Why are seniors no years away
in knowledge of research.

What's at stake is a national issue

here in India in his rhetoric
that you're innocent your source

is regress to existing for

the first time this research research

research here as well as various articles
published more than your own conference.

Well as his research is
currently focused in

which is what he thought was right.

Good afternoon.

Feel free to interrupt me at any time to
ask questions in a way it may be good

that there are no slides because they
are distracted by all the numbers in

the figures going on and
can people at the back hear me.

Well the last row over it.

So they're going to talk about
graphene nano electronics and

knowing that this is a day worth
audience all to type on many

different topics but
if something is not clear.

Then I'd be happy to explain more on that
topic and that's one of the sponsors of

this projects including And
the surf index program.

I've seen Mark on the P. us.

And the work that was done here was
the graduate students in the group.

Brian.

Given Brenner and research into
Thomas back first looking at why.

Graphene is interesting.

So the caucus actually
organized the four different

sections one of the first ones is
looking at why graphing is interesting.

Now I don't know if anybody's in this
audience who has not heard of graphene

before.

So will I want to ask that
question to raise your hand.

But.

Graphene as gotten a lot of
interest in the past few years and

one of the reasons that it is so

interesting is because it's a potential
next generation electronic material

the mobility of graphene start to be
more than forty thousand or less when

measured to be on the order of two hundred
thousand for suspended graph the for every

tax of grapheme some of the layers I've
been shown to be more than two thousand.

So even a square per wall second.

And if we can't for Gratian people are
measured to be more than ten thousand for

exfoliate Agrafena then
why is mobility important.

Well mobility comes in the picture when
you have the odd current of mass for

its large current is
a mixture of mobility and

the carry density in fact
the conductance is equal to

the electronic charge multiplied by the
carry density multiplied by the mobility.

But mobility by itself does not
say the whole story because

as the other term which is scary
density in fact in three five So

my connectors you have excellent mobility
but you have a very poor Kerry did see

which is why you always
find you can't really give

you the same type of our current as
a silicon master even though the mobility

can be maybe ten times or even a hundred
times higher than that of silicon.

But thankfully in graphene we both have
a good mobility as well as a good carry

densities so that points to
the possibility of a good on current and

a good switching speed because on current
besides the switching speed which is C.V..

And capacitance risk.

Parable to that of silicon technology and
using a low swing

switching you can actually get much better
performance with graphene transistors.

So that's one big advantage the high
mobility combined with a high carry

density.

The second advantage goes back
to the interconnect properties

that are highly important for
skilled computer chips.

So when you start scaling down
computer chips the the word of

the lines connecting different transistors
goes down and the when the work goes down

the resistivity goes up resistivity
is supposed to be a bulk

property independent of the dimension
of the material but in reality

when you scale down the dimensions of
let's say copper aluminum there is a still

really was supposed to be dimensional as
a dimension independent actually goes up

with decreasing worked and the reason
goes back to some scattering and

domain while scattering mechanisms
that have been copper and aluminum.

And so when you get down to
twenty nanometer dimensions

copper resistivity is on the order
of five microns centimeter.

Whereas for graphene you can actually have
resistivity on the order of one micron

centimeter which is five times lower
than that of copper and later on I'll

show you exactly what this means in terms
of using it as an intricate material.

Not the other advantage of graphene
as it breaks down current density

which is on the order of ten to
the power of nine percent with a square.

And this is over three orders of
magnitude better than that of copper and

that matters because you're
electromigration properties are going to

be extremely better with grapheme and
that in turn matters for

the current carrying up interconnect So if
you look at a current generation computer

chip it probably carries hundreds of
amperes of current and imagine the amount

of current density that's going through
each one of these tiny copper wires.

Each time it has to carry the current
to the individual trying.

The stor going through
the lifetime of a chip

with a lifetime's probably on
the order of ten years to design life.

As is as it reduces the width
of these copper nano wires their

lifetimes getting increasingly
taxed because the amount of current

They need to carry which is actually
increasing with each skilled generation.

So the mixture of reduced dimension and
the increased current carrying constraints

placed on these wires makes them
very immune to break down early

and so using graphene can hopefully
extend the lifetime of these chips.

The other advantages are more mechanical
which is the young smugglers.

It's about three times that of silicon so

there's hope that grapheme can in fact
be used as a men's type of a device.

The other one is that graph it has in
terms of thermal conductivity it's about

five thousand watts per meter Kilburn
which is actually even better than

diamond much better than graphite
ten times better than copper.

So if you look at heat things being
used currently mostly copper and

some sort of a lawyer that uses the heat.

So they're used to actually
conduct heat away from a chip and

spread it out over large areas.

And graphene can be used in those cases as
a heat spreader but unfortunately graphene

as a wee little Z. axis thermal
conductivity which is only about fifty

watts for a miracle which is one hundred
times lower than that of in plain thermal

conductivity so you can't really use
it as a terrible interface material yet

there are some efforts underway to
see how to align these graphene

sheets in order to make them amenable to
use it as a thermal interface material but

it can be used as a heat spreader.

