Thank you.

And again sorry for the delay here
is there a laser pointer somewhere

as you heard

my plan my hope hopefully sufficient
plan is described for you today.

Some of the processes we use for
the growth of advanced quite emitting

diodes all you're familiar with by
the main gods by now your own maybe

several hundred of them and
the question is what has changed so

that this room will soon be lit by
a ladies instead of these heavily energy

efficient things that Mr Edison
invented so long ago.

So this is a collaboration between
my group here at Georgia Tech and

our colleagues at
the Arizona State University.

I want to talk about very some
very detailed things here.

So if you have questions stop me I'll try
not to get too stuck on the details but

I will start with a general review of.

This material system which we call
the three nitride compound semiconductors

noise of OK.

And as you might imagine.

These are some a conductor
is that are closely

related to silicon but fortunately for
us there are different from silicon.

I'll talk about how we grow
these films of these materials.

These are very precisely controlled.

Structures with a couple of nanometer
layer thickness is required for

many of the device I'm
talking about today and

atomically abrupt interface
is very Thank you.

And I'll talk about one of the problems
we have with efficiency in these devices.

Namely as you drive them harder.

In other words you want to
drive them at current densities

that will yield enough light
to light up this room.

We find that the efficiency of the drops.

It's reduced to the low
current density regime.

Too complex.

There it is we call that group change.

Let's see what we can do to kill this.

So we're talking about droop and how we
can reduce the droop this is a fundamental

problem for Advancement of this technology
so that we can save the planet from

the energy consumption required to
light the world and right now lighting

consumes a large fraction nearly twenty
five percent of the energy in the world.

It's increasing daily because
people leave lights on.

But also.

Because there are more and
more applications for lighting.

In the third world people
are demanding more light.

Also you've probably seen examples
of forced Street Bridge there is

a rather interesting L.E.D. display
along that bridge which changes color.

Over time that series of L.E.D.S.

that are programmed to change their
relative intensity versus time.

So a lot of architectural lighting.

Has been put in place to exploit
the advantages of early days but

that of course increases the overall
energy consumption as well.

So we talked about this issue of group.

I'll talk about a very detailed
structure we're going to use

this turn area in the right it's not
something you'll find in grocery store.

You have to make it yourself in
the laboratory by combining.

The right precursors for indium aluminum
in one thousand in the proper way.

My grad students hopefully have
figured it out let me show you

the properties of these materials.

That cause us to be interested at least
the fundamental properties of the band get

the energy electron volts versus lattice
constant for this set of turn area and

Quaternary Eloise as you
know silicon is column for.

If we mix column five and

column three atoms together we can average
of column four and so everybody's happy.

We're going to get Korea
only bonded semiconductors

The interesting thing about these
semiconductors as you can see.

They cover band energy is from below
one electron volt actually point seven

all the way up to six
point two electron volts.

There is no other semiconductor

technology that covers anything
like this wide range of energies.

This of course represents
the visible range.

So you can see it's
an ideal alloy system for

making visible that is visible to humans.

You might imagine that the.

Important color.

It's where your eyes sensitivity peaks and
that's not surprising because that's where

the sun has a lot of energy for
you to see so humans evolved on this

planet with a certain spectral intensity
versus wavelength from the sun.

Spans enough we call it visible light.

And then up here we have
a region we call visible blind or

it's not there's a pole at all.

Something shorter than about
foreign enemy the I can't see and

then finally up here even higher energy.

Four point five electron volts or
so to a man a meters.

We enter the Solar Blind regime.

And that means that if you go out
today the sun is shining beautifully.

You take a spectrometer outside and

you go look for two inner photons
you won't find any of them.

There are no to any photons around.

That's not because
the sun isn't making and

it's making them in copious quantities.

But they're absorbed by
the atmosphere completely.

So if you're out side today looking
around and you see a photon from the sun.

It's from something else that
humans are doing probably

could be a fire could be a missile
could be something else going on.

Missile plumes have signatures
in this U.V. range so.

You might want to track some
photons of that wavelength L.-Y.

system therefore can cover
from binary indium nitride

that point seven electron volts in
the infrared again humans can see it

all the way to the ultraviolet people for
Violet and clearing the visible range.

Well if you gave us this gift and
then she pulled a trick on us.

She made it virtually impossible
to make what are you going to do.

The other is there trick she gave us is
this stuff is highly lattice mismatched

there's very little chance of taking
in the I'm not sure I'd been growing

it on a little.

Right because of the large
lattice constant.

Mismatch.

So for
stacking atoms in a regular periodic array

starting with a template based on aluminum
nitride you had stuff stacked around

maybe three point one angstrom
by the way these are words.

So these are not cubic structures
at all they have a Z. axis that's

over five angstroms this is the eight
plane a circle a plane that is constant.

Hopefully some of you know about
crystallography these are works I

crystals.

