0:00

In this video we will consider the important process of ion implantation.

a process that is an indispensable part of the fabrication of modern MOS

transistors. And specifically, we will concentrate on

one application of ion implantation, namely the adjustment of the threshold

voltage value. The main uses of ion implantation, are to

make source and drain extensions. Which we have seen, and will see in a

moment again. to make Halo regions.

Which we have also seen. And to adjust the, the threshold voltage

value. Here is a device with source and drain

extensions, shown here and here. I remind you, these are used to avoid

having the bulky source and drain regions, which have to be bulky to give

you low resistance. From being too close to the channel,

because then that would result in charge sharing and dibble effects.

In addition we have the so called these halo regions which are regions doped

heavier then the substrate and this helps contain the depth of the depletion region

for the same reason. To, to reduce the short channel effects.

And finally there is an implanted region here, which modifies the substrate doping

in order to adjust the value of the threshold.

And we will say more about this in a minute.

Now we will discuss implantation which results in a change of the doping

horizontally in this picture, in parallel with the surface.

And this will be called the lateral direction and also implementation that

changes the doping vertically in this picture or in the transverse direction.

Now if we make a device with the same extensions and the same halo adjustment

with a shorter channel length. Then these two regions will apploach,

approach each other. And you see that they kind of merge, and

now the entire substrate below the channel is highly doped.

This effect will be important to incorporate of course in the threshold

voltage because the threshold voltage depends on substrate doping.

We will see how this is done shortly. Here is a three dimensional picture of a

ironing planted device. the vertical access is dopping

concentration. And the x, dimension is the horizontal

dimension along the channel. And this one here is the depth of the, is

the depth below the surface. So the, the surface corresponds to y

equals 0, unless you go down below the surface, it's like this.

Now, the doping of the source and terrain regions is high, near the surface, and it

goes down, for reasons that we will see. Because these regions are implanted and

the implant results in a variation of the doping concentration with depth.

When the N type doping becomes sequel to the P type doping concentration of the

substrate. The two cancel each other out and the net

doping goes to 0, which on the logarithmic axis such that the one we

have here, means that this actually goes down to minus infinity.

So the this dip here defines the juntion boundary.

Now the substrate doping is shown here, you can see it is somewhat higher than 10

to the 17th centimeter to the minus 3. This is a threshold adjust implant next

to the surface these are the extension regions.

And these are the halo implants, and as I have already mentioned if you now make

the channel length shorter now the halo implants approach each other and the

substrate doping becomes large throughout the region under the channel.

So let us now talk about the so called High-Low Profile.

4:12

To motivate, let us start with an un-implanted device always, and most it

has a flat pan voltage given by Fie ms minus Q0 over Cox, as we know.

for an N plus poly gate, pi ms is about minus one volt.

with the clean processes today, Q zero prime is practically zero, and you can

neglect it.so the total value of Vfb turned aught to be about minus 1 volt.

Now the threshold voltage turned aught to be V plus V0 plus gamma squared 2 V0.

Now we want this to be positive, let's say we need it to be about 0.4 volts

because we want to make sure that the device is off, for digital applications

when VGS is zero. want VGS minus Vt to be as large as

possible to give you a large current when the device is on, you don't want to make

Vt zero very large. So 0.4 is a compromise between too low a

value and too high a value. Now, if this is minus 1 vote, be at least

minus 1 vote and you want the results to be plus 0.4 if you make a quick

calculation, you'll find that gamma must be very large.

And to make gamma very large, you would have to use a large substrate doping, and

then undesirable things happen. First of all, you have a large buddy

effect because of a large gamma. And you also have large junction

capacitances is because the depletion regions are narrow and the junction

capacitances are large. So ideally you do not want to dope this

up straight heavily throughout it's extent.

A better way to do it is just change to 0 if you can.

So in other words you make this one to 0 prime negative and with another minus

sign here becomes positive. And it makes the total Vfb less negative

and can give you the right value of the threshold.

