High Frequency Properties and Applications of High Frequency Properties and Applications of Carbon Nanotube (CNT) Carbon Nanotube (CNT)

HaoHao XinXin

ECE Dept. & Physics Dept.ECE Dept. & Physics Dept.

University of ArizonaUniversity of Arizona

OutlineOutline

• Introduction and motivation

• Microwave VNA characterization

• THz time domain free space measurement

• Conclusion

http://ece.arizona.edu/~mwca/http://ece.arizona.edu/~mwca/Research Areas

• Novel Electronically Scanned Antennas• Meta-material for mmW applications• Nano Devices and Antennas• Novel THz Sources and Components• Monolithic Microwave IC (MMIC)• 3-D Integrated mmW Circuits and

Antennas• Power Harvesting Applications• Biological Inspired Direction Finding

Our research focus on microwave / mmW / THz antennas, circuits and applicationsCapabilities including: Complete Design / Simulation / Fabrication / Characterization

Prof. Hao Xinhxin@ece.arizona.edu520-626-6941

60 GHz Packaged Antenna Fully Compatible with Si RFIC

3890 µm

Novel Electronic-Scanned Array

High Frequency Carbon Nanotube Research

Power Amplifier MMIC

Antenna feed

-30

-20

-10

0

0

30

6090

120

150

180

210

240270

300

330

-30

-20

-10

0

Metamaterial Based Antennas

3-D THz Rapid Component Fabrication

100 200 300 400 500 600-45

-40

-35

-30

-25

-20

-15

-10

-5

0

5

Frequency (GHz)

MeasurementSimulation

ARO US Air Force

DoE

ARO US Air Force

DoE

AROARO US Air ForceUS Air Force

DoEDoE

• Complete microwave to THz characterization facilities– Network analyzers (up to 110 GHz), THz equipments– Semiconductor parametric analyzer– Spectrum analyzer, signal generators, power meters,

noise figure meter, etc.– Wafer probe station, antenna anechoic chamber

Agilent E8361A Network Analyzer 10 MHz to 67 GHz, 1.85mm test ports various coaxial and waveguide cal kits

Probe Station with various probe Positioners DC to millimeter probes and calibration substrates

Time Domain THz Spectrometer(TDS) 50 GHz – 2 THz

Testing FacilitiesTesting Facilities

• Full capability clean rooms at engineering and optical science (a brand new SEM & NPGS)

• In-house CVD system for nanotube growth• In-house PCB fabrication equipment

Chemical Vapor Deposition SystemFor Carbon Nanotube Growth

• Access of semiconductor foundries– Raytheon RRFC / Teledyne Scientific Company / MIT

Lincoln Laboratory (GaAs, InP, and CMOS)

EvaporatorSputter E-Beam LithographyMask Aligner

Dry Film PCB Fabrication Equipment

Fabrication FacilitiesFabrication Facilities

• Advanced electromagnetic and circuit simulation tools– Linear / Non-linear circuit simulators– Finite-element frequency domain and time domain

simulators– Agilent Advanced Design Systems, Momentum, HFSS,

CST, SEMCAD++, Cadence, Sonnet …– Work Stations with 12 GB RAM– Evaluating parallel EM simulation tools such as GEMS

Modeling / Simulation CapabilitiesModeling / Simulation Capabilities

3-D structure of MWNT

Images are taken from www.nanotech-now.com

Background and MotivationBackground and Motivation

Carbon Nanotube (CNT) (Diameter ~ nm)

• Discovered by S. Lijima in 1991

• Rolled-up sheet(s) of graphene

• Metallic CNT can have current density 1000 times

greater than metals (> 109 A/cm2 vs. < 107 A/cm2 for Cu)

• One of strongest and stiffest materials

SWNT (Single-Walled)

• Chirality => Semconducting or metallic

• Bandgap can be tuned by the chirality of SWNT

MWNT (Multi-Walled)

• Multi-layers in parallel => Tend to be metallic

Main technology issue: material control / fabrication

Lieber et.al. J. Phys. Chem. B 104, 2794-2809 (2000)

SWNT Young’s modulus: ~ 1TPatensile strength: ~ 30GPa

Structure and properties of SWNTStructure and properties of SWNT

Semiconducting SWNT – active devices

Metallic SWNT – passive interconnects

Possibly: all carbon IC

Metallic

Semiconducting

Field Effect Transistors

- Significant better device performance metric1

- Up to THz intrinsic cutoff frequency predicted2

Passive Applications: Interconnect, Antenna

• Transmission line model of CNTs

• Quantum / Kinetic inductance (~ 10,000 higher than magnetic inductance)

