字幕表 動画を再生する
>> Sticks.
All right, but we're going to talk about spins today.
Not about sticks.
So I want to continue our discussion of the concepts
and theory behind NMR spectroscopy and again,
this is not going to be about math or anything to that extent.
But we're going to be thinking very, very qualitatively.
All right, so when we last left things we had said
that there are two spin states
for a dipolar nucleus like a proton or C13.
There's spin up and spin down,
and if you have an applied magnetic field there's a small
energy difference between the spin up state,
what we're calling the Alpha state and the spin down state,
what we're calling the Beta state.
And because that energy difference is
so small unlike IR spectroscopy or electronic spectroscopy,
UV Vis spectroscopy where the energy differences are very big
and all of your molecules are in the ground state,
here there's only a miniscule number of nuclei
in the lower state more than the number in the upper state.
We said if there are -- if we take 2 million protons,
out of those 2 million protons depending
on the applied magnetic field, it will be 50 or 80
or thereabouts difference in population.
In turns out that difference in population is going
to be extremely important because it is only
that differential population that's going
to be able to get us a signal.
All right, so if we think about things
in an XYZ coordinate frame, and I'll talk more
about NMR spectrometer in a second and how it works.
But imagine for a moment we have some coordinates,
so the X coordinate is coming out of the plane,
the Y coordinate is in the plane and the Z coordinate is pointing
up and we're going to have our applied magnetic field BNOT
pointing upwards.
That kind of makes sense.
These super conducting magnets are always vertical
because you've got this big pot of liquid helium surrounded
by a vacuum vessel, surrounded by liquid nitrogen,
surrounding by a vacuum vessel.
And, those small amounts of population,
that small differential of population with spin up is going
to give rise to a net magnetization.
Now, in other words a way to think about this is for most
of our cases we're going to have one nucleus pointing up,
one nucleus pointing down in their spin
and there's no net vector here.
Those vectors cancel each other out.
But for that small differential of access vectors you're going
to have some net magnetization along the Z-axis.
Now, the ay it works, when you apply a magnetic field is you
actually have those vectors that are processing around.
So in other words they are processing
at that resonant frequency, at the Lamar Frequency
at 500 megahertz for 117,500 Gauss magnet.
So we actually can represent this by saying okay,
we've got spin sort of pointing in every which way,
and I'll just draw two directions.
They're all processing around.
So remember, remember this is only the differential population
that we're worried about because already for every one
where you have one up and you have an opposing one spin
down those vectors are going to cancel each other out.
Now the other thing is they're not quite on axis.
In other words they're not like this.
It's like a gyroscope if you've ever hung it from a string,
the gyroscope doesn't --
who's hung a gyroscope from a string as a kid?
The gyro in physics lab or something.
The gyroscope doesn't hang vertically,
it kind of hangs off axis and goes around like this.
But if you think about it, since those spins are not bunched
up for every one that's processing
like this there's another one that's opposite it.
So in other words if were just like this you'd say,
oh there's a net magnetization along the Z axis
but also a net magnetization along the Y axis.
But there are other spins that are like this
and they're all going around.
So everything is canceling
out except the net magnetization along the Z axis.
All right, what I want you to imagine right now is
that we're going to place a coil along the X axis and we're going
to put energy into that coil.
We're going to apply a magnetic force.
And I want you to think classically
because the quantum mechanical thing is going
to be we'll flip the spins.
I'll show you that in a second.
But you have your net magnetization along the Z axis,
and think back to classical physics.
If I apply a force along the X axis, right hand rule and all
of that good stuff, we rotate our vector downward.
So after we apply a pulse, I'll just say a pulse.
If here's out net magnetization,
when we apply an RF pulse our net magnetization moves along
the Y axis, and so I guess if I want
to actually represent it I'll just say XYZ,
and I'll say here's our net magnetization.
And as you'll see in a moment we're going
to have continued procession.