Unfortunately, there's no way to just change to 0, what you do is you attempt

the p implant, which should be as shallow as possible to add negatively charged

ion. Which is equivalent, if they are very

close to the surface, to make Q0 prime or negative.

6:25

So the result then, is a device with an implanted region.

And in this simplified picture, I show the implant as existing over.

The finite depth, and that is a clearly defined boundary between the implant and

the substrate. The implant characteristics are the

effective dose, which tells you the number of implanted ions per unit area.

Which is 10^12 per square centimeter of gate surface area.

And the average kinetic energy of the ions, the way the ions are implanted is

they are accelerated via an electric field, then they pinch on the surface

they enter with high kinetic energy. which they gradually lose as they hit the

crystal lattice. So, the kinetic energy, as they leave the

implanter, is measured in tens or hundreds of kiloelectron volts.

So these are implant characteristics that will help you determine the final implant

you'll end up with. And if you plot the doping concentration

versus depth from the surface, you get a constant concentration for the, substrate

that you start with. And then you end up with a distribution

of implanted ions that looks almost like a Goshun like this.

And sub i, which is a function of y. Now both of these implants are of the

same type, so you add N A B The initial substrate doping plus an 1 and at the end

of the total doping after the implant. Now because this is too complicated to

handle in simple modules the whole thing is approximated be a step like this so

the final value of the step is NAB, the un implant that the substrate doping.

And this one here is Nab plus Ni, where Ni is a constant value that tends to

approximate this varying profile of Ni of y over here.

So this is called high to low as you go from the surface to the bulk, you go,

start from the high concentration go to a low concentration.

Now, let's assume strong inversion operation.

as you remember, the as you increase VSB, because of the body effect, the threshold

increases. Now, initially, if you assume you have a

channel next to the surface, initially if you don't have much of VSP, the depletion

region depth would be small. And the entire depletion region would be

inside the implant, which corresponds to high substrate doping, and then you can

expect large body affect. But as you increase VSB eventually the

depletion region extends outside the implant.

And now as you vary VSB, the bottom of the depletion region moves inside the

lower doped substrate, in other words, where the doping is only NAB.

And then it turns out this is equivalent to a smaller body effect.

9:31

Now if we denote by V sub I, the value of VSB at which the bottom of the depletion

region reaches the bottom of the implant, in other words, depth d sub I over here.

Then, when you plot VT versus VSB, it turns out you get two branch curves.

Which can be represented as one value of a the threshold versus VSB, when the

implant, when the depletion region is inside the implant, and another one when

it is outside. So you may have two different

expressions, one for VT1 and one for VT2. And use one or the other, depending on

the value of Vsp. But in reality, because the actual doping

profile is not a step profile like this, but it is like that, you can get equally

good results, or even better, by using a simple modification to the classical

body. effect expression, which is this one.

You simply add the term K2VSB, you made K2 negative, so what this does is when

VSB is large, this term becomes significant and subtracts from what you

would normally have which would be continue like this.

Subtracts from that and gives you an approximation with this behavior over

here. So this model is widely used.

10:54

Now let's talk about the opposite case, the low-high profile.

now instead of an acceptor implant, we use a donor implant.

Which means that if we have a p-, p-type substrate here, we have, we implant Donor

ions, and depending on the relative concentration between n and that the p,

the final implanted region here can end up being p with lower doping

concentration, or even n. For now we will assume it stays p so we

have this is the initial substrate doping, this is the implanted donor

concentration. So, you have to take this and subtract

that, because they are of the opposite type, to get the overall concentration.

And again, you can approximate this like that for a simple model.

So as you go from the surface towards the balk?

You start with the low and you go to high.

Now this is why this is called a low high profile, the final, the approximate

doping concentration you end up in the implant with the difference between the

initial substrate concentration you started with minus an I who and I is an

approximation. It's an operative value over here.

The threshold is still given by the same expression as before, only now because

you encounter the low doping and then the high doping has VSB increases and

eventually the depletion rate can extend outside the implant.