So all this advantage is
a mix of graphene very useful

in a number of applications not just
as a policy mass type of switch or

a post copper internet but
really memes to ISIS for.

Storage because it's got a high
surface area for sensors.

Once again because of its excellent.

So for say it a while you.

Can be used for current carrying wires
actually for Par distribution networks

are the other mega scale not really micro
but mega scale participation networks.

And the applications go on and

on depending upon just which property of
graphene you're trying to make use of.

Now in terms of the timeline

of when the graph in some of the graph you
know applications will come to fruition.

When they would hit the market.

You probably have already
heard a lot of stories about.

A Korean university.

Some of you sent in one university
which has gotten these huge

thirty inch graphene sheets and.

What they do is some sort of C.B.D.
growth process in terms of getting

graphene on copper fault and
design the copper falls away and

then using them as electrodes going
back to some other bands of graphene but

I told you what it's got this
excellent conduct D.V.D.

multiply that with the thickness of
graphene that's needed in order to achieve

the kind of beauty in the thickness is
only of maybe a few tens of nanometers

that you need and now you actually
competing with Indian pin oxide which is

used as a transparent electrode in many
of these L.C.D. type of applications.

So I T. O. actually has worst conduct
really and the worst article transference

compared to graphene and Samsung in
the same universe is actually shown that

you can get excellent properties as
a transparent electrode with graphene.

So that may actually be the most
near term application of widespread

graphene usage in transparent electrodes.

And the next one would be
as a thermal material.

And to differentiate graphene thermal
material from graph five graphite is that

being already used as a thermal
material and it has a thermal conduct.

On the order of five hundred to six
hundred watt per metre kill when it's low

cost compared to copper and it's got all
these advantages but graphene a special

it's got a thermal kind of to be a five
thousand what we're made to kill and

which is ten times that of graphite and
in order to make use of this we need

to actually come up with a few engineering
in a way sions I won't go into that but so

that we differentiate
graphene from graphite and

we foresee that that will probably happen
in the next four to five year timeline and

hopefully we'll start seeing some
thermal conductivity applications

based on graphene being used or.

The next applications will be some sort
of composites there's a lot of talk about

using graphene in some sort of
a composite to make the whole

system to be better behaved than if
you have only one of those material So

graphene by itself may have
a good mechanical strength but

how do you use it in
a park like let's say.

A road or of aircraft rather of
a helicopter or or wing of an airplane.

And Scott you all heard of carbon
composites but graphene composites promise

even more because because of
their specific share strength and

space with young smugglers is is much
better than just carbon and graphite.

So there is hope that graphene composites
going to be used in different mechanical

parts in order to strengthen them but
at the same time make them lightweight.

And this would be unlike the seventy
eight year timeline and the same time and

we can probably see other
things like using it for

sensors and using it for maybe hundreds
in storage and then beyond all this.

Maybe in the ten to fifteen year timeline
is when we're really looking at using it

as a post card for materials so
when silicon start scaling down to sub ten

nanometer dimensions as well and copper
become so problematic that people actually

start looking at other alternative
materials to replace copper with them and.

Fien as a whole up there and in terms of
being used as a pro Seymour switch that

that's kind of even further out
maybe fifteen twenty years down when

people realize that OK silicon scaling
going from one generation to the other

is going to be problematic because of
some issues which are going to later on.

And then graphene as hope that
it would replace silicon.

Now one of the challenges of using
graphene one of the biggest challenges.

Just like in carbon
nanotubes was the growth.

So in carbon nanotubes Let's see there was
another big problem which is alignment How

do you actually aligned isn't that
it used to where you need them to be

once you manufacture them and
that was always the big question but

with graphene at least we don't
have that issue we can actually

look at graphically pattern them and
place them where we want to be.

But there is still the issue of getting
the larger graphene sheet that can be

compatible with the the type of materials
and the type of processing that is done

currently otherwise
changing the entire line of

manufacturing process to a completely
new line would be very expensive and

perhaps inhibit the usage of graphene.

Under some of the other challenges
sort of things like integration and

manufacturing and
yield which I think would be so

much easier to solve than the growth
problem that in terms of economic feasible

to you know we've talked about some of
the challenges potential applications but

economic visible T.
is especially important for

using it as a postcard propose
seamless replacement and over there.

The cost of comes in the picture which is
easily upward of four billion right now

and then the question is
what has graphene to offer

that is not already offered by silicon.

So it's not just enough to say
graphene can outperform copper by

let's say fifty percent.

Or it can outperform silicon by two X..

