These are the slides from the May Science Cafe featuring Dr. MVS Chandrashekhar. During this cafe he discussed his work with graphene a new, clean energy source.

1. Clean Energy Lab (CEL)
Towards Plasmonics in Epitaxial Graphene
M.V.S. Chandrashekhar
Department of Electrical and Computer Engineering,
University of South Carolina
USC CMU MPI/Pisa
G.Koley R. Feenstra U. Starke
T.S. Sudarshan N. Srivastava C. Colletti
C. Williams
J. Weidner
1
B.K. Daas
K.M. Daniels
S. Shetu
O. Sabih
A. Obe

2. Clean Energy Lab (CEL) @ USC
OUTLINE
•What is Graphene?
•Why Plasmonics?
• Viability of IR Plasmonics in EG on SiC
• Infrared carrier transport in EG/SiC
• Molecular doping studies using IR
•Interband processes
•Electrochemical Functionalization of EG
•Summary

3. WHAT IS GRAPHENE?
 Single atomic layer of graphitic carbon “discovered” in 2005-
Physics Nobel in 2010 Geim & Novoselov, U. Manchester
 Electrons behave like they have no mass-am I crazy?
 Strongest material known -space elevator E=1.25TPa
 Highest thermal conductivity in-plane
 It is all surfacesensitive to surroundings
 Very transparent and highly conductive-touch screens?

4. Clean Energy Lab (CEL) @ USC
WHAT IS A PLASMON POLARITON?
Polariton: Collective oscillation of electrons (Plasmon), generated by the
electromagnetic field that excites the metal/dielectric interface [1]. It is a near-field
phenomenon. Like waves in water.
Electromagnetic wave Electric or magnetic Dipole
Polariton
(Bosonic-quasiparticles)
Phonon-Polariton (IR photon + Optic phonon)
Exiciton-Polariton ( Visible light + exciton)
Intersubband Polarition (IR photon + intersubband-excition)
Surface plasmon-Polariton , SPP (Surface plasmons +light)
[1] W.L. Barnes, A.Dereux, T.W. Ebbesen, Nature 424 (2003) 824-830

5. Clean Energy Lab (CEL) @ USC
MOTIVATION: THE PLASMONIC CHIP
1. Overcome diffraction limit of light (d<λ/2) using SPP
2. Merge electronics and optics together in nano scaled range
3. Important for data processing, super lensing, sensing etc.
ωp2
ε m (ω ) = 1 − 2
Surface Plasmon Polariton at metal/dielectric interface ω
Whenε m<0, K is imaginary
Surface confinement
5
SPP
CHALLENGE: Couple Collective SPP to Single particle excitations
[2] M. Dragoman, D. Dragoman, Nanoelectronics: Principles and Devices, Artech House, Boston, 2006

6. Clean Energy Lab (CEL) @ USC
HOW DO PLASMONICS WORK?
•SPP propagation mediated by intra band processes
•SPP detection mediated by inter band processes
Graphene
e2
π h2 ∞ ∂f ( E − EF )
σ int ra ( ω ) = i ∫ dE ]E
i −∞ ∂E
ω+
τ
∆2
(1 + 2 )
e2 (ω + iΓ ) ∞ E
σ int er ( ω ) = i ∫∆ dE ( 2E ) 2 − ( hω + iΓ ) 2 [ f ( E − EF ) − f (− E − EF )]
π
Unlike a metal, there is significant interband conductivity even at low energies.
KEY: How to convert plasmon to e-h pair and vice versa?
-high speed computation
-new paradigm in plasmonic light sources

7. Clean Energy Lab (CEL) @ USC
SIC SUBSTRATE DIELECTRIC FUNCTION
ω 2 − ωLO + iΓ1ω
2
ε SiC = ε SiC (ω ) = ε ∞ 2
ω − ωTO + iΓ 2ω
2
WLO= Longitudinal optical phonon (972cm-1)
WTO= Transversal optical phonon (796cm-1)
ε
At high frequency SiC ~6.5 [8]
ε
At low frequency SiC ~9.52
ε (0) ωL
2
LST relation: = 2
ε (∞) ωT Negative dielectric function
n imaginary, damped wave gives
SPP surface confinement
SiC’s negative dielectric function in restrahlen band
 n is imaginary, damped wave
 confines SPP vertically
Role of metal and dielectric reversed.
[8] Dmitriy Korobkin, Yaroslav Urzhumov, and Gennady Shvets; J. Opt. Soc. Am. B, 23,3,468 (2006)

