Electron Transport in Strongly Coupled Molecular Electronic Junctions Richard McCreery, Adam Bergren, Sergio Jimenez Bryan Szeto, Jie Ru, Andriy Kovalenko, Stan Stoyanov University of Alberta National Institute for Nanotechnology
Electron Transport in Strongly Coupled MolecularElectronic Junctions
Richard McCreery, Adam Bergren, Sergio JimenezBryan Szeto, Jie
Ru, Andriy
Kovalenko, Stan Stoyanov
University of AlbertaNational Institute for Nanotechnology
National Institute for Nanotechnology
University of Alberta,Edmonton, Alberta,Canada
Founded 2002Building dedicated 2006
$$$:
National Research CouncilNSERC (Canada)Alberta Ingenuity FundCanada Fund for Innovation
“Molecular electronics”“molecularjunctions:”
Today’s question: How are electrons transported through molecules ??
Note:• molecules become circuit elements• critical dimension is 1-10 nm
E=0 V
E=-0.7
E vs. NHE
Electrochemical reduction:
+0.5
-0.5
0
distance from electrode
LUMO
HOMO
-1.0
+1.0
ionic double layer
+
-
+
-+
-
+
-
+
-+
--
-
-
+
+
molecule insolution
metal electrode
ener
gy re
lativ
e to
vac
uum
, eV
-7
-6
-4
-5
distance
metalliccontacts
Efermi
Molecular junction:
two electrodes, no doublelayer, no solution
ener
gy re
lativ
e to
vac
uum
, eV
Two common electron transport models:
“off-resonant”
e.g. tunneling, Schottky,field emission
-7
-6
-4
-5
distance
metalliccontacts
Efermi
“resonant”
e.g. “resonant tunneling”,“orbital mediated tunneling”
-7
-6
-4
-5
metalliccontacts
LUMO is close (~within kT)to Fermi level
HOMO
Efermi
-+ e-
LUMO
HOMO
-+ e-
The scientific question:
How are electrons transported through 1-5 nm thick molecular layers?
Outline: • fabrication of molecular junctions• characterization• electronic properties• transport mechanism
V
Cu
Au
Things we do differently fromeveryone else:
sp2
carbonvery flat (< 0.5 nm rms)graphitic carbon substrate[Pyrolyzed Photoresist Film, PPF,essentially metallic, with ρ=0.006 Ω-cm]
covalent C-C surfacebond, stable to > 500 oC
conjugated, partially orderedmono-
or multilayer, 1-5 nm thick,108
– 1012
molecules in parallel
slow electron beam deposition ofCu top contact, often covalentlybonded to molecule
Phys. Chem. Chem. Phys, 2006, 8, 2572J. Chem. Phys. 2007, 126, 024704
next slide
J. Phys. Cond. Matter, 20, 374117 (2008)
1 µmmolecular layer (not resolved)
SEM TEM
20 nm
polypyrrole
carbon
metal
EMmount
PPF Echip
4’’
wafer PPF Echip
Clip electrodeused for electro-chemistry
PPF leads
Junction
Microfabricated
“E-chips”
cut aftermoleculedeposition
Bryan SzetoJie Runext slide
Au strip
500 µm
1 mm
Junction area:
2.5 x 2.5 µm to400 x 400 µm
V
Vsense
+-
current amplifier
Cross section of a PPF/NAB/Cu/Au junction (SEM)
Molecules (not resolved)
Si
PPF
SiO2
SiNCu/Au
Left side Right side
Si
SiO2
PPFMolecules (not resolved)
Cu/Au SiN
SiO2
Si
PPFmolecules
Cu/AuSiN SiNSiO2 SiO2
Schematic of junction structure:
molecular layer is really thin compared to metals,does it survive metal deposition??