And again, if you're worried about the fact that all
of our vectors are not lining up, that they're all processing
like this, just think as I apply a pulse
and drive my magnetization from the Z axis onto the Y axis,
the vector sum is right along the Y axis even though there are
some that are like this, I drive it down.
They're countering each other.
There are some like this, I drive it down.
They're countering each other.
And so our net magnetization ends up a long the Y axis.
Does that make sense?
All right, lets come back to our spins to see what this means.
So, the way I was trying last time
to represent this very small difference
between the Alpha state and the Beta state was
to show some vectors, some spins pointing up in the Alpha state.
And some spins pointing down in the Beta state,
and to try to represent this miniscule difference
in population what I did for the purpose
of my drawing was I drew 6 with spin up in the alpha state and 4
with spin down in the Beta state.
>> [Inaudible] nuclei right?
>> Those are representing exactly the spins
of individual nuclei.
So in other words, if we had a mole of --
or more realistically if we had a millimole of CHCL3,
proiochloroform [assumed spelling] in our NMR tube,
what this would represent would be the different,
the nuclei of the hydrogen there and we would have out of
that millimole of nuclei we would have a small access
in the Alpha state and they would all be processing.
All right, so if we apply an RF pulse, and now I'm going
to be a little bit specific, if we apply a pulse long enough
that it's what's called a 90 degree RF pulse or a pi over 2.
That's just radians and degrees, your choice
and they get used interchangeably.
What that does is it equalizes the population of Alpha
and Beta state, so I'll represent
that by 5 spin up and 5 spin down.
And this situation is exactly the situation that we have
at the end of my little drawing
over on the left-hand blackboard.
In other words here's our net magnetization.
And so the key is now we have no net magnetization spin up,
no net magnetization spin down,
but the very important point is we have the net magnetization
focused along the Y axis.
It is not diffuse, it is not pointing in all directions.
We actually have net magnetization in the XY plane.
And if we apply a longer pulse, a more powerful RF pulse,
so again I will represent our 6 little arrows
and 4 little arrows representing our differential populations
of the Beta state.
If we apply a more powerful RF pulse,
what we call a pi RF pulse
or a 180 degree RF pulse I can invert the population.
In other words I will represent that by 4 arrows pointing up
and 6 arrows pointing down.
And if I want to draw that on my diagram,
can anyone tell me what I do with my net magnetization
on my little XYZ diagram at this point?
>> Down.
>> It's going to point down, exactly.
All right, and this is the damming thing.
Well, one of the many damming things
about NMR spectroscopy is no matter what you do
with your pulses you are limited to the difference in population
that occurs between the Alpha and Beta state,
and later when we start to talk
about 2D NMR spectroscopy we're going to learn about one
of the common techniques now, which is to go ahead
and have polarization transfer.
Now, think about what I said before
in the equation relating the Boltzmann Distribution
to the energy difference.
And remember how the magnetogyric ration
for carbon was a quarter the magnetogyric ration for protons.
That means that roughly the Boltzmann Distribution is going
to be a quarter as big on differential population
for carbon as it is for protons.
So when we get into techniques like HMQC,
which is a two-dimensional technique one of the tricks
of this technique is to transfer the larger
but still miniscule population difference
from proton to that of carbon.
But again, what's damming is you never can get away from the fact
that out of 200 million --
out of 2 million protons at 500 megahertz there's no way
to exceed that I think I said 81
out of 2 million population difference with the exception
of some very specialized techniques that involve
for example unpaired electrons and free radicals or xenon atoms
for that matter, and special optical techniques.
All right, so we have our differential population
and we know that if we apply a pulse
of the right length we can drive that population
to have a net magnetization in the XY plane.
Now let's take a look at how we get a signal
out of the spectrometer.
All right, and again we have a coil and I'm going to represent
that coil as being along the x-axis.
That's a little bit of an over simplification.
And now we have our net magnetization in the XY plane.
And as I said, it processes and so I'm trying,
of course it's hard in three dimensions.
I'm trying to represent this as procession in the XY plane.
I'm trying to represent that with a curvy little arrow
but what I'm really saying is you have your net magnetization
and it's moving around, those vectors are moving around
and we'll just take a single nucleus like chloroform.