• Resonance frequency 100 times lower (~ ): much smaller antenna

Potential High Frequency Applications: GHz Potential High Frequency Applications: GHz -- THzTHz

1. J. Guo et al, IEDM, pp. 711 (2002) 2. P. J. Burke, Solid-StateElectron., 40, 1981, (2004)

LC/1

Challenge of Single-tube measurement

• Significant mismatch between single CNT’s high intrinsic impedance and 50Ω microwave testing system

• Parasitics often dominate the intrinsic CNT properties

• Difficulty of test fixture fabrication

• Kinetic inductance at microwave frequency has not been experimentally verified

HighHigh--frequency measurement of CNTfrequency measurement of CNT

CNT

50 Ω50 Ω

RCNT~10kΩmismatchmismatch

A CNT across electrodes using Electrophoresis A CNT across electrodes using Electrophoresis

Alternative approach: measure an ensemble of CNTs (CNT paper, array etc.) and extract intrinsic properties

May be necessary for practical applications (bundles, thin film devices, etc. are currently more promising)

Alternative Measurement ApproachAlternative Measurement Approach

CNT ensembleExtraction

Incident & ScatteredEM waves

Intrinsic CNTProperties

Limitation of reported CNT ensemble measurements

• Measured only transmission or reflection

• The assumption of µ=1 had to be made

• Lack of error analysis

Goal: Study high frequency properties of CNTs

• Vector Network Analyzer (10 MHz – 110 GHz)

• THz Time Domain Spectrometer (50 GHz – 3 THz)

• Characterization of both SWNT and MWNT

Characterization of CNT EnsemblesCharacterization of CNT Ensembles

MWNT Paper Sample MWNT Paper Sample

• Thickness: 3.5 mil (~90 μm)

• MWNTs are randomly oriented

Suspend MWNTsin a fluid

Filter the fluid onto a membrane

Support

Dry the membraneand remove the

MWNT paper

PreparationPreparation

MWNT paper photoMWNT paper photo SEM imageSEM image

VNA Experimental SetupVNA Experimental Setup

• Characterized with state-of-art Agilent E8361A PNA network analyzer (10MHz-67GHz)

• Calibrated with waveguide cal kit before measurement

• MWNT paper sandwiched between two rectangular waveguides

• Both of magnitudes and phases of reflection (S11) and transmission (S21) are measured

Port 1 Port 2

• Nicolson-Ross-Weir (NRW)

approach is implemented

• m=0 is selected and verified

• ε = ε’ - j ε” and μ = μ’ - j μ” are complex permittivity and permeability normalized to free space.

Extracting Intrinsic Material PropertiesExtracting Intrinsic Material Properties

Incident

ReflectedS11

TransmittedS21

t

1 21 11V S S= +

2 21 11V S S= −

1 2

1 2

1 VVXV V+

=+

2 1X Xξ = ± −

2

21VV

ξξ−

Γ =−

0

[ ln | | ( 2 )]rpvZ

j mk t

ημ ξ θ π

′= − +

2 2rpvZ

μεη

=′

Extracted Permittivity and PermeabilityExtracted Permittivity and Permeability

• Resulting conductivity σ=ε”ωε0 ~ 1500 S/m, close to DC conductivity 1000 S/m

• No significant magnetic response

• Negative µ” are due to numerical spill over from ε”

-500

0

500

1000

1500

2000

0 10 20 30 40 50

Frequency (GHz)

ε’

0

1000

2000

3000

4000

0 10 20 30 40 50

Frequency (GHz)

ε”-2

-1

0

1

2

0 10 20 30 40 50

Frequency (GHz)

μ’

-2

-1

0

1

2

0 10 20 30 40 50

Frequency (GHz)

μ”

-500

0

500

1000

1500

2000

0 10 20 30 40 50

Frequency (GHz)

ε’

0

1000

2000

3000

4000

0 10 20 30 40 50

Frequency (GHz)

ε”-2

-1

0

1

2

0 10 20 30 40 50

Frequency (GHz)

μ’

-2

-1

0

1

2

0 10 20 30 40 50

Frequency (GHz)

μ”

1. L. Wang, R. Zhou, and H. Xin, IEEE Trans. Microwave Theory and Tech., Vo. 56, No. 2, pp. 499-506, February, 2008

Verification of the AlgorithmVerification of the Algorithm

assigned with computed ε and µ

Excitation

Excitation

HFSS simulation ModelHFSS simulation Model

• A 3.5-mil thin slab is assigned with extracted ε and µ and sandwiched between two waveguides

• The simulated S-parameters are consistent with measured S-parameters

Ansoft’s HFSS (High Frequency Structure Simulator) is employed for simulation.

