New Directions in Molecular Electronics Research

1
New Directions in Molecular Electronics Research
John C. Bean
UVA research contract history (because in
academia, the money has gotta come first!) 8/02
- 8/06 NSF-NIRT (Nanoscience Interdisciplinary
Research Team) "Merged CMOS/Molecular
Integrated Circuit Fabrication" Lloyd Harriott
(PI), John Bean, Matt Neurock (ChE), Mircea Stan,
Lin Pu (Ch) 8/04 -11/07 DARPA -
MOLEapps "Hybrid Mole Computer Using Vapor
Phase Assembly" Lloyd Harriott (PI), John Bean,
Mircea Stan, Avik Ghosh 6/07-6/11 NSF
NIRT "Surface State Engineering - Charge
Storage and Conduction in Organo- Silicon
Heterostructures as a Basis for Nanoscale Devices
John Bean (PI), Avik Ghosh, Lloyd Harriott, Lin
Pu (Ch), Keith Williams (Phy)
2
Origins of Molecular Electronics
Aviram and Ratner (IBM Research 1974)
Basic organic chemistry ? Local attraction or
emission of charge "Electron Withdrawing
Groups" "Electron Contributing
Groups" Example NO2 group attached to benzene
ring (here stretched out)
REAL arrangement after Oxygen at left steals
electrons from Carbons to fill its outer "shell"
Chemistry textbook Chapter 1 version of valence
electron arrangement
3
But translated into semiconductor speak . . .
"Electron Withdrawing" Acceptor
"Electron contributing" Donor
And what do you get when you put Acceptors sit
beside Donors? A DIODE And THAT is what Aviram
and Ratner proposed
Quino withdrawing group Ethane spacer
Methoxy contributing group P
I N
4
Did it Work?
0.5 nm
Not Exactly Electron clouds spread out too
much (formed "molecular orbitals") Everything
was washed out!
But eventually, molecular switching WAS obtained
using "Nitro-OPE"
5
But only in "hero" test structures
Called "Break Junctions"
Which in reality looked a bit more like this
6
"Maturing" only slightly into structures such as
Yale's "Nanopore" molecular electronic "device"
test structure
Au
Holes etched into wafers from BOTH
sides Molecules then (hopefully) filling minute
hole formed in thinned layer FURTHER From our
"mole" (UVA student who'd worked summers at
leading Yale lab), we learned that Typical
device yield ONE per 100 or more
tried (and that was a GOOD day!)
Si3N4
Si
Si
SiO2
Au
Au
Si3N4
Au
Alkanethiol
7
Bringing us to the 2001 point where we entered
field
BTW To an academic, "entering field" when we
start grubbing for research
The Good News Bad News
Single molecular switch HAD been
demonstrated "Device" was incredibly small 1
nanometer long x tenths nm in cross section But
published circuit results were (peculiarly)
missing Further It was rumored that device
yields were alarmingly close to 0 and . .
. Field produced highest visibility scientific
fraud of the 1990's (Schön - BTL) Nevertheless,
our first grant took at face value proposed
interfacing with Si IC tech "Merged
CMOS/Molecular Integrated Circuit
Fabrication" That is, we (as "Si people") would
just try to add Moletronics sub-circuits to ICs
8
So we first devised our own test device
UVA Nanowell
200 nm Au
5 nm Ti
Self Assembling Monolayer of organothiol
molecules
100 nm SiO2
Focused Ion Beam milled hole (approx. 50nm)
200 nm Au
SiO2/Si

• - Planar device - Easy to fabricate -
  Potential for Circuit Integration
• With which we compiled AND PUBLISHED field's
  first statistics on device yield
• Using liquid phase molecular deposition, we got
  yields ? 5
• We were disappointed - but field leaders
  whispered to us "that's GREAT!"