It's processing at the frequency,
we'll call it V, call it new.
So like at 500 megahertz sometimes it's called the
Lamoure Frequency.
So for example at 500 megahertz for a 117,500 Gauss magnet.
NMR spectroscopy was actually discovered by the physicists
and rejected by the physicists because they figured
that there would be this universal property of a proton,
of how fast it processed in any given magnet and we --
when we come next time to the concept
of chemical shift we'll see that the frequency,
the magnetic field that the proton field is modulated
by the nature of the molecule, by the environment
in the molecule hence different types of protons process
at different frequencies.
Chloroform processes at a different frequency that TMS,
the CH2 groups of ethyl alcohol process
at a slightly different frequency than the CH3 groups.
These differences in frequency are very, very small
but they were upsetting to physicists
because psychics figured this has to be universal property
of protons and when they saw it varied by magnetic environment
in the molecule they gave it the contemptuous term
"chemical shift" [laughter].
Anyway, so okay.
So what happens if you have a coil
and you have a magnet rotating in that coil?
Think back to your physics.
Well eventually you'll have -- we'll get to relaxation
but right now lets just imagine--
[ Inaudible Speaker ]
Anyone ever done this again, as a kid?
Where you take a magnet and you spin it in the coil?
[ Inaudible Speaker ]
An electric thingamajigy.
[Laughter] a current.
Yeah. Yeah, we get a current in the coil, and this is the basis
for all of this stuff.
So while the nuclear generator are working hard
in San Onofre making steam,
what they're basically doing is turning a magnet inside a coil.
In practice it's done with armatures and wires,
but it is the exact reverse of a motor
and the simplest motor you can make involves taking a coil
and taking a magnet on an axel and putting
in alternating current in.
And the reverse, when you spin a magnet inside a coil you get an
alternating current, and so the current looks something
like this.
On the X-axis I'm going to plot voltage
and on the Y-axis I'm going to plot time.
And an AC current is simply the voltage oscillates
in a sinusoidal fashion.
So we get this cosign wave and it goes on.
You can call it a sine way if you want,
and technically I guess I should be at the peak
of my wave right here.
But what that's saying is as the magnetic --
as the vector is like this is relation
to the coil your current comes to a --
your voltage comes to a peak and then
as it gets along the negative X-axis you come down and it gets
to the negative Y-axis and it comes up
and for the 117,500 Gauss magnet we're going ahead
and having this coil,
the frequency be 500 million cycles per second.
It has to be in the XY [inaudible].
[ Inaudible Speaker ]
That comes from the pulse that we apply, so remember we have
out net magnetization along the Z axis.
We apply a pulse that pulse drives the magnetization
into the XY plane along the Y axis and we get procession.
That pulse is equalizing the population of Alpha
and Beta states and doing
that is basically making same number spin up, same number spin
down but it's directing them together in one way,
and in practice when you apply pulses they actually come
in fours.
So for those of you who've run a spectrometer --
how many of you have run a NMR spectrometer?
And you do number scans as equal to four or eight or 16.
That's no accident because we are actually doing what is
called phase cycling, which means in order to cancel
out artifacts we first do --
we do a set of four experiments or eight experiments.
You first do a positive X pulse for example,
then you do a negative X pulse, then a positive Y pulse,
then a negative Y pulse.
And then we average them all together
and that reduces artifacts.
So this big problem with all of this is we're dealing
with really miniscule signal and so the killer
in NMR spectroscopy, we've got very few nuclei
that are available, we've got very small magnetic vectors,
we get very small signals and the whole key is how
to get enough signal out of there
over all the noise that's coming
so that you don't need an entire NMR sample, NMR tube full
of pure sample and you can take just a few milligrams
or less compound in your NMR tube.
So if I want to give my very, very simple diagram
of an NMR spectrometer; so you have a solenoid.
A solenoid is just a coil in which you have electricity.
It's a super conducting coil.