This is a plane that is constant and
this is what we would need for

example to fit one template on
the other if we were growing layers

on a seaplane substrate and believe me.

There isn't much forgiveness here.

So if you mess up and try to put too much
strain in your layer it self destructs

and so very careful design of reactors for
growing use gnomes has to be

made because you need precise control of
the L.A. composition at least for atoms.

Let's move forward.

To look at what options we have for
the basic template for an L.E.D.

based in this system.

We could go to the store and

buy a gallium nitride substrates like
you go by silicon or gallium arsenide.

Except you better have a big wad of
cash in your pocket because a two inch

subsidy is eight thousand
five hundred dollars.

So I tell my students the most expensive
piece because they they go through these

sort of regularly to not be thousand
five hundred dollars you know

personally get more experience out of
that but it's still pretty expensive.

The problem is that you can't buy
much larger than two inch today.

And as I said it sort of runs your bank
account pretty red pretty quickly if

you start to use that substrate you
can say well I buy aluminum and

try to I can maybe get
some advantage because.

Maybe it's cheaper to make.

But then it's of course
very highly strained.

Revenge again nitrate is in
principle you can let us match

this composition of the Quaternary and

the turn airy here in the moonlight and
turn airy this region in this little.

Moon shaped thing is quite here
his indium aluminum gallium in it.

So let's go buy some gallon I try to grow
a laser diode Well actually that's what

everyone does that makes
Blu ray disc laser players.

So you own one of those
by now maybe several.

P S three year Blu ray player on your hard
drive on your descrive on your computer.

Maybe a nice pioneer Blu ray at home.

To feed your twenty P.
seventy inch and he backlit L.C.D..

So you're all happy.

Well those are built on the substrate.

And they're expensive as I told you and

they're made of something around here
in the M gallium nitride in the violet.

Actually they run it for
zero nine animators so.

It's about right here in composition where
the active region of that laser diode

is operating and has a composition
of a little bit of indium in

gallium nitride So it's a turn area
about nine percent in the M. or so

if you choose to nitrites A Well I don't
care I'm going to go buy a lunar year.

You're more serious trouble because
your brain can't even not even worse off

a fifteen by fifteen millimeter substrate

will cost you five thousand
five hundred dollars.

Little square of material.

If you can buy them.

When you have a pretty nice
to people that make them or

you won't be able to buy them
then they sort of you know

they don't like bad people so they won't
sell them to you if you don't have to.

So you have those two choices.

Most people who make at least today.

Don't use either of these Think of
a store and buy sapphire substrates.

How big is sapphire Well
you can get six inch now.

A two inch wafer is about twenty two
dollars if you buy lots of them.

So big change in cost.

So in the commercial area of
application like L E D's and

he's you see in your car taillights
at least in traffic signals.

The green and there are different and
the red are different but

the green ones are all this stuff
operating somewhere around here.

And so it's a different alloy.

More like twenty percent in the I'm
on Sapphire that produces mostly D.'s

let's consider now what you
might want to do to make them

improve live L.E.D. structure and first I
want to tell you Will you make is made of.

Before I get into the details that
let me share what our tools look like

this is a commercial.

M O C D tool that I bought
in two thousand and

three when I joined Georgia Tech at the
time it was virtually state of the art.

These machines were in production in
Korea in Taiwan in Europe United States.

You can see here it chamber growth chamber
which has a lid in the latest flipped open

their pocket here.

You see machine in this surface is
a two inch diameter wafer pocket.

This is a graphite scepter and

that scepter is a three zone
resistance heater the top lid.

Has a very complex geometry.

That top little cost about thirty five
thousand dollars if you break one.

How do you break one when we
actually have broken them

show you why this has four thousand
one hundred Labor machine holes in it.

There are three point arms above this.

Surface different gas flow.

So there are three different zones of
gas flow above those they all feed down.

A given set of these small pipes that are
laser welded into this laser machined when

these other holes this hole here is used
for optical monitoring of these pockets.

These three holes so

many wrote on this you know how do you
train people to write on the board.

The screen and
I would have three holes here.

Are four optical monitoring
with parameter e.

We don't use those holes during groceries
this one to monitor during growth and

we can tell the temperature
of the scepter.

The growth rate of
the film on Sapphire and

all sorts of things that happen when you
change from material a to material B..

So if you want to measure.

Peanut butter layer in a jelly layer
they're going to give you a different

optical signal using the tool we have.

This whole thing is in
a nitrogen glove box so

you see this gloved hand here
that's not Michael Jackson.

That's my student.

We're going to rubber glove.

And over here.

This is the chamber closed and
you can see these three optical ports

plugged because we don't use them during
growth and this thing here is stacked on

top of that hole that's the optical
monitoring tool that we use for measuring.

The growth in real time.

So interesting labels here.