9
Leading to 2nd proposal's fundamental changes in
processing
"Hybrid Mole Computer Using Vapor Phase
Assembly" Still spoke of Molecular / IC hybrids,
but with new technology "Vapor Phase" - Based
on my own MBE apparatus and experience
Deposition Chamber 1 Original semiconductor
process
Deposition Chamber 2 Converted to Organics
Center Loading Processing Chamber
10
Why Vapor Phase?
1) More compatible with state-of-the-art IC
processing 2) Much higher purity - Why
Chemical synthesis is a triumph / nightmare!
Excerpt from seventeen step process used to
synthesize a molecular switch Each step has only
30-80 success rate ? need to filter out
balance BOTTOM LINE a) Typical purity 99
(99.9 GREAT!) b) Practical MOLEtronics
switch involves 2000-10,000 molecules Odds of
getting SINGLE pure switch?
11
Whereas vapor phase is intrinsically pure
We evaporate! Means we Distill (automatically!)
High vapor pressure contaminants outgassed
beforehand Low vapor pressure contaminants
never liberated This UVA Vapor process produced
first TRUE Self-Assembled Monolayers
Liquid phase deposition pseudo SAM" 1) As
depositied 2) Rinse with solvent 3)
Left w/ chemisorbed layer Vapor phase deposition
TRUE self-assembled monolayer That's
it, done! (excess re-evaporates) Single
ordered monolayer! Ready to proceed with
device fabrication! AND VAPOR PHASE DEVICE
YIELDS ? 30 (vs. liquid state-of the art 1-5)
12
But there were lingering problems
Molecules attached via S-H groups to atoms of
Gold substrate layer But gold is neither single
crystal nor atomically smooth ? "Low"
magnification AFM image Near atomic
resolution STM image ?
Leading to almost unavoidable short circuits at
grain boundaries
13
2006 - Targeted attachment to Silicon single
crystal
State-of-the-art Si surfaces are "hydrogen
passivated" Si (100) Si (111) To
attach organic molecule to Si, must eliminate or
displace hydrogen layer Which we did using a
process known as HYDROSILYLATION
Unhappy 2nd carbon steals H from neighboring Si ?
chain reaction
Organic molecule w/ terminal double/triple bond,
breaks one of those bonds to attach to bare Si
Heat or UV blows H off one Si
14
Hydroilylation known to sort of work in liquid
phase . . .
But would it produce 100 molecular
coverage? And would in work it VAPOR
PHASE? We tried with a variety of molecules
(liquid vapor, with heat and UV)
Phenylacetylene (precursor to MANY known
switching molecules) Shortened version of the
switching standard, Nitro-OPE 4-ethynyl
2-nitroaniline Biphenyl dinitro
(BPDN) Ethynyl ferrocene
15
Results?
The GOOD NEWS Hydrosilylation DOES produce
near 100 coverage for all molecules
tested Vapor deposited layers are again
self-limiting TRUE SAMs monolayers The BAD
NEWS Silicon atoms are spaced more widely (and
thus so are attached molecules) Many
electrically switching molecules are large and
don't fit tightly together Still possible for
top metal to short out device via gaps
between molecules And in some cases,
electrically active subunits lost in process of
hydrosilylation
16
So we took stock
Classic molecular electronic device devilishly
difficult to produce Only a one molecule thick
(1nm) gap between metal contacts Nevertheless,
molecular electronics demonstrated
that Molecules CAN conduct Molecules CAN
produce subtle electron withdrawal or
contribution And at UVA, we had figured out how
to Grow much higher purity molecular layers
using vapor phase Grow them as TRUE
self-limiting single molecular layers Attach
them densely to a silicon surface How to
otherwise undermine "Brick Wall" facing
microelectronics?
17
Apply to Nano-MOSFETs
Semiconductors depend on part per million to part
per thousand dopants In nanoFET, may not be
enough volume to assure presence of dopant Or
concentration subject to large statistical
variations Similar statistical problems with
transport Electron "mobility" involves
statistically averaged scattering on ions And
as charges pressed closer together, repulsion ?
MUCH stronger Producing effects like "Coulomb
Blockades" OR Random Telegraph
Signals (fluctuations in FET current with trap
charging and discharging)
18
Molecules on Si surface might be far more subtle
Use molecular charge withdrawal / contribution in
lieu of doping Wouldn't so much dope as fine
tune position of Fermi level And could open door
to new quantum conduction effects Wave
functions spill over entire molecules Thing
that scuttled Aviram and Ratner's original
proposal! When molecule is bound to silicon,
get new wave function Hybrid of silicon wave
function molecular wave function Wave Wave
? Beating and interference "Fano scattering"
19
Labeled this new Molecular Electronics
Research "Surface State Engineering"
Recalling "Bandgap Engineering" of early
semiconductor heterostructures Which would be
applied to silicon on insulator (SOI)
NanoFETS Team John Bean (PI) and Lin
Pu (Ch) Molecular Attachment to Si Avik Ghosh
Modeling of hybrid Si molecular
conduction Lloyd Harriott Fabrication and
testing of NanoFETS Keith Williams (Phy)
Evaluation of quantum conduction effects
20
With additional team members
George Celler - Chief Scientist of
SOITEC World's leading producer of SOI wafers
for the integrated circuit industry And former
Bell Labs collaborator of mine and Lloyd's ?
Source of expertise (and, we hope,
wafers) And scientists at the National
Institute of Standards and Technology Nadine
Gergel-Hackett and her colleagues Dave Seiler and
Curt Richter (Nadine is recent Ph.D.
graduate from our molecular electronics group)
? Sources of advanced analytical
capabilities
21
And proposed outreach component
That would build upon UVA Virtual Lab science
education website www.virlab.virginia.edu -
Animated 3D presentations on over 50 micro and
nano topics - Over 3.5 million hits since
2005 2005 NSF Nanoscience Undergraduate
Education Grant - 200,000 used to acquire
six STM's and AFMs for new "Hands-on
Introduction to Nanoscience" Freshman/sophomore
class for students of ANY major www.virlab.virgi
nia.edu/Nanoscience_class/Nanoscience_class.htm

22
That would reach out across the Commonwealth
Via partnership with the Science Museum of
Virginia (SMV) To Develop new museum
exhibits and displays Help with design of new
student-project-oriented NOVA museum
site Participate in Museum's K-12 Science
teacher education programs
23
Result of efforts?
Against at least 101 odds, we were funded last
June 1.3 million dollars, spanning four
years Team now up and running Already have
preliminary basic science theory exper.
results NanoFET device test structure design
under way And outreach is also rolling UVA
SMV K-12 science teacher course planned summer
2008 Compressed version of UVA "Hands-on"
nanoclass With state backing, class also being
transferred to other schools Danville
Community College to support their Nanotech
business incubator
24
Which we might also be able to leverage via . . .
New DARPA initiative on Bioelectronics RFP to
be issued next March We've been in consultation
with key DARPA players since February Last June
we briefed them on idea we have developed Way
of taming DNA for bioelectronics Which DAPRA
has since repeatedly cited as "Best idea we've
heard yet!" So STAY TUNED "Molecular
Electronics" may not survive in the form
originally envisaged But its spin-offs are very
much alive and well