But we should be really getting an order
of magnitude advantage from graphene

compared to what are medieval three
placing if there's going to be

a big change either in the front
end of line or the back end of line

fab because just the cost of the building
the been manufacturing them in large

quantities is going to slow down
the adoption of graphene and

we need to be mindful of how to
integrate some of the processing

of graphene we're processing that's
currently going on in the silicon five.

So in terms of.

In terms of just looking
at the power density of

computer chips if you if you go back and
think about the one nine hundred sixty S.

we used to have these are mostly
in sixty's and seventy's and

eighty's the bipolar transistor slowly
started propagating and then by the early

eighty's it became clear that the bipolar
technology had a very high current

density in fact somebody has done the
study where the plot the module heat flux.

Bipolar boards and
then they see an interesting trend so

those were only to use the white board so
this is the module heat flux

and this is let's say the time and this is
the nineteen eighty and ninety nine the.

And what what the study
shows is that by and

by the early eighty's your margin
he had read a really high level and

why does that matter because that's
going to heat up your system.

Reduce it's reliable the and then
place constraints on other things like

maybe instead of air calling you may need
some sort of a special cooling module.

And that's one seamless came in
the picture because now you don't have

the base current in the bipolar.

So you could just use see mass
with very little gate leakage and

so on that maybe in the early
one nine hundred eighty S.

The Master core and
it's been going ever since.

But now we're reaching almost the same
level of module heat flux which is

a computer chip you're
probably dissipating more than

one hundred to one twenty watts.

And you barely have a centimeter square or
so of area.

So really reaching
the limits of air calling

there's already talk about
going to water cooling.

But all of those are expensive options
because of reliable to constraints.

So for the mass market for portable
applications for desktop computers.

What sort of technology.

Do we need that can reduce the seat flux.

Well unfortunately we don't have
a technology here yet that can take or see

mass and bring down the marginal heat load
so this will probably go higher limit and

maybe saturate as we go into multi-core
technology and this is one of the biggest.

Motivators for finding a replacement for

silicon and if you look at the our
what constrains the scalable T.

of silicon it's not
manufacturing technology.

So if you think that the people
can do thirty nanometer or

twenty nanometer technology.

Well that they can already do it.

People already working on
twenty two nanometer and

below technology generations.

In fact.

I don't think there's going to going to
be any problem even going down to sub

twenty nanometer or
maybe even twelve nanometer technology so

scaling down technology from
a manufacturing point a few is

never the problem you can do that with
silicon you can do that with the fab and

the hopefully that you we technology
will come up if not even a father.

If you will take over for
the smallest dimension so doing

the manufacturing was never the problem
the problem goes back to some fundamental.

Limits which is one of them
being the heat dissipation.

If you look at how much energy.

Seamers transistor currently needs to
switch on the order of ten thousand K.T..

Where as well there's a person in
this room demining who is shown that

you only need a lot of to.

The switch.

So we actually have that big room between
Katie lot of two and ten thousand K.T..

In which we can switch but
what we are ten thousand Guinea because of

various constraints like power supplies
noise and switching noise and so on.

And and
where is all this ten thousand K.T.

going to it's all going to
heating up your chip and when

you have a billion of these transistors
in a chip you're going to dissipate

upwards of one hundred watts to one
twenty watt with a billion transistors.

And then looking forward to
more multi-core more density

you're only getting more and
more density and

that in turn reduces even
further the amount of advantage.

You're getting from each
scaling generation.

So manufacturing is never
limiting your Skilling

it's basically all these other
fundamental limits that are limiting at

one of them the primary one
being that he dissipation.

Now another big bottleneck
maybe from one device to

the other word a lot of talk is
going on in terms of design for

manufacturability In fact many of the CAD
tools that are shipped nowadays include

beer from modules that you do when
you steps of design spaces so for

example designing for the worst case the
designing for the the Mino designing for

best case so you can actually mix and
match this critical paths.

So that you have a design that looks
at what the variable to use doing

to your design and then map out some
other channel lengths and channel worth.

According to what the.

What the tool predicts would be a.

Let's say case case by case scenario.

So all the stock was because.

Just looking at if grow.

Can Replace silicon we look at
the economic feasibility and

I want to go into what
does a new technology need

in order to replace silicon
as a switching device.

So I told you about ten thousand
Katie right now being required but

if you go down to the end of the road map.

There's a road map usually put out by the
idea as the international technology road

map for solar connectors and the latest
edition was back in two thousand and

nine when The predict that by two thousand
and twenty four or so is when the end of

silicon scaling would be that end of
the silicon scaling region regime.

You are switching energy would be on
the order of thousand Katie which is

ten times lower than what we have today.

But a new technology that is
to replace graphene would have

I'm sorry to replace silicon would have
to be at least ten times better than that

if it has any chance of replacing it
with all the fab cars and with all

the different manufacturing constraints
that the new technology may impose so

which means we really need on the
something on the order of one hundred K.T.

to be the switching energy but still that
leaves a lot of room for us from hundred

K.B. all the way down to Katy lot of two
which is the fundamental switching limit.