One of our tools has this other
thing hung on the top in this whole

colon at the curve.

We can measure the bowing of a two inch
wafer in real time as we heat sapphire up.

It tends to flex and
when you grow strained layers on it.

Like we are growing.

It flexes a different way.

And actually people have
movies of these things.

It looks like Tater ships.

Flexing in and out.

So you can make the sapphire that
you're using into a nice potato chip

fact if you heat this thing fast enough to
pay off the scepter so we don't do that.

All right let's move on to.

The commercial side of things today if you
want to store you wouldn't buy my tool.

That's only for universities these
days you buy something like this.

This is the same technology called the
showerhead reactor for obvious reasons.

Here's a showerhead is that got probably
ten of the fourth war weary area.

Holes in mind as thirty five two
inch wafers loaded on this chamber.

Then if you choose an alternate
technology commercially available.

Slightly different geometry
will go into the details but

this is not planetary reactor.

They're obviously.

A set of wafers on each of these
rotating disks there are six wafers.

In the on the periphery one in the center.

On six rotating this the larger
carrier rotates one way the disco tape

the other way.

So it's kind of rotation and
it averages out to growth chemistry so

we get uniform films and

both of these tools will run you
about three million dollars today.

This is a top view of that.

Tool to give you an idea
of what's happening.

Samsung.

Who as you know is one of the premier
L.C.D. T.V. manufacturers.

Making L.E.D.
backlit T.V.'s they ordered one hundred

reactors like this at one time and
they're installing more than that now so.

Korea is basically being
filled up with stainless steel

Taiwan in a similar Taiwan is going
to start sinking soon because

the weight of a stainless steel is going
to force it back down below the ocean.

Especially when global warming occurs.

So consider carefully if for
you guys land in Taiwan.

OK So these are very large reactors
you spend three billion dollars and

you get an elegy fab and
that small potatoes relative to silicon

Intel just spent eight billion
dollars on their latest fab.

So still of course much different
quantities of stuff that Intel's making.

Right now the technology going from this
two inch style substrate to six inch.

So this would hold six six inch
wafers sapphire wafers and

regrow simultaneously on those wafers the
more advanced tools that are available now

have robots that load these wafers
humans don't handle them anymore because

humans break things and
they're difficult to train.

Sometimes there is a list of
the current state of the art products

from four of the major manufacturers these
are the big four making white L.E.D.S.

this is a white L.E.D.
that you would use in your home.

It's a warm color one white Ltd.

Here's a cartoon view of one of them and
these others.

This is a creamy paste early
start up out of North Carolina

some four crazy grad students from N.C.
State started it many years ago.

They're now multi-billionaires or
something.

Happy friends of mine so
I get to talk to them.

The color temperature.

This is what we call warm white
it's around four thousand kelvin

equivalent temperature so he something
four thousand kelvin the globe is white.

This is the color you'll see this is
about what an incandescent bulb does

a little bit.

These are a little bit cooler
than this on these lights here.

This is the the voltage for

Operation fifty million or so three and
have three point seven volts.

At a current of three and fifty million.

That's the forward part of this diode
you can see they're all about that.

This is the luminous flux or
the useful light coming.

This chip.

At three hundred fifty million amps.

It's about one hundred lumens hundred
thirty lumens somewhere in that range.

An important measurement
of the performance is this

efficacy lumens per watt.

When you go to a store you go to
Home Depot you you know go and

find one hundred women like Bob right.

You say.

And once an equal into
a sixty watt light bulb.

So what the sixty watt light bulb
look like in terms of lumens per watt

is pretty crappy.

Actually very crappy over there much
better than any any incandescent bulb or

any fluorescent bulb you're going to find.

Let me now go into some
very exotic details.

Hopefully some of you have had some
basic semiconductor physics or

something like that.

Quantum Mechanics and
you understand that confining a particle

in a box like this is the potential
well and some particle in it.

Same lecture on a potential well has
a ground state in an excited state.

If I ply electric field so
that quantum Well this box.

I change the potential
energy across the box.

Because of the electric field in the
relation to the gradient of the potential

so constant years a green potential
You can see there is a shift

in the distribution of probability
functions for electrons in that box.

This is a box drawn for the holes in
a semiconductor this is a box drawn for

the electrons This is the energy gap for

that semiconductor and
you can see that in this case.

The states are four
holes are dramatically.

Confined to one side of the box.

Electrons are confined
the other side of arcs and

that's probably bad news for
electrons and holes to recombine.

You want their way functions
to overlap in real space.

So they're integrated probability of real.

Combination is high and this electric
field is detrimental to that process.

Sadly for us.

These materials are highly
placed electric and so

if you grow them in a strange fashion.

You get a free electric field and

it's it's a pain to deal with because you
can't do anything about it it's there.