0 20 40 60-0.5

-0.4

-0.3

-0.2

-0.1

Frequency (GHz)

0 20 40 60-180

-179

-178

-177

-176

Frequency (GHz)

0 20 40 60-34

-32

-30

-28

Frequency (GHz)

0 20 40 60-40

-30

-20

-10

0

Frequency (GHz)

Measured

Simulated

Measured

Simulated

Measured

Simulated

Measured

Simulated

dB (S

11)

dB (S

21)

Pha

se (S

11) (

deg)

Pha

se (S

21) (

deg)

0 20 40 60-0.5

-0.4

-0.3

-0.2

-0.1

Frequency (GHz)

0 20 40 60-180

-179

-178

-177

-176

Frequency (GHz)

0 20 40 60-34

-32

-30

-28

Frequency (GHz)

0 20 40 60-40

-30

-20

-10

0

Frequency (GHz)

Measured

Simulated

Measured

Simulated

Measured

Simulated

Measured

Simulated

dB (S

11)

dB (S

21)

Pha

se (S

11) (

deg)

Pha

se (S

21) (

deg)

Error AnalysisError Analysis

2 2 2 2( | 11|) ( 11) ( | 21|) ( 21)| 11| 11 | 21| 21

S S S SS S S Sε ε ε εε′ ′ ′ ′∂ ∂ ∂ ∂′Δ = Δ + Δ∠ + Δ + Δ∠

∂ ∂∠ ∂ ∂∠

2 2 2 2( | 11|) ( 11) ( | 21|) ( 21)| 11| 11 | 21| 21

S S S SS S S Sε ε ε εε′′ ′′ ′′ ′′∂ ∂ ∂ ∂′′Δ = Δ + Δ∠ + Δ + Δ∠

∂ ∂∠ ∂ ∂∠

2 2 2 2( | 11|) ( 11) ( | 21|) ( 21)| 11| 11 | 21| 21

S S S SS S S Sμ μ μ μμ′ ′ ′ ′∂ ∂ ∂ ∂′Δ = Δ + Δ∠ + Δ + Δ∠

∂ ∂∠ ∂ ∂∠

2 2 2 2( | 11|) ( 11) ( | 21|) ( 21)| 11| 11 | 21| 21

S S S SS S S Sμ μ μ μμ′′ ′′ ′′ ′′∂ ∂ ∂ ∂′′Δ = Δ + Δ∠ + Δ + Δ∠

∂ ∂∠ ∂ ∂∠

• The partial derivatives are computed numerically. For example, to calculate ∂ε’/∂|S11|, we evaluate ε’ with |S11|+δ while all other parameters remaining the same (δ: small number, selected to be 1E-4), then . The S-parameters uncertainties are provided by Agilent.

/ | 11| [ (| 11| ) (| 11|)] /S S Sε ε δ ε δ′ ′ ′∂ ∂ = + −

Error Analysis (contError Analysis (cont’’))

The systematic errors of ε” (less than +/- 15%) are much smaller than ε’ , μ’ and μ”.

ε” extracted from different data sets are fairly close (within +/-6.5%), but ε’ , μ’ and μ” change dramatically. Consistent with the above theoretical prediction.

Overestimation of errors ( 6.5%<15% ) is due to

• Worst-case S-parameter uncertainties

• Numerical partial derivative evaluations

Systematic errors are more sensitive on S11 uncertainties than S21uncertainties. Consistent with expectation.

• Most of energy is reflected. Change of material properties leads to larger difference on S21 than S11, in other words, little variation in S11 will cause dramatic change on ε and μ.