Next there is the.

On current or let's say
the the power which is dissipated.

When the devices on and that power is on
the order of a few hundred watts right now

like I told you for a billion devices but
once again we need to be on the order of

a few tens of watts and then we got onto
standby power which is when let's say

the device is not doing any active
complication What's the leakage.

Multiply that by the wall that you
know that's a standby power and

that's on easily on the order of a few
micro or even Milli Watts right now for

some high performance to us but

we needed to be less than a nano wire and
the way we would achieve all these

reduction in the end what the active
on the standby power would be.

By looking at some of the basic
factors that go into designing

the future transistors.

So just by replacing your channel which
is made of silicon with the channel

that's made of different materials like
graphene is obviously not going to do it.

You need a completely new
switching paradigm and and

that's a central message when you look at

what is needed in order requests silicon
it's not going to be just a mass for

it where you just take out a channel
material and replace it with grapheme but

it has to be more than that which is
using the new material in new ways.

So graphene in fact displays a lot
of novel effects the tools and

off promise to make us feel that
using those techniques would make

it highly amenable to using a switching
device so I'll talk about a few examples.

Later on but the point of this is
just replacing the channel material

of a silicon master but some of
the material is not going to do the trick.

They are looking at the two major topics
that I want to focus on one of them

is graph in transistors and the other one
is graphene Internet so first we look at.

Graphene transistors how are.

Graphing transistors fabricated in
research labs currently one of the most

common ways is to use explode in a graph
in which you take a block of graphite and

the user Scott statement had to peel
off single layers of graphene and

then deposit them on an ox substrate
pattern electrodes on them and

then you can actually
pattern even further channels

on them to make them into graphene or
a bins and then you characterize them.

Well that's easier said than done because
it's not just about the the fabrication

process but it's also about
getting high quality devices.

So for example when you play a car for
a big chunk from.

The graphite piece How do you know that's
a good quality flick will you need to do

some RAM on imaging on it to make sure
that there is no defects on it and

even after you do there are many amazing
you're not really sure how many domains.

There are or what the domain sizes so you
need to actually fabricate the device look

at the electrical properties from that you
would know whether it's a good quality or

a moderate or even a bad quality device.

So using clear graphene as been

very good in terms of getting some of
the pioneering results in graphene So for

example some of the early work which
started back in two thousand and four and

flamed graphene showed a lot
of the interesting efforts in.

For example climb tunneling and
quantum Hall Effect and

very Spears and all of these work in.

Graphene was using play graphing and.

The other types of doing graphene
are using grown graphene for example and

one of the major efforts headed by Dennis
and all the here is the Georgia Tech more

Cirque which is doing the graph in growth
using primarily a Paxil graphene and

silicon carbide so you probably heard
a couple of talks already on that and.

What the method uses a silicon
carbide starting substrate.

And that's heated up to the right
temperature in order to get

the right number of grapheme layers and
they've shown extremely high mobility and

good transfer properties and
in addition to that there are other

ways of getting graphene people have
used chemical really derive graphene So

you take a block of graphite and
then put it in self

you're cast to take it through
a few chemical processing steps and

you break it down into really ten and
narrow that Arubans and

people have been able to actually measure
band gaps on these narrow ribbons.

But on the other hand you're back
into the same area of carbon

nanotubes where you where you have
a solution now obvious graphene flicks and

you have to now align to
each one of these can.

Dr mannerable So the chemical method
is not really a growth with that but

more to study some interesting properties
of graphene So in addition to that is

a C.V.T. method which I told you
a little about before and over there.

What's done is you take a.

Metallic catalyst so

usually it's a copper copper is known
to be better than other metals.

But sometimes nickel is also used or

the beauty about using copper is
it's a self limiting process so

you take foil of copper and it can be
either on a substrate or by itself and

then you put it into a form as
heat up to a few hundred degree C.

like seven hundred eight hundred
pass a carbon containing gas and

then the graphene starts to form by
the interaction of the metal catalyst.

And the carbon containing
gas flowing over it.

It's been shown that you can
actually get modern graphene and

copper whereas in nickel it
forms domains of copper that

I'm sorry domains of grapheme that may or
may not be continuous and

it's also not a self limiting process
a lot of the work that I've seen with.

C.B.D. graphene is mostly on copper foils.

Look at looking at the different
properties between these different

types of graphene so

flick graphene people are measured of
these on the order of ten thousand fifteen

thousand maybe even twenty thousand
centimeters corporate world second and

you compare that to the mobility
of silicon which is and you guess.

They're just a range.

OK Anybody can guess the range for
me for the silicon mobility.

Which is an OK So the range is anywhere

from one hundred to can be even less than
a hundred but just with an easy example.

A hundred to a thousand when it's under
you can have mobility of maybe a thousand.