8. Clean Energy Lab (CEL) @ USC
Viability of Plasmonics in EG on SiC
TM modes are found by assuming that the electric field
has the form as..
When x>0 Ex = Beiqz −Q x and Ez = Aeiqz −Q x E y = 0
1 1
When x<0 Ex = Deiqz +Q x and Ez = Ceiqz +Q x
2 2
Ey = 0
Dispersion relation for TM mode is given by
ε1 ε2 σ (ω , q)i
+ =
ε1ω 2 ε 2ω 2 ωε 0
q − 2
2
q − 2
2
c c
Assuming we are in low q, so q<w/c, SPP
dispersion relation is.
ω2 1
q = 2 [1 −
2
]
c σ (ω , q )
( + ε2 ) 2
450
ε 0c
ω
Free space dispersion relation is q =
c
Fig: SPP dispersion relation plot with free space dispersion
8
SPP dispersion intersects the free space dispersion -coupling of
SPP into free space radiation- SiC substrate essential.

9. Clean Energy Lab (CEL) @ USC
Viability of Plasmonics in Epitaxial Graphene
q= wave vector
Coupling between SPP and Single Particle Excitations ω= frequency
•Intersection between SPP and free space ω1 = vF q
•Coupling to free space
•Intersection region has to be dominated by
interband scattering
•Energy to create e-h pairs, not heat
•SPP detection
•Potential for tuning this process
•Change Ef by gating to suppress e-h
•SPP guiding.
ω2 = 0 q < 2k F ω2 = γ q − 2 EF q > 2 k F
Applying single particle excitation boundary
condition for intra and inter band scattering
Comes from graphene E-k bands 9
(developed by S.Das Sarma)

10. Clean Energy Lab (CEL) @ USC
MODULATING EPITAXIAL GRAPHENE
PLASMON WAVEGUIDE BY DOPING
‘OFF’: When Ef is low, only ‘ON’: When Ef is high,
interband transitions allowed. interband transitions not
Can transform plasmon to DC allowed. Can propagate signal
current and vice-versa. without significant damping.
Electrical manipulation of
plasmonic signals.

11. Clean Energy Lab (CEL) @ USC
Graphene
Exfoliated graphene Epitaxial graphene
( single layer) (single or multi layer)
Silicon (Si) GaAs 4H-SiC Metal Graphene
(Ag)
Supporting TE --- --- ---- No Yes [2]
mode
Dispersion Parabolic parabolic parabolic parabolic linear –EHP at
relation
any wavelength
Band gap 1.12eV 1.42eV 3.23eV 0 0
Electron Mobility <1400 <8500 <900 200000
(cm2/v-s)
RMS roughness --- ---- ------- ~1nm <0.5nm
SPP Detection ----- ------ -------- Metal to Single material
and guiding
guide, for guiding and
materials 11
Semi to detection,
detect
[3] L A Falkovsky “Optical properties of graphene” . Phys.: Conf. Ser., Volume 129, Number 1 (2008)
[4] M.Jablan, H. buljan, M. Soljacic “Plasmonics in Graphene at infrared frequencies” Phy.ReV. B 80 245435 (2009)

12. Clean Energy Lab (CEL) @ USC
Epitaxial Graphene Growth
Raman XPS & ARPES
6H-SiC Graphene
A
B
D peak (1345 cm-1)…..due to induced
disorder
C
A G peak (1585cm-1)… due to in plane
vibration
C
B
2D peak (2670cm-1)…..due to double
A
resonant process
A B C
FiG: Realization of Graphene from 6H-SiC ID/IG…Disorder ratio <0.2 [5]
12
[5] A.C Ferrari and J. Robertson “Interpretation of Raman spectra of disordered and amorphous carbon” Phys. Rev B 61 vol 61 num 20 (2000)
[6] P.J.Cumpson; “The Thickogram: a method for easy film thickness measurement in XPS”Surf.Interface.Anal,29,403 (2000)