5 10 15 20 25 30 35 40 45 50500

1000

1500

Frequency (GHz)

ε′

5 10 15 20 25 30 35 40 45 500

2000

4000

6000

8000

Frequency (GHz)

ε′′

Intrinsic Permittivity of MWNTIntrinsic Permittivity of MWNT

• Bruggeman Effective Medium Theory (EMT) applied to remove the effect of air in MWNT paper and extract the intrinsic permittivity of MWNT

• ε’ and ε” are roughly doubled

0a eff b eff

a b

a eff b eff

f fK K

ε ε ε ε

ε ε ε ε

− −+ =

+ +

Composite with a & b

1 qK

q

−=

2

1 2

1/

1/ 2 /

aq

a a=

+

f’s are volume fractions

a’s are diameter & lengthof individual MWNT

THz Time Domain Spectrometer (TDS) SetupTHz Time Domain Spectrometer (TDS) Setup

• Femto-second laser excites a low-temp grown GaAs substrate

• THz pulse generated and transmitted to sample

• Same laser pulse optically delayed to detector: coherent measurement

• Transmission and reflection magnitudes & phases can be measured1. Z. Wu, L. Wang, A. Young, S. Seraphin, and H. Xin, J. App. Phys., Vol. 103, No. 9, May, 2008 (This article was also published in the June 2, 2008 issue of the Virtual Journal of Nanoscale Science & Technology (www.vjnano.org)

• Photoconductive THz-TDS system (50GHz-1.5THz)• Transmission Setup

– Sample-out as reference spectrum– Sample-in as sample spectrum

• Reflection Setup– Metal plate as reference object (reflection -1)– CNT sample and metal plate surfaces placed at the same

position (misalignment <12.7 μm)

– Incidence angle 26°

7

Detailed THz Experimental SetupDetailed THz Experimental Setup

• Large reflection from MWNT sample• Small transmitted signal through the sample• Higher order multiple reflections ignored

8

Metal PlateMWNT

Free SpaceMWNT

ReferenceMWNT Refl.MWNT Trans.

Frequency DomainTransmissionReflection

THzTHz--TDS ResultsTDS Results

' "z z jz= −

01( , 0)1i

zR zz

θ −≈ =

+

))1(2(exp2)1(

4)( 00 d/λnπjz

zn,zT −+

=2

2

0)/(sin1cos

)/(sin1cos),(

nz

nzznR

ii

ii

θθ

θθ

−+

−−=

' "n n jn= −Complex refractive index: Wave impedance:

First order complex transmission and reflection coefficients:

n and z numerically solved from T0 and R0

No need to assume n = 1 / z (or µ = 1)

If 0°, z and thus n can be analytically solved from T0 and R0iθ ≈

(Normal-incidence approximation)

/n zε = n zμ = ⋅Permeability:Complex permittivity:

9

Material Parameters Extraction AlgorithmMaterial Parameters Extraction Algorithm

Multiple branches of n’ solution due to the phase ambiguity in measured T0

10

• Good consistency between VNA and TDS measured n’ solution branches

• Analytically solving n’solutions with approximation gets close results

0iθ ≈ o

0 020

1 ( 2 )1 2

Tn mR d

λππ

′= − ∠ +−50 100 150 200 250 300 350

-500

-400

-300

-200

-100

0

100

200

300

Frequency (GHz)

Mul

tiple

bra

nche

s of

n′ s

olut

ions

50 100 150 200 250 300 350-500

-400

-300

-200

-100

0

100

200

300

Frequency (GHz)

Mul

tiple

bra

nche

s of

n′ s

olut

ions

MultipleMultiple--Branch Branch SoultionsSoultions of Index of Refractionof Index of Refraction

• m = 0 branch in TDS results is the physically correct branch

• The continuity of n’ at the boundary frequency is kept

• Good consistency of n’ and n”between VNA and TDS results

• Statistical error bars show good repeatability

11

Extracted nExtracted n’’ and nand n”” of MWNT Samplesof MWNT Samples

0

1000

2000

3000

4000

0 100 200 300 400

VNATHz-TDS

Frequency (GHz)

Extr

acte

d ε"

0

200

400

600

800

1000

0 100 200 300 400

VNATHz-TDS

Frequency (GHz)

Extr

acte

d ε'

-2

0

2

4

6

8

10

0 100 200 300 400

VNATHz-TDS

Frequency (GHz)

Extr

acte

d μ"

-4

-2

0

2

4

6

8

10

0 100 200 300 400

VNATHz-TDS

Frequency (GHz)

Extr

acte

d μ'

(a) (b)

(c) (d)

0

1000

2000

3000

4000

0 100 200 300 400

VNATHz-TDS

Frequency (GHz)

Extr

acte

d ε"

0

200

400

600

800

1000

0 100 200 300 400

VNATHz-TDS

Frequency (GHz)

Extr

acte

d ε'

-2

0

2

4

6

8

10

0 100 200 300 400

VNATHz-TDS

Frequency (GHz)

Extr

acte

d μ"

-4

-2

0

2

4

6

8

10

0 100 200 300 400

VNATHz-TDS

Frequency (GHz)