Super conducting so that you have the electricity flow
forever and you don't have
to keep putting more electricity in it.
To do this you have a cold and liquid helium.
In order to minimize the evaporation
of the liquid helium you have a vacuum around it.
That's a Doer vessel and if you have a vacuum thermos
for your coffee, it has that.
But in order to further minimize the loss you have
that Doer contained in liquid nitrogen and a second Doer
around there because liquid helium's expensive
and you don't want to replace it that often.
So you have your solenoid and then you have your NMR sample,
and an NMR tube, and then you can think of it as your coil.
God this is a horrible drawing [laughter].
So your coil goes to an amplifier.
So this is just like a radio at this point.
Your coil goes to an amplifier
because you're getting a miniscule signal.
Who's ever opened an AM or FM radio and looked inside of it?
So the first thing you see is some sort of metal coil,
right, on an armature.
That coil, one of them is tuned for the AM frequencies
and there's a different one tuned for the FM frequencies,
or if you have a stereo and you have one coil that's your FM
antenna and one coil that's your AM antenna.
Obviously none of this is internet radio.
So anyway, you have two different coils.
That comes back to what I was saying before about having --
remember I mentioned broadband detection and the proton coil?
So in general different coil shapes work well
for different frequencies, and so one coil's on the inside
and so if we are doing proton NMR it's best
to have your proton coil on the inside.
If you're trying to get the best carbon NMR, in general it's best
to have a coil tuned to carbon's frequency.
Remember they differ by a factor of four from the inside.
Okay, so then for modern NMR, what you do is you go
from an amplifier we're going to go digital.
So after you get a signal, the signal is analog, after you go
to the signal in order
to process it we're going to digitize it.
That simply means convert it to bits and bytes.
So in other words instead of having a voltage here that's,
you know, 1.007823 millivolts you're basically going
to convert your voltage to binary
and say that's 110011 etcetera.
So we go to an ADC or analog to digital converter,
and then that goes to a computer and to a printer.
And for those of you who've run an NMR spectrometer you probably
know the command RG, receiver game.
Who's heard that one?
What you're doing there is basically saying, okay,
we want to have -- we want to fill up, have as big a number,
if we had an eight bit to analog converter,
in other words eight digits of zero
or one you want your biggest signal to fill up that thing,
to be as close as possible to 11111111,
whatever, eight one's is.
But if it's bigger you're going to saturate it
and then you're going to get all sorts of artifacts and clipping.
But if it's too small, if you're representing your maximum signal
by 00000111, then by the time you're
down t very small signals you just don't have the digital
capacity for it.
So that's what you're doing when you're adjusting receiver game.
Okay, so that's my pigeon diagram
of an NMR spectrometer and what's happening.
And as I've said, it's more complicated
because we have coils in all four directions
and you're going positive X to negative X and so forth.
All right, the big advance which of course is accepted
as ubiquitous in NMR is the Fourier Transform.
And I can guarantee all of you are going to be able
to intuit what this is with zero mathematics.
I've nothing against math, but there's incredible power
to being able to actually understand stuff rather
than calculate it.
Extras needed back here?
All right.
I'll give you the simplified version then I'll explain a
few details.
But let's start with the simple version.
All right, so this is a cosign wave corresponding
to a procession at one cycle per second.
In other words, every second we go around once.
If we take this function and Fourier Transform it we end
up having an amplitude axis and a time axis,
and what the Fourier Transform does is converts the time axis
to a frequency axis.
So we still have amplitude and now we've gone
from time to frequency.
And so if I write a little graph, 0123, I can represent --
God that's lousy and uneven.
I can represent the Fourier Transform
as a peak at one hertz.
One cycle per second.
And to a first order
of approximation that's all there is to a Fourier Transform.
It is taking that oscillation
and saying what's the frequency of the oscillation?
So if I take this second graph here
and we Fourier Transform that,
what's the Fourier Transform of that second graph?
A peak at two hertz.
Now if you've ever looked at your free induction decay --
we'll come to FID but you collect an NMR spectrum,
you see that wiggly thing.