But when it's deliberate reduces So two
hundred is probably what we have now with

the current level of doping and
the silicon transistors and

graphene with the carry density of on
the order of ten to the power of twelve

percent in a square which is equal
in terms of the power of eighteen

percent in a Q.B. actually have
a mobility as high as fifteen thousand.

And that was four X. one unit graphene.

Now for

other types of graphene perceivably
grown graphene the mobility is lower.

It's on the order of maybe
two thousand at the most but

most of them is actually lower.

So using it as a material for
post CMOS transistors is not yet.

A possibility but hopefully that
process will get better and better and

the mobility will get higher and compete
with other types of graphene and whatever

Taxil graphene the sheets of graphene show
five thousand ten thousand mobility easily

and chemically graphene
the Nano ribbons show mobility

of about one hundred sheets don't matter
there because they don't do sheets and

chemically derived but
for the nano ribbon.

Chemically derived.

Graphene they're actually sure mobility
of one hundred to one fifty which

actually competes OK with silicon but
still not

good enough to actually replace silicon
for any application like a transistor.

OK

that's a good question.

Hopefully nobody heard it.

Why C. would be grown graphicness
lower mobility and so.

There's a little less
a few cases one of them as

the seemingly grown graphene is no defect.

It's a pristine CD graphene yet

it has a lower mobility that may
just be because the domain size.

Domains are really small and the order
of five micron ten micron domains then

I'm sorry not for my kind of huge
five nanometer ten nanometer domains

you're really not going to
get much of a mobility.

Whereas a production graph in people have
shown the domain size to be on the order

of a few hundreds of nanometers
unexploited there is no limit you can

actually get single crystal domains
are as big as tens of microns on a side.

So it just comes back to what's
a domain so as in the biggest I've seen

with CD gone is probably on
the order of four or five microns.

But they don't achieve that
consistently it's very occasionally

people are able to achieve.

Micron size domains but most are time
it's tens of nanometers with C.V.T. and

hundreds of nanometers with every tax will
grapheme and maybe tens of microns with

expletive grapheme and that may explain
some of the mobility the difference.

Yeah

yeah the domains
are are pretty much like a.

Poly type of material that
if you're powerless silicon

it's made up of crystal in regions but
still it's not a continuous Crystal and

regional So whenever the electron the
carriers in the cross these domains then

can't have this extra scattering So if
you look at the mobility within a domain.

It would be really high but
as a assume is across the main boundary

the extra scattering would
reduce the mobility.

You're absolutely.

But still a theoretical pursuit I don't
think there is something called single

crystal carp or people who have looked
at single crystal nickel but there is a.

Like I said there is no self limiting.

It will be of the nickel.

So they haven't been able to grow
in layers of graphene on it.

So we're going on.

Grapheme transistors.

So what I mentioned was the starting
material could be any one of these.

And now.

What are some of the mechanisms
that limit the transport in

graphene because in the context of
either using it as a transistor or

an interconnect you're
always interested in.

OK if I get this material to
be the best possible material

then what's the type of
transport that it will exhibit.

Well one of the one of the main
attractions of grapheme is that

apparently there is no phone and
scattering mysteriously there is no

phone on scaring so
the mobility can be as high as millions.

Although people have measured it to be
only on the order of two hundred thousand

for suspended graphene and that's partly
because of some manmade defects and

not really the the limit
of graphene mobility.

And now when you put
graphene on a substrate.

There is going to be that interaction
with substrate foreigners which limits

the mobility further than that for
silicon dioxide it's on the order

of forty thousand for silicon carbide
fifty thousand B.C. to sixty thousand

depending on with substrate you put in it
depends upon what type of acoustic four

nights interact with the graphene
carriers and that would limit mobility.

Accordingly But

let's say it's in there are a few tens
of thousands depending on the substrate.

And coming down even further.

You have impurities scattering and that's
really all the manmade type of scattering

that occurs when you fabricated devices so
when you go down into the lab and

put a forest on your substrate and
then slab graphene on it.

The impurities from the substrate are
going to interact with the carriers and

grapheme and reduce its mobility
because of the scattering potential and

used in the graphing layer and
then you add to that.

The width scaling as they reduce the width
of these nine ribbons you're going to get

more and
more scattering from the edges and

that in turn induces the linage
roughness scattering

which can limit mobility even more so
in a realistic device.

The intrinsic one and
scattering of millions

of mobility are the substratum
that scattering tens of thousands.

Don't even come into picture.

We are mostly faced with impurity limited
scattering which can limit mobility to

down to like two thousand three thousand
and that's one of the scenarios which

the see when you go on a graph you
may also face because over there.

It's a lot of the processing that we have
to do in order to remove that copper and

put it on some sort of a resistor and
then slap it on oxide.

So there could be lots of impurities
scattering that limiting immobility In

addition there is also the the edge
roughness scattering which could be coming

in from the edges when you
reduce the line with abuse.