So this is been one of the problems
with improving the efficiency for

commercial allergies beyond
the values I showed you that

the so-called quantum confined
stark effect in these materials

to the electric field which is due to
pierce electricity of these materials

you all have probably heard now many
times about the A long spears like your

generators zinc oxide generators and
things that's the same thing.

Think outside like these it's
very similar material and

we have at least as large pieces electric
fields in these structures as he has.

Another problem we have is a high
injection levels you saw those

characteristics I showed you.

At three hundred fifty million amps.

Well for light in this room.

We're going to use over two
amps of current So guess what.

This is sort of meaningless.

When used to put two amps through
a much higher resistance.

And you we have the effect shown here
peak lumens versus current driving amps.

And this is a large ship by the way this
isn't a small chip This is a one square

centimeter L.E.D. just in the people make
them one square centimeter one device.

And you can see what happens as
you increase the current here.

The efficiency of the peak lumens So
in the rules over

this plot is this the wall plug
efficiency to power lumens per watt.

And you can see it drops dramatically so
this is what we call group.

How do you fix Troop Well troop
is a is a quantum mechanical

problem we have to solve there
are many different aspects to it.

Let me show you a commercial easy again
this is our friends from Germany Siemens

luminous flux for a white L.E.D.

versus current notice it
goes to one point four amps.

This is a one square millimeter L.E.D.

So it's much smaller than that one
square centimeter really show you

these are sort of cool white five
thousand kelvin color temperature.

So this is something very close to
a fluorescent lamp in its color output and

again you see this is
a commercial drive it hired.

The luminous flux no longer
starts to roll over and

this is a plot of the wall plug
efficiency again lumens per watt.

Peaks here extremely low currents and
then just sort of tails off.

And I notice this is only one hundred
forty years per square centimeter.

Not really high current density.

But this is a problem.

We want to drive these harder when
you go to two amps or three amps.

You're obviously dropping a lot of power
and your efficiency is going down.

So this is the so-called
Golden Dragon plus

again from a recent website of our friends

in Germany from Ostrom German company.

And they use the same technology I showed
you on Sapphire to make these early days

notice their peak efficiency is around
three asterisk or a centimeter and

we're out here as a forty
seven percent drop in

the lumens per watt or
wall plug efficiency.

How do you fix that and
we've got fundamental problems but.

We also have some design
parameters we can work with we

can't really change the material too much.

That's emitting light.

Because we're stuck we want.

Actually these are blue really D's
with a yellow FOS for stuck on them.

So it's actually a very complex package.

What you see here.

Is a little bit of an example
there's a chip down here and

this yellow blob here is a special U.V.

proxy that takes the blue light absorb
some of it and makes yellow and red light.

So your eyeball thinks its point here.

This device made by us from
the yellow color here is

the foster put on top of the chip directly
and not in the and in the plastic.

All right so how do we fix this problem.

Well my students and I have been
working on fixing this problem and

so we've decided to look at using

a different turn area than everyone else
in the whole world uses as far as we know.

We're using indium aluminum nitrite.

Or Mind you that Lattice
matchable can be lies matched.

In our guy RAM here is here's
the kind of things we were growing.

Today in sort of cartoon view and

some micrograph of our structure
is taken at Arizona State.

You see.

Let's look at these layers
this is a thin layer.

Actually it's four point six nanometers
thick eight letters constants of

indium aluminum.

This layer the light layer
is gallium nitride it's

nine point two seven any mistake
it's eighteen that is constant.

So Kelly nitrite.

And there are several of these wells there
was a lower mag picture of the structure

typically or five wells in our ladies.

So how do we improve the structure let me
look at the structure itself starts with

the plain Sapphire the plain vanilla
twenty dollars apiece kind of thing.

We start with our first layer
is gallium nitride but it's.

Growing at low temperature.

This is entirely a silly thing to do.

It took years to figure this out
because it was sort of stupid

any real Christgau know
that's the wrong thing to do.

But by accident people discovered
it was the right thing to do.

Then we switched to higher
temperature to grow another where

I'm going to be a buffer layer for our
use of the word buffer it's something to

buffer the defects that
come from the sapphire.

Then there's silicon dope to layer.

This is an Indian gallium nitride
turn every layer of silicon doped.

These are all and type.

And then the quantum models themselves.

Kelly nitride.

In the beyond.

We call these two larger
bandgap materials.

The barriers.

And then this layer is one
we're going to focus on.

It's aluminum gallium nitride in most.

We're going to use in the moon.

Right right right there.

And the contact is made to P.

type gallium nitride up with magnesium
we use a lot of Indian players instead.

So those are some innovations
that we have put in this device.

Course the first job is to
focus on the quantum wells and

make sure they're correct and right.

Then we have to figure out our P.
type doping up here.

Consider what we need do with this
layer which is a critical layer.

Called electron blocking layer.

For those of you who have
had some device physics.

You know this and type region.