That is when you see the large concentration and gamma becomes more

important. So this k2, it has now a positive value

for the low/high profile in order to make this one an effective correction to the

rest of the body effect equation. Now, real profiles of course are as I

mentioned. The approximate Gaussian curves.

They are too complicated to get manageable analytical results, so people

use approximations. And these approximations work because of

several reasons. First of all, the variation of the doping

with depth is gradual. It's not true that as you increase Vsb in

the depletional region. Becomes deeper and deeper.

Suddenly, you reach the limit point between the implanted and the unimplanted

region. You just see a gradually changing doping

concentration with depth, nothing drastic happens.

Also the details of the Qb variation are obscured by the integral in our drift

equation. I'll remind you that the drain current

has a drift term and a diffusion term. And the drift term looks like this.

And what matters int his current is not QB itself, it's only the integral of it.

So the details of QB prime are obscured inside this integral.

The diffusion equation looks like this. Here, you do see QB by itself.

But in both of these equations, the main terms, which is the first-order term here

and the square-law terms over there, the first-order term here.

All of these terms do not involve Q'B prime, and that is why If you don't model

Qb prime very accurately, you can still get accurate expressions for the count.

Now's some models use an effective doping, which is made a fraction of the

surface potential in an effort to model implanted devices.

Now If you do this, if you've got an effective doping that depends on the

surface potential, effectively you're making the body effect coefficient a

function of Vsb. This is just a mathematical convenience

to help you model the device. In other models, what they do is they

make an effective doping a function of Vgb.

Again, this is just a mathematical convenience.

Of course, it doesn't mean that the concentration itself changes as you

change VGB, but in equations that we used to have that relates VGB to the surface

potential. there is a term that depends on substrate

doping. They make that term a function of VGB.

And now for a given VGB, you can use the same equation to get the value of the

surface potential, which we used to do for the old region model.

Some problems that can occur with the iron implantation, if the device is not

designed properly or sometimes even if it is.

you can get a log I versus Vgs, here we emphasize the weak inversion region, and

when you have a large Vsb then you get the curve on the right, here.

And this happens because for large Vsb the bottom of the depletion region is

outside the implant,it is in the lightly doped substrate.

And a lively dope substrate means that you have a small gamma, and therefore a

small factor n, which affects the slope of the weak inversion region, so it looks

like that. Now, when VSB is 0, the depletion region

is shallow and it is inside the implant. And then gamma, because of the large

doping consideration there, gamma is large and n is large and you get a

smaller slope. But in addition, as you vary VGS the

depletion region depth varies, an it moves over various parts of the implant.

In other words, it goes over, a region where the doping concentration of the

implant gradually changes. From high to low.

And, therefore, as you increase VGS, the slope even changes.

So when you see things like that, there may be an indication that the implant has

something to do with such effects. Now, pMOS devices are like nMOS devices

but here they have a p type source and drain, an n type body.

And when these regions are implanted, the pole is also implanted, so the pole

typically p, p plus. And if you implant.

This one with extra donors you can adjust the effective body doping near the

surface and modify the threshold of the pMOS devices.

I would very briefly like to mention buried-channel devices.

These are devices that are not used today.

They used to be used for, to make depletion-mode devices, and in device

research they are being considered even today.

Now, the implant type for such devices is the same as low to high, you remember to

get the low to high profile. You start with a, a P-type substrate and

you implant donors. But now, in this case, we implant with

even more donors than we have acceptors, so in other words, the implant, here, the

peak of NI is above NAB. This is the substrate concentration.

This is the implant concentration versus depth.

So then in order to get the com, combined concentration of the two, you have to

subtract the two. And because this is larger in [INAUDIBLE]

than this one, you end up with a net donor concentration near the surface.

And and accept the consideration in the bulk.

So, the device looks like it is p type in the bulk and N type near the surface.

Now, depending on the value of the gate potential, you can even deplete the

surface here. And then you end up with a channel that

is formed by the n type region, except for the depleted part near the top.

So you have a channel where the current flows below the surface.