Now Reubens So that just shows you
what are some of the limiting factors

in a realistic graphene sample.

So now what we have done

in one of our experiments is to extract
the word dependence of the mobility.

Just to see how it behaves for
example you get a graph in and

while I use a papa johns to show you
two areas of zucchini new graph.

So here

what we have is the word which would
be on the log scale one micron.

And this would be the mobility.

So even a square.

I can.

That's the human.

So if you look at mobility four.

X. four year a graph in samples what
we see here is it starts off at about

three thousand to five thousand.

Come start and
as you start reaching sixty nanometers

you actually start seeing
the degradation like that.

And when you reach about twenty nanometers
you probably have mobility of about

one hundred or so
one hundred to two hundred maybe.

And.

This goes back to the lines roughness
used scattering that I was talking about

earlier.

And when you start reducing
the dimensions of this group and

you start making a carrier score through
more and more scattering events.

Now there is some theory that
carries some traffic in or not and

libelled the backscattering But but
this is what we're seeing for samples

with impurity density of about ten to
the point of twelve percent with a square.

And once again why we're seeing only
like three thousand to five thousand

mobility here goes back to the impurities
scattering in some of our samples and.

The reason that there is that with
dependence is because there is more and

more interaction of the carriers
with the edges for similarly

grown graphene where we have domains
you have more and more interaction

the carriers with the domain boundaries
and that in turn would reduce mobility.

So in those cases it's a little more
complicated all these things are for

mostly single crystal exploded grapheme So

when you have domain boundaries
now you have an overlap

of the sized opponents with respect
to the work as well as the dependence

because the domain boundaries which
I want to get into in the stock.

So one of the other results that we showed
was that we can actually dove graphene

using a commonly used as a so-called
hydrogen through the screen arcs and

so I'll spend like a minute talking
about that what we did was just Q.

which is actually initially used
as a locus but on dielectric and

then people found out it's an.

And even with Aga few to resist and

what we found was news had
to skew on graph in a can.

The graph in either P.
type or and now you know how

in silicon you have substitutional
dropping you can have a beer and

type dope in the opening up on
the donation or acceptance of a charge but

in graphene the most of the studies you
see are not substitutional doping they're

mostly surface charge transfer building or
they can be a little static doping.

Maybe occasionally you'll see
intercalated there being but

very rarely substitutional doping.

So this particular to skew doping
mechanism is a surface charge transfer

mechanism where a molecule of some type
in this case such as Q. comes close to

graphene and either donate or accept the
charge and the ops graphene accordingly.

So the doping density would
be very very low even for

a high density of the starting molecule.

So if you had the power of like let's
say eighteen percent of the cube of

the starting material that you're talking
density could only be the power of

fourteen and the power of the inserts
basically point zero or one electron or

whole or molecule of the source species
that some of the doping then see is

into graphene So it turns out that when
you say to skew the molecule molecule or

composition is determined by how much
energy has been internet askew and

there are two types of bonds in it.

Askew that are actually competing with
each other one is a silicon hydrogen bond

which has a lower bond strength
of about four electron Walt and

there's the silicon arcs and
one which is a panther That's about nine

electron walled and at very low
instant energies for it to skew and

the energy could be from thermal energy
plasma or any one of these methods so

when you slap a disk you on graphene
give it a little bit of energy.

You're actually releasing silicon because
of I'm sorry hydrogen because of the so

it can hydrogen bonds breaking and
that in turn don't.

Graphing to be entered.

And then as a as a start increasing
the amount of energy incident are they to

skewer the silicon arcs in Bonn
which has a higher points and

that starts breaking the hydrogen
is our guest in arcs and

takes over the doping and
action develops graphene to be beat up.

And that's why using a single material
you're able to get either end or

point out doping of graphene.

So now let me briefly touch
upon graphene interconnects So

first we look at why graphene.

Is just.

OK I think that's a very good question
why is there are no discussion about

band gap in the stock a los are probably
mention it at least in one slide but

another to raise the question.

Graphene does not have a band.

It's a semi metal and
if you look at its band structure.

There's a linear E.-K.

spectrum where the meat almost
perfectly and the bandgap is zero.

Whereas graphite has actually Or like it's
got more from a metallic structure with

a conduction of will and spans overlap but

the question that you're asking is
irrespective of does bandgap or not.

Well you thought it was a small bank but
it's actually a zero about Gap

which is actually worse now what do we
do about using it for a transistor.

One of the ways this quantum confinement.

When you let me mention the four or five
different methods people have looked at

for bandgap one of the simplest is quantum
confinement when you start reducing

the weight of your graphene then you start
confining the morgues of the carriers and

that naturally opens a bandgap but
we don't know if that will be the method.

Favorable method for

electronic complications because of
issues there are some fundamental of

manufacturing issues how do we actually
make them reliably you need about

ten nanometer wide nano ribbons in order
to open up about a hundred million.