13. NON-POLAR FACE GROWTH-6H SIC
EG on Si face EG on C face
5µm× 5µm×
5µm 5µm
What Growth
Growth mechanism is
happens
mechanism is defect&step
in
step flow mediated [**]
between?
mediated [*]
[*] M. Hupalo, E. Conrad, M. C. Tringides http://arxiv.org/abs/0809.3619
[**] Appl. Phys. Lett. 96, 222103 (2010)

14. Clean Energy Lab (CEL) @
USC 13000C 13500C 14000C 14500C
Si face
A plane
M plane
C face

15. Clean Energy Lab (CEL) @
USC
Raman Characterization
Si face
C face
All peaks are red shifted with increasing temp.
What would a H2 etch do? Decreasing stress with temperature increase
2D peaks narrow with increasing temperature

16. Clean Energy Lab (CEL) @ USC
Surface Plasmon Polariton (SPP) in Epitaxial Graphene
Our approach
Mathematical Model [7]
Experiment:
Blank SiC is used as reference. ω 2 − ωLO + iΓ1ω
2
ε 2 = ε 2(ω ) = ε ∞ 2
ω − ωTO + iΓ 2ω
2
∆2
(1 +)
e 2 (ω + iΓ ) ∞ E2
σ int er ( ω ) = i ∫∆ dE ( 2 E ) 2 − ( hω + iΓ ) 2 [ f ( E − EF ) − f (− E − EF )]
π
e2
∂f ( E − EF )
σ int ra ( ω ) = i π h
2 ∞
i ∫
−∞
dE
∂E
]E
ω+
τ
2
ε1Nσ (ω) ×cos(Φ
( ε1ε 2ε 0 / α + ) c
1)
−ε1ε 0
Fig: Schematic view of FTIR differential R= 2
reflection spectra setup ε1Nσ (ω) ×cos(Φ
( ε1ε 2ε 0 / α + ) c
1)
+ε1ε 0
n1
1 − [( sin Φ1)]2
α= n2
cos Φ1 16
[7] T. Stauber, N.M.R Peres, A.K. Geim; “Optical conductivity of graphene in the visible region of the spectrum”Phy.Rev. B 78 085432 (2008)

17. Clean Energy Lab (CEL) @ USC
Surface Plasmon Polariton (SPP) in Epitaxial Graphene….(Cont.)
Results of developed mathematical model
Fig: Variation of number of layer Fig: Variation of Fermi level
2
ε1Nσ (ω) ×cos(Φ
( )
ε1ε 2ε 0 / α +
c
1)
−ε1ε 0
R= 2
ε1Nσ (ω) ×cos(Φ
( )
ε1ε 2ε 0 / α +
c
1)
+ε1ε 0
Variable Parameter
Number of Layer, N
Fermi Energy Ef
17
Scattering time τ
Fig: Variation of scattering time

18. Clean Energy Lab (CEL) @ USC
Surface Plasmon Polariton (SPP) in EG/SiC interface
Experimental results from FTIR: Evidence of SPP at EG/SiC interface
Fig: AFM image of SiC Substrate Fig: IR reflection of SiC Substrate with SiC as reference
ωLO
18
ωTO
Fig: AFM image of EG (2ML)on SiC Fig: IR reflection of EG with SiC as reference

19. Clean Energy Lab (CEL) @ USC
EG transport properties extraction using FTIR
Extracted Parameters:
•No of Layer N=2-17
•Fermi Energy Ef=10535meV
•Scattering time, τ=4-17fs
Interband broadening is assumed
constant=10meV i.e. only intraband
scattering considered.
Extracted No of layer matches well with
XPS measurements.
Fig: IR reflection measurement and mathematical
model are consistent

