with fast edge. . 2.. Digitize Input and OutputSimultaneously~·· .

3. Take Derivative of Fast Edges (yields Impulse).

4. Take FFT of Impulses on Input and Output.

5. . Take·Ratio of OutPl;Jt over Input to Normalize.

6. Subtract Phase of Output from Phase of Input.

"IEEE Slide 11

1:1:

1:1:

r=+ R I::;

I:;

I:;

~

~

E:;

c c c

Example Step Input r:;

c ~

1:1

t:I

C

C

t:l

~

t:l tj

~

6 C IC

c

Page 114: 2006 IEEE Radio and Wireless Symposium

Derivative of Edge.·

"IEEE Slide 13·

7

Page 115: 2006 IEEE Radio and Wireless Symposium

IEEe Radio F d ~ S· I!~~ ast E ge I or tlmu us

10,OftB/dlv '~,oo.~~rt ~H3PBf~t .C.i/.#'4'j5,1MV

"IEEE Slide 15 • I

•I

Bias T added

I

I

I

I

8

Page 116: 2006 IEEE Radio and Wireless Symposium

IEEE Rildi(l Derivatives of edges-~ ~j;(tn;~G~ _..~~~~~

~':" .... ,;,..~ .,I>-4~." .. _< :" •.! ,~, '_"'<""~'M< 'c,*, ,} L.i::\!:~.:,~ j..,,,,.. <'-~~'W!' ,:", .. '. .. """,~,. '"::" .J:"'-",.",.,.""." .....".,."..,.."..,.'"",.-",-:._,--~,-'it';,-_.- --,-_ .•-,-_.,."",..,.rr..,.,,,.,,,;{ -,~ •

~~.. ~ ....

ro'.t/ltIV..Q:Ellttlm\1J:t $&.tNV4S.~ 2QAm~~

"'IEEE SIIde17

Adding Bias Teest:A

flIe EdIt ltm HlniMq DIg IllsPIOY ""­ _ _ _ _ Ml!S'oopo l.Itl_ IlQ> _

9

Page 117: 2006 IEEE Radio and Wireless Symposium

FFT ofGate & Drain of normal FET

.IEEE Slide 19

calibration with Through Structure

·;".·.·2?<'::":,·y,: ,

-i<\->:,~~~;; ~,.

10

Page 118: 2006 IEEE Radio and Wireless Symposium

CNT resistor Experiment

150 u Probes on ISS TH RU ~'::::':::-::'~~:~"~.::~ II' P ut T H R U .......

--.~ .,_ " . ""-,

OulpulTHRU \

\ 1\ \ i/~ /\.

V \\/ N . 5

Frequency (GHz)

.IEEE S1lde21

resistor Experiment . . . ., .

150 u Probes onlSSTH R U

155

150

lD '45

~ '40

£l "5'c

::l -e '30

« "5

'20

115

11

Page 119: 2006 IEEE Radio and Wireless Symposium

~x~~~

:g -20 :::> . .

·".:;~~:':30·

resistor Experiment

Multi-tube CNT resistor ~ /"-"" .

\~ \\_/"." '\ Inpu! THRU

\

\ /\ //' v \/

Output THRU Iv\ ~ \,

~ N100 +-.,..,...-...---,.---.--r--.----.-~---.-'L..,--"'r--o

o 12345 Frequency (GHz)

"IEEE Slide 23

Radio CNT 'sto' Ex ., t'. . resl ·r '. .penmen

o -, ..\ .. Normaiized ouiputfromCNT-JD-RE~.m -10

\",.~

12

Page 120: 2006 IEEE Radio and Wireless Symposium

~~ eNT resistor Experiment

10 1Multi-tube Resistor Nonnalized Output

5

01- ==::::,..---,iil ~ -5 u..r:: -10 ·4dB@4GHz

-15

-20 BandWidth limited by the Syste

0.01 . 0.1 1 Frequency (GHz)

.IEEE S1ide25

IEEE Radio apdWi...... Symposium ~1.,'t9Ht!_~.{'!",)(,"'"""'l<>"' ...;.;. SWNTFETs _

• Courtesy of Islamshah Amlani, Motorola Inc.

SWNT

........ ..' ~ .." " ..._._.~............. V.t="O.1

~:"~"::::;:::.=-="-:i':":-=" '.' VI:OA'

-0.5' .' ~1;O:;-1;5::-2.0 "VO;(V} '-..

~20

-15

-25

-10

NVAI eat8lyst

-1UOnmAu . Q

~50 :nrnPI .' ,.:,O'ri;'" AlO; .

' ·Gate:sia6t, )-~->.,

104 S ...-. VI=-2'gm -- . Il ;;..-;:: VI= -1.7 . ;r-:...... Vt= -1.4

;00...........--... VI= -1.1

..........._- .VI=-o.ll

~........._~............ V';;-o.5? -..--......-- ........

. ............................ Vt=-o2.

13

Page 121: 2006 IEEE Radio and Wireless Symposium

IEEE· Radio dd I b. I~;~A itiona "esu ts"''''0'_'''' _

At the time of submission of this document, additional experiments were in progress. The slides and results will be presented at the workshop on January 15, 2006...

• IEEE Slide 27

GtJ':\rJ ;

-- ~~~C2- 50 fJ-4 =t

• A .. f)~~ (Ctfl(V\~AN.Y( ) ~~. (V~~~ \ ~~

a~

14