But how much of a bang get you need for

a transistor Well that goes back to your
honor for a show with about twenty six

million trying Walt energy from thermal
from the thermal the thermal induced

carriers you need to be about this
twenty six grain so a band get for

at least one hundred million looked
around the world as required.

But if you ask people around the around
in the silicon area they would say more

like two hundred three hundred M. e v and
even that may be low by some standards but

quantum confinement is one method and
other people have shown by layer graphene

has a good band gap to spend
using an asymmetric field and

that's done by basically having a dark get
a bottom gate or some sort of symmetric

doping which splits the symmetry
of the in the some lattice and

graphene and that opens a bandgap but
it has its own issues as well.

And one of the other methods is to use one
or two layers of graphene on the silicon

face of silicon carbide and other not
direct electrical measurements but

some sort of indirect studies have
shown that that may open up a band.

That people love the graphene
measures where the whether you some

some sort of Palestinian molecules and
graphene to induce

a mess type of structure in graphene that
opens up a band but mobility suffers.

So there are all these methods.

Each one has their own disadvantage.

And it's not clear if any
one of them will work but

what is clear is just opening a bank
is not going to solve the issue

because like I was saying earlier you
don't want just a drop in replacement for

silicon in a mass wedding in fact
we don't even want to master it.

So there are many people who are looking.

How to Build a transistor without a band
gap and some other examples are using B.

and junctions for
example using Klein tunneling

it's a very interesting quantum phenomena
which I'll talk about in a minute.

Or by using let's say a by layer
condensate where you are you make use of

tunneling of the condensate particles
from one one layer to the other or

you can make use of spent transport all of
these don't require any bandgap you can

just without a band get you can actually
switch from one face to the other phrase.

So there is hope that and

because it will tied in with the fact
that we don't want a mass for

tree placement for silicon longer
the fact that graphene Does not sure.

But has excellent properties in two D.
seem to go kind of

hand in hand as an opportunity there
to find other ways of switching.

So going on to the next topic that we
just looked at why graphene looked

at graphing transistors a little bit.

So let's look at the graphic
interconnects and

I mentioned briefly about how
graphene in terms of its.

Interconnect properties are shown
very good resistivity and

to actually pinpoint some
numbers that were measure

on the order of about five to
ten microns centimeter for

there is a studio of graphene nano
Riverdance with the width of about twenty

nanometers Now how does this compare with
copper all copper also has about a five to

seven Micron centimeter resistivity when
you get down to twenty nanometer worked

and the industry continues to keep
scaling these down further and

further and further So the latest overt
is in a companies are able to make for

fourteen fifteen nanometer a couple wires
with the resistivity is on the order of

seven Micron centimeter and they're
looking to improve it even further.

So at least till the twelve or
thirteen nanometer generation.

I think copper is going to continue
to be the material of choice.

And then looking at the current carrying
capacity one of the interesting.

Things for this.

I wish I had the slide up but
what that shows is a technique which we

use to do some of the properties of
graphene which is you induce a high

current in the graphene wire and look at
the saturation properties of the ivy and

from that you actually are able to get
what is the thermal conductivity of

the Nano wires What is the current
carrying capacity and a whole lot of

other information and what we found was
thermal conductivity of nano wires on

the order of five hundred to a thousand
watt per meter kill when the current

carrying capacity is easily to the point
of nine percent with a square and.

So for the low resistivity nano wires.

It looks like they have a high
current carrying capacity and

a good thermal conductivity
which is good news.

And so
finally let me leave you all with this

one concert which alluded to
earlier which is Cline pearling So

the graphene shows a lot of interesting
properties now how do we make use of these

properties that are very very
different from other materials.

So one of the things that
arises out of this so

interesting system is
something called the.

Super spend.

Well first let me draw the E.Q. spectrum
of graphene I was telling you it's

a linear E.Q. spectrum and
this is called the direct point.

So there's I want to explain
much except to say this.

Led to a lot of thinking
from different theories and

the set have the graph you know the
perfect bend stop material to experiment

with you don't mean Larson Krantz you
don't need all these fancy equipment in

order to experiment with some
of the fundamental physics.

So not only is that interesting to
us engineers for a novel switch but

it's also into interesting to a lot of
physicists because of the opportunities it

presents for Bronze Star physics
European junction and well

let's not even call it a pinch actually
mentioned I just call it a barrier.

So if it's a classical You all
heard what happens to a classical.

Barry when when a particle is
instant on a classical barrier.

It just reflects off.

Now what about quantum mechanics
What does quantum mechanics tell us

when a particle is uncertain about
it does that reflect after for

it really well there is a certain finite
probable to have transmission and

that probably or perhaps machines
dependent upon the energy of the particle

is dependent upon
the barrier height maybe.

In case of import from barriers
that may be dependent upon

the width of the barrier.

But it turns out that in the class
of materials the graphene belongs to

this particle that's
incident on the barrier.