20. Clean Energy Lab (CEL) @ USC
EG transport properties extraction using FTIR
B,K. Daas…MVS et al JAP (2012)
∞
Carrier density ns = ∫ D( E ) f ( E − EF )dE
0
D ( E ) = 2 E / π ( hv F ) 2
Fig: Fermi level Vs No of layer
1
τ = k1( ) / vF
ns
1
Short range scattering[9] τ∝
ns
Fig: Scattering time Vs avg. carrier density
Coulomb scattering[9] τ ∝ ns
Mobility, µ= eτ vF / EF
2
20
Fitting value of k1=0.6 suggests our EG is
Mobility (1000-10,000) cm /V-s 2
dominated by short-range scattering.
[9] L A Falkovsky “Optical properties of graphene” . Phys.: Conf. Ser., Volume 129, Number 1 (2008)

21. CORRELATION WITH ULTRAFAST
SPECTROSCOPY OF EPITAXIAL
GRAPHENE
If states are occupied by pump,
probe signal will not be absorbed,
transmission increases
 85fs, ~10nJ 785nm laser, pump &probe
 Measures ENERGY relaxation time, not momentum
 τenergy>>τmomentum, supports short range scattering

22. THZ PROBE, OPTICAL PUMP
 Non-linear power dependence, quadratic fit works
well-intervalley phonon scattering & Auger dominate
 Explains full behavior, withτrec~200fs , B~1-3cm2/s

23. MOLECULAR DOPING OF EG-LONG
Clean Energy Lab (CEL) @ USC
RANGE?
Mirror
Collecting Incoming
•Pure N2 - inert gas
light light
signal source
•15ppm NO2 -electron accepting gas
•500ppmNH3 -electron donating gas
Sensing
element
Graphene
SiC Substrate SPP
Graphene
Fig: Experimental setup
Findings:
Reflection amplitude changes
-Looks like change of thickness
but thickness can’t change 23

24. Clean Energy Lab (CEL) @ USC
Conductivity Matching:
Optical Conductivity:
∆2
(1 + 2 )
e2 (ω + iΓ ) ∞ E
σ int er ( ω ) = i ∫∆ ( 2E ) 2 − ( hω + iΓ ) 2 [ f ( E − EF ) − f (− E − EF )]
dE
π
e2
σ int ra ( ω ) = i π h 2 ∞ dE ∂f ( E − EF ) ] E
i ∫−∞ ∂E
ω+
RPA approximation: τ
e2 ns n F [4rs / (2 − π rs )]
σ RPA
T =0 = [ + i ]
π h ni G[4rs / (2 − π rs )] 4ns
Fig: Dielectric function of SiC
Intraband-low f Interband high f
2π
e2 2 2π
sin θ
2 x2 (1 − cos θ ) 2
rs =
4πε 0ε SiC vF h
G ( x) =
x
8 ∫ θ
dθ F ( x) =
8 ∫ θ
dθ
0 (sin + x) 2 0 (sin + x) 2
2 2
Here, Γ=h/2πτintra is not taken as constant but is allowed to vary.
This is needed to get a good fit to the data
Interband scattering
Extracted parameter ni
matters even at DC.

25. Clean Energy Lab (CEL) @ USC
C-FACE IR REFLECTIVITY
• Adsorbed molecules transfer
charge  charged scatterers
• As ni increases, inter/intra
band scattering increase
• τ ~1/n i.e.
i,
conductivity decreases
• Assume each ni is an
adsorbed molecule
• From ΔEf, we can extract
carriers induced, n, using
D(E)
• 0.01e charge donated by
each NO2 molecule
Agrees with Kelvin probe
measurements

26. Clean Energy Lab (CEL) @ USC
No of Gas Fermi ni/ML Intra band Avg. Inter band
Layer level (cm-2) scattering scattering
(meV) time (fs) time(fs)
34 N2 25 2x1011 90-280 185 27-60
NH3 30 6x1012 60-90 75 1.6-2
NO2 35 2x1013 2-9 5 0.3-0.5
22 N2 45 3x1011 10-17 14 9-17
NH3 65 7.5x1012 2-9 5.5 0.2-2
NO2 95 6x1013 0.9 0.9 0.1-0.2
9 N2 70 5.1x1011 10-20 15 3-4
NH3 90 5.5x1013 0.8-1 0.9 0.2-0.5
NO2 120 1.5x1014 0.4-0.5 0.45 0.1-0.3