That wiggly thing is the free induction decay.
That's what's going into your coil at each cycle.
If you've ever looked at it it's not a simple sign wave.
It is a simple sign wave if you only have one type of proton.
So if you do it on pure CDCL3 that has a little CHCL3,
and not a lot of water,
not a lot of TMS you'll just see a sign wave that decays
and I'll tell you about that in a second.
But normally what you see is something
that has more complications to it.
And so here what I did literally, this was just done
in Excel as an example,
is I took our one cycle per second graph
and I took our two cycle per second graph
and I added them together to get this red curve.
And so if we take the Fourier Transform of this red curve,
again we get frequency.
I'll just represent that as 0123 etcetera.
But now the Fourier Transform is going to be a peak at two hertz,
or peak at two, and a peak at one.
In other words, basically what's that saying is
that this is just the super position
of a one cycle per second current
and a two cycle per second current.
And obviously in an NMR spectrum
in which your ethanol may have four peaks for the CH2 group
and three peaks for the CH3 group it's going to be a heck
of a lot more complicated than that.
[ Inaudible Speaker ]
Parts per million is like cycles per second and more specifically
if you have 500 megahertz per session then you're going
to have 500 hertz as one ppm, a 1,000 hertz is two ppm
and 1,500 hertz is three ppm.
And if you have 300 megahertz per session,
if you have that 70,000 Gauss magnet I talked
about last time you're going to have 300 hertz is one ppm,
600 hertz is two ppm, 900 hertz is three ppm.
All right, now I'm not exactly playing honest with you.
Because if you have a frequency at a certain --
so remember, this is amplitude and time,
and the Fourier Transform transforms the time axis
to frequency axis.
If this goes on forever then you end up with a line
of infinite sharpness.
What actually gives rise to the sort
of peak you see is something at this frequency but that dies off
with an exponential decay.
And so when you Fourier Transform a cosign wave dying
off with an exponential decay you actually get something
that looks like an NMR peak and I'm not a great artist
but I will try my best
to represent the shape of this peak.
You have little wings coming out
and this is what you call a Laurencin Line Shape.
A Laurencin is just a mathematical function
that corresponds.
Its Y equals 1 over 1 over X squared plus 1,
and you'll later one see I have a simulation program
that actually incorporates this we'll get to play with.
However, the main point is peaks are not infinitely sharp
and this doesn't go on forever.
The reason this doesn't go
on forever is what's called relaxation.
So there are two types of relaxation.
There's longitudinal relaxation, sometimes called spin lattice.
Sometimes you'll also refer --
see it referred to as T1 relaxation.
And what this involves is re-equilibration
of the Alpha and Beta states.
My transfer of energy to thermal motions in the sample.
Now, easy way to think about this is
when we applied our pulse we drove our magnetization
down into the XY plane.
In other words what that means is we took the Alpha
and Beta states that were in the Boltzmann Distribution,
a natural distribution to be in a magnet and we forced them
to an unnatural distribution.
But eventually due to spin lattice relaxation your nuclei
flipped their spins back to these -- to the Z axis.
Or flipped their -- the population returns.
And now you see why you have an exponential decay.
It's a half-life process.
Any given nucleus has a finite probability
of having its spin flipped back to the natural population states
and so that's like radioactive decay.
That occurs with a half-life.
It's called the relaxation time
or more specifically the T1 relaxation time.
So that gives rise to our exponential decay.
That gives rise to our exponential fall off.
Now, there's a second type.
T1 relaxation is a little more important
in small molecule NMR spectroscopy
but there's a second type
of relaxation that's also important.
Transverse relaxation, sometimes called spin relaxation,
sometimes called T2 relaxation.
And what transverse relaxation does is involves interaction
of spins with other spins in the sample leading to an unbunching
of spins in the XY plane.
[ Inaudible Speaker ]
Unbunching of spins in the XY plane.
So interaction with spins of other nuclei leads to unbunching
of spins in the XY plane.
What does that mean?