Goes through perfectly.

There is no reflection.

There is no finite
probability of transmission.

But really there is one hundred
percent probability of transmission.

Now let's ask ourselves
why that's the case.

Normally we look at this spectrum.

From this alone.

You can deduce.

Why this is happening we have to look
into something more fundamental which

is the carnal nature of the electron so

when graphene us propagating the crystal
depending upon the Kyra ality of

the carriers propagating the crystal
they can have different properties and

that property that we're interested in
rate know is called the pseudo spin.

So this branch of the energy is
different from this branch in

the sense that each branch
has its own spend so

there is not the true spin and
that's why it's called the pseudo spin.

So when you have a particle
going in here and

so for example let's say a particle
is humor and it hits the barrier and

if there's perfect backscattering it has
to then come to this right because it's

got the same momentum but the same
energy but the opposite momentum but

unfortunately that particular Bryant has
a different pseudo spin and this is north

both the flip and a source as well as the
flip of momentum which is just impossible.

Discus you can either
have it here flippantly.

Momentum or
a flip in the Source been but not both.

Usually in fact a flip and sort of spin
takes more energy than a flip of momentum.

So what happens here is
the back scattering actually

occurs such that it goes
to the same branch and

a different energy like the one over there
so it actually becomes a hole on the side.

When you have a particle incident
like this it just continues on with

the same moment because it's a whole.

No it's a negative moment.

But because it's a whole it's
traveling the same direction.

And it continues out aren't like that.

And why did this just happen because it
has to have either the same momentum or

the same sort of spin and
like I said to the spinners.

More difficult to flip so
it maintains a so the same

sort of spin flips the momentum becomes
a whole continues on the same direction.

An even more interesting.

This seems to be highly angled the planet.

So if you have a can you coming
in from a different angle.

Let's say that angle can actually be
perfectly reflected depending upon

the energy of the carrier and you can
actually draw up a plot of the angle

of the plane and properties of this
one and you get something like this.

So for a set of angles.

They plus or minus forty five degree or so

what you're doing is you're
transmitting the carriers perfectly.

So this is actually
the transmission problem.

So for a small range of angles.

You're perfectly transmitting
a carrier and plus or

minus forty five degree there is
a finer transmission probable the but

there is other angles in which
the transmission probably is zero.

But making use of
something like the earth.

It is possible to construct a switch.

So imagine you have a gate.

That makes a barrier go up and
down and let's forget about this and

of the peninsula let's
just assume we are some.

Liable to collimate the carrier so they're
always instrument in a certain direction

then but
because of the barrier property alone.

You can you can have a fine.

Transmission between the and
state of the asteroid.

Now you bring in the angle component and

let's say in one more of
the gate carriers are incident.

Normally in the other mode of the gate
Kerrison are obviously incident

then that presents you with even
more opportunities to have a better.

Arnulf ratio so I'm not trying
to get structure here too and

learn a little Imagine
that a little bit but

by having a gate that can control how the
carriers are coming in either normally or

an oblique angle you can actually control
the on off properties of the switch.

So this was meant to really highlight some
of the novel properties of graphene and

the opportunity present.

In thinking about new switches so
just using it as a replacement

channel material for a silicon master
is not going to do the trick because of

fundamental limits and that's why we're
looking at some of the fundamental

novelties which will lead us to a better
performance from a grapheme switch.

So it's probably been difficult for
you sitting through this with.

Without any slides and
just looking at the piece of.

Scribbled drawings on the board but

feel free to shoot me
an email with any questions.

If you have and
I'm also point two questions right now

Green

Power Point presentation morning video
of the market share or one of the down.

Well.

There's been one one or two demonstrations
offered one from folk I'm Scroope

at Columbia and
Goldhaber garden and stand for.

I wouldn't say it was a demonstration but
it's not the demonstration.

So what they show is the signature of
Klein tunneling and using some sort of

a tarp get it in a bottom good structure
and that shows wiggles in your I.V.

and there and they say that the wiggle is
coming from the sprinkler mechanism but

nobody has demonstrated
a switch out of the sea up.

I think the only constraint is
your collimation of carriers.

So if you're able to collimate these
carriers down to up to room temperature.

Then you should be able.

There's no problem.

You know there's that angular
dependence that I spoke about.

So usually carriers are not collimated
like a beam of light particles that

are different wavelengths whereas a laser
has all particles in a very narrow range

of wavelengths.

Similarly when you inject carriers at
a contact here they would have different

directions and that direction distribution
is dependent upon the energy of induction.

But it is possible to collimate
the carriers by using a super light

a structure or a point contact
a Quantum Point contact and

then all your carriers will be
inside in the same direction and

the Quantum Point contact would probably
not behave ideally at room temperature.

It's not known and

the super collimation structure also
we don't know how to behave at room

temperature but all of these we have very
well that a lot of preachers we know about

things all.