27. CORRELATION WITH ‘DC’
MEASUREMENTS
4ppm
 NO2 makes the C-face more p-type
 Implied δp~1012-13cm-2 -is this possible?
M. Qazi….MVS, Koley et al., Appl. Phys. Exp., 3, 075101 (2010)

28. CORRELATION WITH KELVIN
PROBE
~60% or more change in conductivity expected
Scattering from impurities not enough to explain
measured change in optical conductivity
Electron affinity of NO2 dominates!
 Consistent with F.Schedin’s result of G/SiO2
 Assume ΔEf~10meV for 4ppm. μchem ill-defined.

29. Clean Energy Lab (CEL) @ USC
No of Gas Fermi ni/ML Intra band Avg. Inter band
Layer level (cm-2) scattering scattering
(meV) time (fs) time(fs)
34 N2 25 2x1011 90-280 185 27-60
NH3 30 6x1012 60-90 75 1.6-2 From FTIR
NO2 35 2x1013 2-9 5 0.3-0.5
22 N2 45 3x1011 10-17 14 9-17
NH3 65 7.5x1012 2-9 5.5 0.2-2
NO2 95 6x1013 0.9 0.9 0.1-0.2
9 N2 70 5.1x1011 100-200 150 3-4
NH3 90 5.5x1013 0.8-1 0.9 0.2-0.5
NO2 120 1.5x1014 0.4-0.5 0.45 0.1-0.3
 From ΔEf, we know δp(n)
 Assume each ni is an NO2 molecule
 So, each NO2 molecule donates δp/ni ~1%e for all
thicknesses-same as SKPM!
 ~(ΔEf/ΔSWF)2~0.3-2%e over various samples.
 ni decrease with thickness-diffusion in C-face?
 NOTE: interband broadening as large as 1eV!

30. REMEMBER PLASMONICS?
 If interband broadening is large, even metallic
graphene plasmons will be damped, must control.
 Periodic structures enable tuning using localized
plasmons-enable conversion of plasmon to e-h pair

31. SUMMARY FOR PART I
 Plasmonic devices possible on EG/SiC
 How clean is as-grown EG?
 Gaseous molecular doping useful for transport
studies over wide energy range near K-point.
 For FET’s, interband scattering could be
important at high carrier concentration, even at
DC. May influence realizing plasmonics.
 Will we be able to convert SPP into e-h pair in
controllable fashion?

32. PART II:
FUNCTIONALIZATION

33. ELECTROCHEMICAL
FUNCTIONALIZATION-SI FACE
RMS: 0.57nm
Scale:
8nm
Before
RMS: 1.00nm
Scale:
8nm
After
 H+ attracted to graphene cathode 1V, 1hr.
 Can it react? V<1.2V, H2 formation potential
 Goal: Bandgap in diamond-like graphanes.

34. FUNCTIONALIZATION BY RAMAN
SPECTROSCOPY
 Single monolayer of graphene is more reactive than bulk
graphite
 Up to ten times more reactive than bi-layer and multilayer graphene
 Substrate
enhanced electron transfer
 Emergence of D-peak indicates reaction in graphene
1200 D-peak red-shifts 1354-1335
cm-1.
1000
Raman Intensity (arb. units)
G peak broadens and
800 slightly blue shifts ~3 cm-1
New peak at ~2930
600
2 Indicative of C-
400 Hbond
G GraphaneD
200 D
Graphene
0
1200 1600 2000 2400 2800
-1
Wavenumber (cm )
34
• R. Sharma, et. al. Anomalously Large Reactivity of Single Graphene Layers and Edges toward Electron Transfer Chemistries, Nano Letters 10, 398-405 (2010)