Will inject majority care luck Trons
from this side into the quantum Wells

this region will inject majority
care holes into the quantum well.

So holes come from the top
electrons from the bottom and

our goal is to keep the electrons
coming from the bottom.

From going over this these quantum
wells and recombining up here

where all the holes are so this is
a barrier and energy bearer for electrons.

So that they can escape.

Are quantum Miles.

So that's important.

This layer here also we have played with
because these quantum wells are highly

strained you are using another approach
to reduce the strain in those layers.

Finally if you're if you
really have lots of money.

You can consider substring Sapphire for

gallium nitride substrates
of various kinds and cost.

Let's talk about this
electron blocking layer.

Again you slide you've seen it
shows you the droop effect.

And here is one reason why it carries me.

Go from the inside to the peace side and
not recombine in the quantum wells and

therefore be useless in terms of
active layer carrier spillover so

injected carriers just go
right over the active region.

They spill over into the layer and

we never see him again higher electron
ability than holes for these materials.

Tells us that electrons move farther
on your given current flow so

they look I'm talking there is important.

We're going to use instead of
the standard aluminum gallium nitride.

We're going to use in the I'm and
this material which is typical.

Has some limitations in preventing
efficiency group as you can see from

these commercial devices.

So we're going to take our P. type layer

in our active layer where the quantum
is our and insert something in there.

Called electron blocking layer.

We're going to use two kinds illumined
gallon nitrate as our control and in

the image and I try to as the experimental
Tiriel all remind you what we're trying

to do this is our friend or enemy actually
the diagram I showed you earlier.

And here is gallium nitride.

If we grow aluminum gallium
nitride on gallium nitride.

It's under tensile strain.

That's probably not really good if
we grow this material right here

on the indium aluminum nitride Turner
line that's about eighteen percent

in the I'm that layer with the lattice
match to go in one thread and

therefore have no or
let's say produced strain.

Let me also talk about
the other part of this L.E.D.

and that is the active layers you
know these are blue and green.

So they're over here and

they're under the go forward ones
they're under compressive strain.

So this is our friend.

Our Savior the Indian I tried
eighteen percent last match to Gan.

We can grow the quantum wells
under compressive strain.

And we could also grow electron blocking
layer with the same turn area under

compressed strain in this compass range.

So this alloy that we're
talking about here

gives us the freedom of design in terms
of strain control strain management

that this aluminum gallium
nitride system does not give us.

And whenever you've got
trouble in River City.

There were options you
have to get out of town.

With your skin.

I'm on you.

The better off you are so.

We're going to try to explore that.

Let me show you again a detail diagram.

So I for this exotic.

Talk.

In some sense this is like trying to
energy diagram versus position for

the quantum wells in the light
emitting part of this device.

I showed you earlier the effect
of appears like a field.

That causes these things to have
high fields in their tilt at these

are the quantum wells.

You will.

There are things here.

These are the barriers the things
of the various Rosell dippy things.

With the aluminum gallium nitride electron
bucket where the red diagram applies.

And with our standard material
you can see there is a.

Where electrons if they have
a fancy can get over this barrier.

There's also other things that I want to
talk about that are that are negative

about this design but
if we use the lattice matched

in the overnight try turning right
which is in the black curve.

We have a larger barrier for
electrons melech transit coming from

this side the end region going over
hopefully landing in these quantum wells.

But not going over here and

high current densities the electrons
have a large enough energy to get over

these barriers if you're if you
have designed them correctly.

So here's our experimental structure.

Again in simple form.

We're minute we're changing this yellow
region here electron blocking layer.

We're going to grow.

Green and Blue Ivy's using
different electron blocking layers.

And we're going to use different P.
type players as well and for

control we're using
the standard twenty percent.

Well this slide talks about this
turning area last night again.

And this is a rather detailed slide so
I'm going to just move very quickly.

Basically we grew several kinds of
L.E.D.S. blue and green L.E.D.S. with

we characterize their I.V.
curve their electronics characteristics at

low current densities with direct current
and high current pulse current So

there are some students in
my lab who take our wafers.

And fabricate L.E.D.S.

out of them standard simple easy products
we can sell these on the street.

Because they're just not ready for
prime time that package at all.

Let me talk about the crystal
growth aspect of this.

Because growing at
Turnberry is not trivial.

It will probably give someone a nice Ph D.

actually because we're
the only ones in the unit.

You know of course you never
know what the Clintons are up to

you know I suspect they've known this for
a long time.

But you know they don't
talk to us much about this.

So it's hard to tell anyone
composition in human rights

versus growth temperature to
pressures in that chamber pressure.

Is a variable we use
in our growth chamber.

It operates in a reasonably useful way for
about twenty two or

three hundred tour and
then the gas dynamics of that chamber

sort of fail you and
you end up with well a mess.

You don't get the film you want.