Cu/Au
PPF
SiN/SiO2
to scale:
1-5 nm
> 100 nm
Au 30nm
Cu 120nmSiN
70nmSiO2 50nm
PPF 1µm
(not to scale)
~4 nm NAB
~10 nm PPF
Quartz substrate(0.13 mm)
514.5 nm laser
excitation at
45°Collect Raman
scattered light
normal to
substrate
Au
CuNO2
N=N
(~ 4 nm thickmultilayer)
“backside”
Raman of buried interface:
~50% transparent
Quartz/PPF/NAB/Cu/Au(after metal deposition)
Quartz/PPF/NAB, no
Cu or Au
3000
5000
7000
9000
11000
13000
15000
17000
600 1100 1600
Intensity (counts/30 s)
Raman Shift (cm-1)
Ram
an in
tens
ity (3
0 se
c, 1
9 m
W)
NO2
N=N
Adam BergrenAmr
Mahmoud
(Monday, 4:20 PM)
NAB on Au/Ti
Au/Ti NAB
AuTi
-log
(R/R
o)
wavenumber, cm-1
FTIR of buried interface:
Ti is “primer layer”
forNAB bonding to Au
wavenumber, cm-1
-log
(R/R
o) Au
IR transparent Si
IR beam
after 100 nm Au deposition:
Adam BergrenAmr
Mahmoud
(Monday, 4:20 PM)
carbon/molecule
V
Vsense
+-
current amplifier
Cu
Au
= fluorene
carbon on SiO2
12 nm
20 nm
1.7 -
2 nm
2 μm
V
Electronic behavior:
Labview
-25
-20
-15
-10
-5
0
5
10
15
20
-0.2 -0.1 0 0.1 0.2
ControlFL-SiO2
PPF/SiO2
/Cu
PPF/Cu slope= 1.6 Ω
curr
ent d
ensi
ty, J
, A/c
m2 V
SiO2
Cu
10 nm
V, carbon relative to Au
-0.002
-0.0015
-0.001
-0.0005
0
0.0005
0.001
0.0015
0.002
-4 -3 -2 -1 0 1 2 3 4
J, A
/cm
breakdown~ 3 MV/cm
slope= 420,000 Ω
Start with something familiar:
VPPF
Cu
SiO2
Au
-3
-2
-1
0
1
2
3
-2.4 -1.8 -1.2 -0.6 0 0.6 1.2 1.8 2.4
NBP film muchmore conducting than SiO2
How about a molecule instead of SiO2
?
NO2
nitrobiphenyl multilayer4.6 nm thick:
Cu
NO2
J, A
/cm
2
PPF
4.6 nm
V, carbon relative to Au or Cu
NO2
-3
-2
-1
0
1
2
3
-2.4 -1.8 -1.2 -0.6 0 0.6 1.2 1.8 2.4
V
J, A
/cm
2
NBP4.5
FL-SiO2
NBP 2.8 nm
NBP 1.6
Nitrobiphenyl junctions of differing thickness: V
NBP
Au
4.6 nm2.8 nm1.6 nm
SiO2rsd
5-15%yield > 90%
typical errorbar
7 A/cm2
~ 106
e-/sec/molecule
J. Phys. Chem. B, 2005, 109, 11163
• No obvious shape change from 1.6 to 4.6 nm thickness
• Symmetric with minimal hysteresis
• Repeatable > 108
cycles
-6
-5
-4
-3
-2
-1
0
1
2
3
-1 .5 -0 .5 0 .5 1 .5
V , V
ln J
, A/c
m2
V
Same data, on a log scale:NBP(1.6 nm)
NBP(2.8 nm)
NBP(4.6 nm)
ln(J, 0.1 V)
-16
-13
-10
-7
Ln I
(V=0
.1 V
), A
1 2 3 4 5d= thickness, nm
β
= 0.22 Å-1
NO2
β
= 0.21 Å-1
J = B e-βd
Literature: β• alkanes
(echem
or junctions) 0.8 Å-1
• aromatic (echem, 1999) 0.22• conjugated (echem, SAM)
0.3 to 0.6
• polyene* (2005) 0.22• oligothiophene* (2008) 0.1• oligoporphyrin* (2008) 0.04• oligophenyleneimine+
(2008) 0.3
* single molecule junction+ ~100 molecule junction
d
β
is smaller for aromatic structures(i.e. conjugated molecules are
better “conductors”)
‐15‐10‐5051015202530
‐1 ‐0.5 0 0.5 1V (PPF relative to Au)
j(A cm
‐2)
no molecule
(2.0 nm)(1.4 nm)
(1.5 nm)
NO2
(3.7 nm multilayer)
(1.9 nmmultilayer)
NO2
strong effectof structure and
thickness on conduction
Is thisbehaviormolecular?