35. H-FUNCTIONALIZATION SHOWN BY RAMAN
SLOPE
 Increasing photoluminescence
background
 Increasing hydrogen content
 Ratio between slope m of the
linear background and the
intensity of the G peak
D peak  m/I(G)
Raman Intensity
 Measure of the bonded H content
G peak
 Based on amourphous carbon
S≈ 18µm
results
Wavenumber
(cm-1)
 maybe dominated by grain
Florescence is not seen in boundaries
carbon only hydrocarbons!!!
•B. Marchon, et.al. Photoluminescence and Raman Spectroscopy in Hydrogenated Carbon Films. IEEE Transactions on Magnetics, Vol. 33, NO. 5, Sept. 1997.

36. FLUORESCENCE BACKGROUND TO ESTIMATE
H-CONTENT
Damage distinguished from functionalization by a) damage has
unmesurable slope for a given D/G ratio b) D peak position
36

37. SUBSTRATE DEPENDENCE OF
FUNCTIONALIZATION
Table 1: Average Parameters From Each Substrate in Study
Substrate D-peak D-peak D/G D/G Normalized Normalized
Position Position Ratio Ratio Slope Slope
Before After Before After Before (µm) After(μm)
(cm-1) (cm-1)
SI(1°) 1348 1330 0.21 1.91 3.66 14.4
SI2(on) 1344 1332 0.17 1.32 4.24 18.9
SI3(0.5) 1347 1331 0.13 0.6 3.93 4.42
* All substrate averages contain at least three samples
• Substrate Limited Functionalization
– Possible Causes
• Off-cut angle
• Substrate Resistivity
• Residual Damage in Graphene
 Problem: Issue with conversion control?
 Solution: Enhance reactivity with metal? 37

38. RAMAN SPECTRA OF
FUNCTIONALIZATION WITH AND
WITHOUT PT NANOPARTICLES
 Chemically Deposited • Raman Shows:
Platinum – Incredibly large D/G ratio~4.5 38
– Emergence of Fluorescence
 H2PtCl6 · 6H2O + DI water – Addition to D’ shoulder peak
– C-H peak at ~2930

39. RESULTS OF EVAPORATED METAL
CATALYSIS FUNCTIONALIZATION
Increased reactivity seen in Au and Pt enhanced conversions
 D/G ratio>1.0 for Au and Pt
 Fluorescence> Noise Threshold (5 µm) 39

40. SUMMARY: METAL CATALYSIS
D Position D Position ID/IG ID/IG Normalized Normalized
Before After Ratio Ratio Slope Slope
(cm-1) (cm-1) Before After Before (µm) After (µm)
SI 1348 1330 0.21 1.91 3.66 14.4
SI2 1344 1332 0.17 1.32 4.24 18.9
SI3 1347 1331 0.13 0.6 3.93 4.42
SI3 Au
Avg 1342 1330 0.22 1.05 4.42 7.86
SI3 Pt
Avg 1364 1330 0.086 1.24 3.81 17.69
Increased functionalization with metal catalyst
40
Increase in fluorescence  bandgap?

41. SCANNING TUNNELING
SPECTROSCOPY
K.M. Daniels, …MVS, R. Feenstra… et.al, presented at EMC2011
accepted, JAP
 Evidence of localized states
functionalized
unfunctionalized
*8x8mm
More evidence required to distinguish from damage
What are these states?
41

42. CYCLIC VOLTAMMETRY
 Clear substrate dependence
 Qualitatively different from bulk carbon
 Clear peaks, not double-layer charging
 Still investigating peak assignments

43. SUMMARY OF PART II
 Electrochemical functionalization possible.
 Evidence for hydrogen incorporation
 More clarification needed
 Functionalization is substrate dependent
 Metal catalysts enhance functionalization
 Evidence for localized states by STS

44. MASTER SUMMARY
 Plasmonics in EG proposed
 IR transport studies with molecular dopants
 Electrochemical functionalization of EG
 Evidence of localized states
We also gratefully acknowledge the
Southeastern Center for EE Education for support of this work