So here's here's a serious I
tore data here's the data for

the same growth system using
very similar growth conditions.

You can see the higher pressure it
gives us higher Indian composition and

a given set of growth conditions.

So once we know this performance property
of the chamber and the growth process

we're using we can use an optimized high
temperature regime around forty or not.

My low temperature regime
around seventy eighty to grow.

The SAME electron blocking layer
with about the same composition

somewhere here around eighteen percent.

So we call the low pressure
low temperature version.

Surprisingly enough.

And that's a high temperature high
pressure version where you caught me and

Lisa.

But this crime more descriptive
not nearly as friendly looking no

here's what you really want to grow of
course is these layers that are atomically

smooth as there is lettuce primer
stick after all you've got too much.

And you've got you've got trouble.

So this is important.

They are most roughness in the animator's
there's a grow temperature for

different pressures seventy five to one
hundred two or three tours mentor and

you can see in blue.

Here is our senior tour data there.

Smooth layers for a wide range of
growth temperatures much smoother than

the low pressure versions so guess what
my students like to use this set of

conditions to grow those layers
because this data was taken for

a thick layer of one and a meter stick
the device has a twenty and your layer.

So we purposefully grow a thick
thing is the all rough.

It's going to be get a rough idea of
what it's going to be when it's thinner.

So another thing happens when you go
to these different growth conditions

as you realize from my schematic diagram.

The electron blocking there is
grown after the quantum wells.

Now if you start here and you grow up

through all this stuff you grow your
beautiful quantum wells they're perfect.

They're pretty green when
you do full luminescence and

then you make the device you grow
another wafer with this stuff on top and

your wafer turns black and
it doesn't emit light at all.

And when it doesn't light we call it a
dark emitting diode or a D.V.D. or a DEAD.

So we don't make deads LEDs.

Of course once you get into
this process you figure.

You know no world dark energy
is in your dead L.E.D. So

you solve the problem of the universe.

Whereas all the dark energy.

It's in our bad way for us probably.

OK So dads are not to be
not to be used for advice.

Here's an F.M. of the low pressure low
temperature turn area sample I showed you.

Here's an F.M.

of the high temperature high pressure one
again these are got on and in your steak.

This is the X. ray diffraction.

Rock in curved.

Symmetric.

Make a tooth a to curve for this.

This is the gallium nitride
peak this major peak here is

the peak related to the turn
composition these nice oscillations

that are picked by theory are Pendleton
fringes which are due to thickness

oscillations due to different index of
refraction at the X. ray wavelength.

The blue curve is experimental data and
the red is simulation and

you can see we can fit this.

Actually this is a block you may or
that we measured here.

This is this data fits an eighteen percent
in Jim Turner L I with twenty one and so

this is an actual lecture on documenting
we grew as a calibration layer and

I guess these are these
are twenty these I'm sorry.

These are twenty as well.

So this is what you would see
as your lecture blocking layer.

For the low temperature
low pressure version.

This is a high temperature high pressure
version of a two twenty meters thick

blocking layer.

So quite smooth point two
five nanometers are immense.

And that's not too bad for
stuff on Sapphire Here's some D.N.A.

from our friends that Arizona State for
these different structures.

One thing I'll start with here is
you see the five quantum wells like

I showed you earlier.

This is the case of the high temperature
turn area blocking layer this

light region here is that blocking layer.

It's basically on the scale
these perfectly smooth.

This is low temperature one and
you can see the quantum is again.

But a rather rough and strange
interface that corresponds to that F.M.

image I showed you in the last line.

So for a lot of reasons we think
this is a good choice for E.P.O..

I'm going to initial
time in my lack of being

prompt I'm going to skip this we compared
these ideas I won't go into detail.

Except to show you the data.

This shows you three
are standard early days.

All emitting about the same green
wavelength around three fifteen animators

and it shows you.

From five million amps in black up here
to eighty millions in this sort of

purple color.

These three devices this is the L.C.D.

with no A.B.L. this is one of the in
the human rights or low temperature and

this is a high temperature one so
you can see we get much more light out

one point six times that of the device
without the election by here.

This tells us that we are indeed blocking
electrons in this case they just

went over the quantum wells and landed
in the air and didn't get any photons.

So those were dead electrons.

Here we capture a few more and
got some photons out.

And if you plot to condense
you get the same sort of.

Interesting dependence to bust your.

These with high temperature high
pressure back in there now.

Let's compare the standard device.

So we'll do that.

Here's a three structures are going
to compare the with no blocking

layer a little gal and that's right.

This is sort of the commercial standard.

This is our turn every invention
in the United has a comparison.

You can see again the winner
is our friend in the room.

Much larger than the same structure
growing the same reactor by the same hands

with Ghana.

And these are pulse mode measurements they
go up to four in twenty million amps for

a small fifty microns square L.E.D..