Well remember, I said when we applied our pi pulse all
of our net magnetization was along the XY plane
and all the spins of the same type are processing together.
But if they unbunch, if due to getting tickled by other nuclei,
some process a little faster and some process a little slower.
I'm talking now for one type of nuclei
like the hydrogen and chloroform.
What happens to the net magnetization from that vector?
So look at where the vectors are starting
to cancel each other out.
We're still in the XY plane.
In other words out population of Alpha
and Beta states hasn't been perturbed.
We still have that none Boltzmann population
from the initial pulse where it's in this case
for pi pulse; 50% up and 50% down.
But now we're losing our focus in the XY plane.
As they unbunch our magnetization gets smaller
and smaller and our signal falls off.
And so through these relaxation processes, through T1 relaxation
and through T2 relaxation we have a falloff in our peak.
In our intensity.
And so we get a line that has a width to it.
All right, there are two concepts
that are closely related.
One concept we saw is the whole idea
of the Fourier Transform gives rise
to a peak with a width to it.
But the other thing is that line width is related
to the uncertainty principle.
So we could always go ahead and blame quantum mechanics.
And just of the uncertainty principle
as you've probably heard it is that you cannot know
with exact accuracy both the position
and the velocity of an object.
To put it in other terms the longer you can make a
measurement the more accurately you can know the velocity
or the longer you make a measurement the more accurately
you can know the angle of velocity.
That's the same ideas in infinitely sharp line.
If I have something processing and we can watch it forever,
we can know that this is processing
at 500.0003215 cycles per second.
However, if you only get to look at it
for a little bit you say well it was moving fast.
It wasn't moving at 100 cycles per second.
It wasn't moving at 1,000 cycles per second.
It was somewhere around 500 cycles per second.
And express mathematically what we get is
that Delta nu times Tao is equal to 1 over root 2 pi.
What this is, is the -- I'll call this the half line width
and you'll see why in a second.
And this is -- Tao is the half-life of the spin.
And so let's come back to our Laurencin line shape
and let's come to some hypothetical idea.
So remember we're talking frequency here.
All right, so our hypothetical ideal is this is the
hypothetical exact frequency.
But the point is if we can't measure the frequency exactly
because we're not measuring it for infinitely long,
because the frequency is -- the line is relaxing,
we can only tell well, it's somewhere around here.
So you get a peak that has some width to it.
And this is our Delta nu and from your point
of view what you often think of and what I think of when I look,
is this is what I like to call the line width.
In other words it's sort of at half height.
In other words it's 2 Delta nu.
Because here we're saying well it's within plus
or minus Delta nu of this center value but we can't tell exactly.
Here I look and I say, okay, this line is fat.
All right, so what does that mean?
If we have a Tao equals 2 seconds that leads
to Delta nu is equal to 1 over root 2.
Root 2 pi is equal to .11 hertz.
That's .22 hertz line width.
If I have Tao, and this is of course seconds,
if I have Tao equals one sec then Delta nu is equal to one
over -- well, I'll just skip the equation here --
it's equal to .22 hertz is equal --
and that's .44 hertz line width.
Now the relaxation of protons typically occurs on the order
of one or two seconds.
So there is a real theoretical limit
to how sharp your peaks can be.
Because that theoretical limit is going
to be determined by the relaxation.
Carbons are funny because they relax more slowly
and there the reason that you end up not having big peaks
for quads are always going -- your quandrinary carbons,
your carbon yields and carbon [inaudible] active are always
very short is because between pulses you don't go ahead
and have full return of magnetization.
>> Oh, how [inaudible] spin?
[Inaudible].
>> No, you don't have control of it.
I mean you have a little bit of control.
If you remove paramagnetic --
if you add paramagnetic impurities then --
which we actually do in experiments
like the inadequate experiment then you can increase --
you can decrease the half-life.
if you remove paramagnetic impurities for example,
if you by freeze pump [inaudible] degassing remove
dissolved oxygen which is paramagnetic from your sample,
you can decrease T1 relaxation
because remember it's the interaction of nuclei
that flips the spin and so you have oxygen as paramagnetic.