Extr

acte

d μ'

(a) (b)

(c) (d)

12

Extracted MWNT Permittivity and PermeabilityExtracted MWNT Permittivity and Permeability

2 21

2 21 1( )

p pc i i

ω ωε ε

ω ω ω ω ω= − +

− Γ − + Γ +

ε’ and ε” simultaneously fittedCovering microwave to THz

13

DrudeDrude--Lorentz Model FittingLorentz Model Fitting

0 50 100 150 200 250 300 350-25

-20

-15

-10

-5

0

5

10

15

Frequency (GHz)

Ext

ract

ed n′

Transmission and ReflectionTransmission onlyVNA measurement

0 50 100 150 200 250 300 3500

5

10

15

20

25

30

35

40

Frequency (GHz)

Ext

ract

ed n′′

Transmission and ReflectionTransmission onlyVNA measurement

0 50 100 150 200 250 300 350-25

-20

-15

-10

-5

0

5

10

15

Frequency (GHz)

Ext

ract

ed n′

Transmission and ReflectionTransmission onlyVNA measurement

0 50 100 150 200 250 300 3500

5

10

15

20

25

30

35

40

Frequency (GHz)

Ext

ract

ed n′′

Transmission and ReflectionTransmission onlyVNA measurement

))1(2(exp2)1(

4)( 00 d/λnπjn

nn,zT −+

=

With n = 1 / z (or µ = 1) assumption:

Solve complex n from T0

14

Transmission Measurement Only EvaluationTransmission Measurement Only Evaluation

21 exp( 4 " / )1

zLossFactor mag n fd cz

π⎡ ⎤−⎛ ⎞= −⎢ ⎥⎜ ⎟+⎝ ⎠⎢ ⎥⎣ ⎦

Due to one round trip ofinternal reflections in sample(Transmission/ Reflection)

Loss factor calculated with extracted n” and z

Higher order signal < 10% oflast order signal magnitude for~ 60GHz and up

Verified extraction algorithms

15

Multiple Reflections Effect in Sample Multiple Reflections Effect in Sample

Characterization of SWNT Thin Film on Substrates Characterization of SWNT Thin Film on Substrates

Consider CNT layer as a boundary providing surface current

Air(Medium 1)

SubMedium 2

Reflection coeff:

Transmission coeff:

s

s

YY

σσ

−

+

−Γ =

+

12

s

YTY σ+

=+

1 20

1, nY Y Y YZ Z± = ± = =With

SurfaceConductivity

SWNT Thin Film: ~ 300 nm thickness

3.2mm Pyrex glass

5mm thick window glass3.2mm thin window glass

Thick substrate necessary to avoid multiple reflections

Characterization of SWNT Thin Film on Substrates Characterization of SWNT Thin Film on Substrates

100 200 300 400 500 600 700-0.05

-0.04

-0.03

-0.02

-0.01

0

0.01

0.02

0.03

Freq, GHz

surfa

ce c

ondu

ctiv

ity, r

eal p

art

05200611

100 200 300 400 500 600 700-0.05

-0.04

-0.03

-0.02

-0.01

0

0.01

Freq, GHz

surf

ace

cond

uctiv

ity, i

mag

inar

y pa

rt

05200611

• More accurate real conductivity• Measured surface conductivity

– σ*t = 0.015 S (THz) and 0.0023 S (DC 4-point measurement)– σ ~ 105 Simens

• Peaks measured in conductivity– Systematic error / plasmonic resonances?– More study is under way

Under Study Under Study –– Single Tube MeasurementSingle Tube Measurement

Fabricate Grids on Insulating SubstrateFabricate Grids on Insulating Substrate Disperse CNT and Locate Using AFMDisperse CNT and Locate Using AFMEBL Fabrication of Designed Test FixtureEBL Fabrication of Designed Test Fixture

1 um1 um

Microwave Property Probing and ExtractionMicrowave Property Probing and Extraction

-90

-70

-50

-30

0 20 40 60 80 100

without Tubewith Tube

Frequency (GHz)

S21

(dB

)

-150

-100

-50

0

50

100

0 20 40 60 80 100

with Tubewithout Tube

S21

Phas

e (D

eg)

ConclusionsConclusions

• Carbon nanotube maybe useful for high frequency applications

• Bulk (paper / thin film) materials characterized (8 – 400 GHz)

• Intrinsic material properties (permittivity & permeability) extracted

• Microwave & THz data are consistent

• Drude-Lorentz model fits measured permittivity

• Correlation of single tube and ensemble results under way