Adam Bergren, J Phys. Cond. Matt. 2008, 20, 374117
Various transport mechanisms:
Field emission (Fowler Nordheim)
Incoherent, diffusive tunneling
Coherent tunneling, "superexchange"
Weakly Temperature dependent: distance dependence:
exp(-
β
d)
exp(-
β’d)
(V/d) exp(-
a d)
“hopping”, including redox exchange(Marcus-Levich)
Poole-Frenkel effect (“coulombic traps")
Thermionic (Schottky) emissionStrongly Temperature dependent (“activated”):
d-1
exp (-c d1/2)
exp (-c’d1/2)
400 K 325 K250 K
100,120,150 K
-0.2 0.0 0.2 0.4V
-3.0
-2.0
-1.0
0.0
1.0
2.0
3.0
-0.4
J, A
/cm
2
j (0.2 V)
J. Phys. Cond. Matter, 20, 374117 (2008)
“activated”
temperatureindependent
A good probe of mechanism:Temperature dependence
‐4.5
‐4
‐3.5
‐3
‐2.5
‐2
‐1.5
‐1
‐0.5
0
0 50 100 150 200 250
ln j 0
.2 V (A
cm
‐2)
1000 T‐1 (K‐1)
BP(1.4) FL(1.8) NAB(3.3) AB(3.2)
5 K10 K20 K40 K
Arrhenius plotsArrhenius plots
AJB 25
(1.8 nm)
(1.4)
NO2
N=N
(3.3 nm)
‐4.5‐4
‐3.5‐3
‐2.5‐2
‐1.5‐1
‐0.50
0 5 10 15 20 25
1000 T ‐1 (K‐1)
100 K200 K
26 µeV
56 µeV
50 µeVNO2
N=N
46 meV
31
137(AB)
40 meV(NAB)
‐10
‐5
0
5
10
15
‐1.5 ‐1 ‐0.5 0 0.5 1 1.5
V
j (A/cm
2 )
(4.5 nm multilayer)
NO2
N=N
T = 5 K
‐0.01
‐0.005
0
0.005
0.01
‐0.1 ‐0.05 0 0.05 0.1
V
j (A/cm
2 )
need to explain:
• apparent ohmic
contact• T-independent, despite:• 4.5 nm thick (too thick
for tunneling)
(T independent, 5 –
250 K)
NAB experimental (2.6 nm)
-100-80
-60-40-20
0
2040
6080
100
-1 -0.5 0 0.5 1
Applied Voltage (V)
J (A
/cm
2) PPF/NAB (2.6nm )/Cu
Simmons withimage charge
Simmons, φ
= 0.85 eV(i.e. tunneling through arectangular barrier)
Field emission (Fowler Nordheim)
All common off-resonance tunneling mechanisms fail:
-2
-7
-6
-4
-3
-5
Simmons, J. G. Journal of Applied Physics (1963), 34, 1793‐1803
m* = effective electron mass, where mass of charge carrier = m* x 9.1 x 10‐31
kg
Experimental data collected at 5 or 6 K
( )⎪⎭
⎪⎬⎫
⎪⎩
⎪⎨⎧
⎟⎟⎠
⎞⎜⎜⎝
⎛⎟⎠⎞
⎜⎝⎛ +−×⎟
⎠⎞
⎜⎝⎛ +−
⎥⎥⎦
⎤
⎢⎢⎣
⎡⎟⎠⎞
⎜⎝⎛ −−×⎟
⎠⎞
⎜⎝⎛ −= dqVmqVdqVmqV
dqJ
2*22exp
22*22exp
24 22 φφφφπ ηηη
Tunneling with reduced electron mass (modified Simmons equation):
Φ
= 1.1 eVm* = 0.4 me
‐3
‐2
‐1
0
1
2
3
4
‐1.5 ‐1 ‐0.5 0 0.5 1 1.5
NAB(3.3), experiment
Simmons Modeld= 3.3 nm
“off resonant”
tunneling just doesn’t work
‐0.3
‐0.2
‐0.1
0
0.1
0.2
0.3
0.4
‐1.5 ‐1 ‐0.5 0 0.5 1 1.5
NAB(4.5), experiment
Simmons Model
Φ
= 1.1 eVm* = 0.4 me
d= 4.5 nm
*experimental, from Kelvin probe
-2
-6
-4
-3
-5
-7
NAB LUMO (-3.0 eV
, from DFT)
HOMO (-6.6 , DFT)
Cu (-4.7)*PPF(-4.9)*
eVvs
vacu
um
we expect these levels tobe broadened, by:
• electronic coupling to substrate• intermolecular interactions• variable bonding geometry• uncertainty (i.e. lifetime) broadening
NO2
N=N
0.02
0.04
0.06
200 300 400 500 600 700
NAB in hexane (X.02)
wavelength, nm
Opt
ical
Abs
orba
nce
Some evidence for broadening:
NAB bonded to carbon
Appl. Spectros. 2007, 61, 1246-1253
NO2
N=N
An alternative approach:
NAB
-6
-4
-2
E, eV
DFT with periodic boundary conditions for graphene
Stan StoyanovKirill
KoshelevAndriy
KovalenkoNINT
HOMO
LUMO
NAB-graphene
HOMO and LUMO energies vary with torsion angle
Modeling of both contacts and molecule
Koshelev
-6
-4
-2
C_NAB_Au
E , eV
HOMO
LUMO
C_NAB_Cu strong interactionof both Cu andgraphene
with NAB
ener
gy re
lativ
e to
vac
uum
, eV
a range of orbital energies
-2
-7
-6
-4
-3
-5
distance
LUMOs
HOMOs
filled statesin metal
Efermi
Zero bias: positive bias:
+-
next slide
V=0
Note that more HOMOsbecome accessible for
higher bias, causing upwardcurvature
e-
HOMO fills again fromnegative electrode,
effectively “hole transport”
HOMOs
once + bias createsempty metal orbitals,electrons can leave
HOMO
e-
+ bias
+
-
2.6 nm
more HOMOs
becomingaccessible with increased bias
NO2
N=N
-70
-50
-30
-10
10
30
50
70
90
-1 -0.5 0 0.5 1Applied Voltage (V)
J (A
/cm
2)
observedgaussian
distributionσ
= 0.52 eVEf
– EHOMO
= 1.7 eVNchan
= 105
sech
distributionσ
= 0.31 eVEf
– EHOMO
= 1.7 eVNchan
= 105
σ
= half width of orbital energy distributionEf
– EHOMO
= Fermi level to orbital offsetNchan
= total number of active channels
Sergio JimenezAdam Bergren
PPF/NAB (2.6 nm)/Cu
-5-4-3-2-1012345
-1 -0.5 0 0.5 1
Applied Voltage (V)
Ln (J
)
NO2
N=N
• Ef
– EHOMO• HOMO “linewidth”
(σ)• and number of channels (N)• molecular layer thickness (d)
Main parameters of the model:
sech
observed
gaussian
for the electrochemists:NOT Marcus/Butler-Volmer; similarity due to distributionof orbital energies ratherthan thermal fluctuations
Important notes:
HOMO
LUMO
Ef
ener
gy
σ
EF
- EHOMO
• broadening caused by coupling and localenvironment, not thermal fluctuations
• main parameters are distribution width (σ),energy offset (EF
– EHOMO
), and thickness
• overlap of metal and molecule orbitals may(and probably does) occur at zero bias
• depending on offset between molecular orbitals andFermi level, we can greatly vary conductance
The punch line: strong interactions between molecule and contactsresult in resonant
electron transport rather than classical tunneling
Adam Bergren (NRC)Sergio Jimenez (visit. prof.)Andriy
KovalenkoStan StoyanovKirill
KoshelevJie Ru (Uof
Alberta)Bryan Szeto
Also :
Rory Chisholm (2:00 PM Monday)Mark McDermott
NINT