And if we plot these integrate you know

intensity versus injection
current density.

You can see what happens here and
high current densities.

Device out the bell.

The light saturates just like through
the case of a stream as I showed you with

the Allegheny device it saturates
as well like the awesome.

And this new turn.

Here you know we're going to be able
does a much better job at high current

densities Now remember I showed you the.

Awesome data.

Moved to this which was at one hundred
forty Astra square centimeter back here.

So we're testing is that higher
current densities then us from there.

And this shows that same data plotted
quantify versus current density for

the black case the no you the blue
case the Allegheny bell and

the red case the in human
right again you see droop for

all three but only eighteen
percent up to three and fifty and

three square centimeter actually
about three sixty for the case.

So we see improve performance and again.

I remind you that Ostrom had tested
there is up to only about one forty and

they saw a dramatic decrease in
the quotation see forty seven percent.

And of course in our case.

We've only dropped

a few percent for the bucking layer L.E.D.

that when forty amps
are square centimeter so

here's another interesting thing
we look at the peak current and

see other pecan efficiency for
those three cases and again we found.

A higher peak current density for
the peak on efficiency.

Which again we think is due to
the improved performance so

it's not a blocking there so
I've done this quite fast.

I hope you've gotten something out of it.

I think overall we believe that.

The nitrite really the market is
going to continue to take off.

It's going to be billions of dollars and

you know there's a joke about how many Ga
Ga students takes to change a light bulb.

You heard that probably use
one to hold a light bulb and

to him to turn ladder around and
they have to screw with my palm in

that's what turns green programs going to
generate against that kind of engineer.

At some point in their future normal.

The standard joke is you never place.

You know you remodel your kitchen
you put your lights in but

you never change a light bulb that
stupid will last longer than you do

you know you'll die was a mite bald.

If you're born with at some point.

So thanks again for your attention.

Sorry for the delay This is
a picture of one of our ladies under

test making a green L.E.D.
this is actually a blue laser dye we made

in my group operating cedar
room temperature in the blue.

This is the far field pattern
showing that a fraction of the beam

from the quantum Well part of the laser.

So happy to answer questions you can
email me if you want to come visit or

show you some tools.

I'm just buying a new reactor that's
going in another part of my lab for

one point three million dollars It's
a cheaper one hundred ninety three million

but it doesn't matter it's doing things
that these tools don't do it goes to

higher pressure and
much higher temperature.

So we're going to get to more
parameter space with that.

Thank you for your patience.

Then you know apologize and
come off the wall in my schedule I'm

leaving tomorrow to go to put on a quest
so I know I'm just very bad things to do.

Sorry.

Thanks everybody.

Yeah that's for a wax material yes right

where

me or is it strictly primarily its
composition because we have all.

Well you obviously can
change the composition and

the piers I could feel
in the quantum wells.

We've looked at that we have ways of
virtual limiting the piers like feel

in one corner while.

Well not all of them yet.

That's another adventure we have to
undergo but there is an option and

I sort of passed over all of this
work that I showed you is on seaplane

base a plane sapphire if you say
well heck I'm going to use plane or

a plane or
even some crazy semi polar plane.

Like I know you might try
something weird like a semi polar

between and see people understand

the piers actually fields are orientation
dependent because the polarization

comes from this large city between
a gallium one thousand atomic bond.

So if you rotate the sucker so
there's gallium nitrogen in the plane.

So averages out.

So that's why I'm playing so good and
plain has no place like it for you today.

No one has shown that employing L.E.D.S.
any better than a seaplane.

Even in spite of a lot of effort.

Although I should say some
people who I don't trust

will tell you that they have solved
that problem and they can show you data.

That sort of fishy data it's you know
it's what I call the P.T. Barnum.

Version of science.

There's a sucker born every minute.

And if you're careful you can sucker a lot
of people and even though the experts.

Are there looking at your data but

it's your data and
some of it can be manufactured or

massaged that's why papers get removed
from journals occasionally right.

So I'm not accusing anybody but
it looks funny to me.

But yes.

Now if you want to say OK I'm going
to make a company makes employees.

How many million dollars you need for

substrate you can't buy
a two inch diameter wafer.

The same plane and if you could buy
a piece of semiconductor plane.

I have some of them.

I'll show you there are only fifty
millions long and five million.

And they cost me over two thousand
dollars so a pretty high price spread.

You go in if you know
that seaplane samphire.

And if you go to employ Maine.

You're really going to spend
a lot of money away for

bonding virtually all the ideas you buy

today have something short of wafer
bonding they're actually subject moved.

So you take the sapphire off after you
grow this L.E.D. you take a high power

laser you raster it over the back of
the wafer through the sapphire U.V. or

infrared laser you melt the gallium
nitride at the interface.

And I think pops off and

if you're clever you do it right the whole
two inch think pops off in one piece.