All right, so the point is exponential decay leads
to line broadening and if you have a little bit
of exponential decay like so, then you get a sharp line.
If you have a lot of exponential decay you get a broader line.
So these are both of Fourier Transform.
Yeah?
>> Do we reduce the fact of the other species
by aligning those nuclei's up?
>> You can do various tricks.
So, in the case of molecules
that have very restricted motions like solids,
T2 relaxation becomes critical and in the case of solids
to get good spectra you have to spin the sample at what's called
"the magic angle" to reduce relaxation, T2 relaxation.
So there are little tricks.
But for the most part relaxation isn't a problem.
Proton NMR, it's a good thing because it allows you
to repeat your experiments, and that brings us to the next
and I think perhaps the last thing that I will talk about.
And as I said, carbon NMR can be a pain because it --
the relaxation is so slow that it makes your peak small.
All right, so the last thing I'll talk
about is signal averaging.
Now, I've already hinted that you have this notion
of phase cycling and that your really have to do experiments
in sets of four or eight with the exception
of certain pulse field gradient experiments that I'll talk
about later that reduces phase cycling.
This becomes important in 2D NMR.
But you go to all the trouble to synthesize a compound,
make up an NMR sample, you collect the spectrum;
it's no big deal to collect date
for a minute instead of for five seconds.
So that's not a big deal.
But the big problem that we have, as I said,
NMR is a very insensitive technique.
You have a very low signal.
You're not fighting the low signal.
What you're fighting is the noise
and we talked about the cyroprobe.
The cryoprobe the other day, doesn't increase the amount
of signal but it does decrease the amount
of noise, electronic noise.
Think of noise as static.
You tune to an FM radio station that's far away,
you hear a lot of static.
Doesn't occur on internet radio because it's all digital.
But you tune to a station that's far away there's a lot
of static.
So what can you do?
The static is random, so if you go ahead
and collect repeated signal you can average it out.
The signal to noise varies as the square root
of the number it scans.
In other words if I collect more data I get more signal to noise,
but the noise is going up as well.
It's just going up randomly.
So in other words, if I go from 16 scans
which is a very reasonable number,
to 64 scans I don't quadruple my signal to noise,
I double the signal to noise.
And so if you're collecting a C13 NMR spectrum
and you've got a lot of noise you say, oh well if I go ahead
and I want to make my spectrum twice as good I've got
to collect data four times as long.
If it's midnight and you've been collecting since 10:00 pm,
know that your spectrum's only going
to get a little bit better, only 1.4 times better
if you wait till 2:00 in the morning.
If you wait till 6:00
in the morning it's going to get twice as good.
If I double the concentration I'll also double the signal
to noise.
So I'll say 2X.
So again, I'm sitting there at midnight and thinking, my God,
I don't want to be sitting here till 6:00 am.
I run up to the laboratory, I dump more sample in my NMR tube
and now by 2:00 in the morning I have twice as good A spectrum.
So that's the general gist.
There are a number of other aspects.
I think I gave us a reading in Claridge that I'd
like you to look over.
There are a number of other aspects
in Fourier NMR Spectroscopy including digital resolution.
But suffice it to say, right now we collect data
for a few seconds, the data is becoming
as your signal is falling off like so.
We have more noise here relative to signal so we don't want
to collect data forever.
So in the end we strike a compromise.
We collect for a few seconds, we average the data,
we perform a few mathematical operations to smooth it
out because the Fourier Transform of a truncated signal,
if I just truncate my signal the Fourier Transform actually looks
more like this where we have a couple of wiggly things
around the peak so we apply weighting functions.
That's called the exponential multiplication.
We apply weighting functions so you don't just truncate,
you actually drive the signal down so
that you get a better line shape
and you can read a little bit more about that in Clairedge.
All right, we'll pick up next time talking
about chemical shift and at that point we'll start to talk
about differences between different types
of protons within a molecule. ------------------------------fa4c97f04076--