If you do it wrong it pops
out in a million pieces and

you've got nothing left because you've
got a very large gradient as this laser

scanning through a two inch wafer and you
can illuminate the whole wafer at one time

you get a focused laser down and scan it.

So people do it.

They make money at it and
it's a production process so

you'll do pretty high.

Now but it does take some care.

The way they are you
know that's a deep and

dark secret Probably someone does do that.

Maybe everyone.

I just don't know about I have my students
work for me lives in San Jose doing this.

They don't tell me much but
time and a lot of secrets they're

like in one year.

Yeah I know you are on

Earth out of the way right
now where my grad students

are not running at twenty four seven and
they're not pushing the envelope.

So until you do you like I showed
you takes four or five hours.

Do.

In production.

I'm sure it's down at least half of that
because they're running these suckers

higher growth rates and they're they're
running a hot to the reason for

example you have robot loading or wafers.

You never cool a scepter below three
hundred seventy grain you load hot.

You don't have that wasted time to
heat that big mass that is a big mass

of graphite you have
heat cost you money time.

So a lot of it.

Five hundred degrees or
four degrees centigrade.

And of course that sort
of burns your fingers.

If you're not careful.

So robots do loading heart for

example when you save time and all

that well

there's no subpoena streams are indeed to
put other layers in like we're talking

about but yeah once it's in production
unproven and under stable control.

It's just it's a small amount of
cents per chip maybe hundreds of

cents per chip who knows.

Again the part of the cost is just

development the chemicals
themselves aren't cheap either.

In fact right now there's a deficiency
of trying to gallium on the planet.

No one can make enough of this compound

we used to start to gallon precursor and
so

people are going crazy chemical engineers
across the planet going crazy trying to

build these plants that are more efficient
but this is not garden variety or

gamma tele chemistry this is all tri
parity six or seven nine they're going to

metallics power for chemicals things
you really got to be careful with or

you end up dead or they end up
being unsellable because no one's

going to buy a crummy time out
the gallon you can forget about it.

It.

Yes.

OK Well one simple explanation for
the blue versus green and the green having

more droop than the blue is that there's a
higher place like a field in the Green One

of miles and we know that separates
the charge carriers more and as we fill up

these quantum wells we just don't there's
no place for those electrons to go.

They bubble right out because
they're up at the top of that.

Well.
So you start Philip that well.

You fill up the first you go second state
suddenly there bubbling over the barriers.

And the green well seem to have more
non-uniformity is in the composition

as well.

This is not a perfect alloy
It's a pencil you believe it.

It's actually something it will be fairly
locally composition in compositionally

varying alloy with with quantum dots of
Indian gallium nitride surrounded by other

and that was part of the lore early on I
think that's pretty much been debunked by

using the way

we believe versus Well

here's the bottom line how
many lumens per watt you yet

if you take a blue photon
pump a yellow Foster.

You're giving up energy.

If you want to assign the world's best
R.G.B. T.V. with the lowest power like

the Cowboys Stadium has ten point
five million L.E.D.S. in one T.V.

in Cowboy Stadium use goes red green and
blue individualities

because that's the only way you get
the energy efficiency you need.

So you want all the white you actually
got to go back and take four or

five different chips and combine
them to get the best power fish and

see promise to cost you a whole lot.

So low cost solution is take a blue L.E.D.

let some blue light
leak out put the right.

Magic sauce on it right.

Foster that has both yellow and red and

maybe a little green even and
get a warm white.

Like these.

The reason these are so warm is the red so
the red has been the trick for

phosphorous technology to get it
efficiently generating red light and

of course again taking a blue.

Photon and creating a red photon you're
giving up all sorts of energy when

you do that transformation.

It goes into heat.

Ouch.

Yes you

know right now it's not the problem
because most of the structure as

I showed you is gallium nitride
there's a smidgen of Indian here and

there very thin layers for
nanometers to stick.

Maybe five of those so

your big problem right now is
where you get the gallium and.

Several years ago gallon was so
scarce and so

valuable that there was an interesting
deft attempted in Russia in Russia.

There is a large pile of gallium at
the bottom of the mine it's used for

neutrino detection and

some guys a question cough when
they're trying to get the gallium out.

Now what how they're going to get this.

Hundred kilos or more a gallon out of
a whole new ground I don't know but

they went in there raided this place.

And they had police came in and
kill them all but

they were after the gallium
in this detector.

So you know.

Hope doesn't get to that but
Kelly it was hard to find on the planet.

First of all but

try mentally gallium electronic purity
is where the hen's teeth right now.

You don't have a friend making it.

You could be in trouble.

Samsung has a locked in
contract with several vendors.

So they've got a supply.

And that user bells hang and dry
are difficult paying high prices for it.

They're sellers do selling as three or
four times buy some celery and

sell you are OK thank you.