What I want to do today is talk about just various aspects

of these correlation experiments that we've been talking about

and using now, COSY experiments and HMQC and HMBC experiments.

We're not going to become like super experts

on these experiments, but we've got a lot

of concepts floating around.

We've got the concept of inverse detection,

we've got some concepts of digital resolution that I'd

like to bring to bear.

We have various delays.

We've already seen when we were talking

about the depth experiment how important delay parameters are

and I'd like us to get a little bit of a feeling of that.

Down in the spec lab you're using gradient-based experiments

and without getting to be super technical I'd like to talk

about the benefits of that and benefits

on phase cycling and experiment time.

So let's start, and I also want to talk about variants

of experiments because although I've said, you know,

we're going to take this core of experiments

and it's not many experiments, I want to talk about some

of the variants of these experiments so that you can see

as you encounter specific problems what other tools you

can unpack from your toolbox to address those problems.

So, let's start by talking about the COSY experiment.

I'll give you the general pulse sequence here and then talk

about some variations of the experiments.

So in general, your experiments start

with a delay that we'll call D1.

That's a relaxation delay.

Remember we were talking about return

of magnetization to the Z axis?

I said normally your relaxation time, T1 relaxation time,

capital T1 relaxation time, is on the order of a second or two.

So when you pulse, normally it takes a few seconds for most

of your magnetization to return on the Z axis

and that's your DIOF [phonetic] and your FID.

Now when you're doing a normal 1D experiment,

that's not a big issue because you're collecting data

for a few seconds to get the typical digital resolutions

that you get in the 1D experiment.

Your 2D experiments there's an inverse relationship

between the amount of time you're collecting data

in your digital resolution and that's kind

of your uncertainty principle.

That gives you your, you know,

how accurately you can know your peak positions.

In a 2D experiment, you don't generally need super high

digital resolution.

So your acquisition times are typically shorter

like .17 seconds for say a typical COSY experiment.

So you don't want to be banging away every .17 seconds

because none of your magnetization will return

to the Z axis.

So most experiments, even your 1D experiments,

have a little relaxation delay.

So that's generally 1 to 2 seconds.

That's basically allowing your magnetization and that's,

of course, not allowing all your magnetization

to return to the Z axis.

It's allowing basically half of it or, you know,

or to allow 1 eth life so to allow 60%

of your magnetization to return.

All right so your pulse sequence is going to run

through relaxation delay then pulse then you wait

and you wait T1, that's your time, you increment this time

and you increment this time up to 1 over your sweep

with in the F1 dimension.

So we'll call that SW1 and remember we talked

about the 2 dimensional Fourier transform

where you're Fourier transforming with respect

to both the non-real dimension to the incremented dimension,

F1 dimension, and the F2 dimension, so you're going

to get periodicity from this weight just

as we see periodicity in the FID.

So then like most 2D experiments the general gist is pulse wait,

pulse observe sometimes with multiple pulses

but remember that's my general sort of simplified thing,

observe, and that's your T2, and then when you Fourier transform

and you get your 2D spectrum,

this is the F2 axis, this is the F1 axis.

So remember, the real axis,

the real 1 for each FID is the F2 axis

and then this one is coming from your incremented time here.

So you typically increment in usually it's a power of 2

so it's usually like in 256 or 512 or 1024 increments.

So, in other words, when you're collecting a COSY experiment

at minimum you're doing 256 or 512 or 1024 repeats

of this whole process.

Now, the more increments, the more digital resolution

[ Writing on board ]

in F1. So if you have 256 increments

and let's say F1 is 6,000 hertz.

In other words, let's say it's 12 PPM

on a 500 megahertz spectrometer that's 6,000 hertz,

then your digital resolution in F1 is going

to be 6,000 divided by 256.

In other words, your digital resolution is going

to be about 20 hertz.

That's pretty coarse because you think of say a typical multiplet

like a triplet and let's say your coupling constant is 7

hertz, so your multiple is 14 hertz wide.

So basically so that's sort of the bare minimum

on digital resolution because your digital resolution is going

to be on the order of like 20 hertz there.

Now there are various tricks with 0 filling.

So even if you don't, so if you collected 1024 increments,

you'd say, okay, my digital resolution would be 6 hertz.

It would be 6,000 divided by 1024, 6 hertz.

So that's sort of more like a typical peak size.

So some of the tricks that you can use are 0 filling

which adds data points artificially

but doesn't actually add new data, which can tighten

up your digital resolution.

Typically that's being done downstairs so typically you're

at least 0 filling to 1024 to sort

of artificially get your digital resolution

to about 6 hertz in this dimension.

All right I want to talk about, we'll talk about the time

for this experiment in just a second.

I just want to talk about some variations

of the COSY experiment

and so there's a variation called a long range COSY

and long range doesn't mean that you're picking

up long-range couplings or necessarily that you're going

and picking up small, you're picking up through 4 bonds.

Remember I said long-range coupling is typically more

than 3 bonds.

What a long-range COSY means is it picks up the small js better.

Why can't I write today?

I'm a mess here.

[ Writing on board ]

We've already seen this problem in COSY.

COSY is great if you have tall peaks,

it'll pick up any coupling, you know,

heck if you've got methyl singlets

that have invisibly small coupling,

you may get a cross peak over them from a tall methyl singlet,

but if you have a multiplet like this and your js are small

and you're coupling with another multiplet and your js are just

like less than 3 hertz, often it's hard to pick

up a cross peak and you saw that in the COSY

of the hydroxyl prolene, the one that we were talking

about in discussion where you saw that, for example,

your geminal protons you would only get a cross peak off of 1

of them because the other had a small coupling

and you could see the small coupling,

you could see a little splitting, remember this?

You saw a little splitting and yet only one

of those 2 diastereotopic methyls was giving you

a coupling.

So, this is like multiplets with, I hate to put a number

on it, but let's say j is less than or equal to 3.

It's sometimes hard to pick up the cross peak.

So a long-range COSY adds an extra delay, it's a fixed delay

that gives these js better and so the sequence is just

as we saw before it's D1 pulse, D1 is as above, pulse,

T1 so these are just as before,

but now you add one more fixed delay we'll call it D2

and then you pulse and you observe.

What the fixed delay does is it makes the experiment pick

up these small js better.

Now there's a price, you say why don't you use it all the time?

There's a small price that you pay.

Your fixed delay is typically let's say 100

to 400 milliseconds.

Longer is going to be better for picking

up small js but there's caveat.

What's happening during that 100 to 400 milliseconds?

>> Relaxation.

>> Relaxation.

So, you're losing signal intensity

because your magnetization is returning to Z axis

so there's a point of diminishing returns

but this would be an experiment that you would do

if you're saying I'm trying to pick up a coupling,

I'm not seeing it in my COSY, I think it's there, I'm confused

about my connectivity because of this and usually the places

that you're going to see it are places

where you have say a methine and you have bad geometry

to say another methine proton

because if you have a methyl group, a CH3CH,

you'll always have a good coupling.

You'll always have a good coupling with CH3CH

because the methyl is always going to have 1 or 2 protons

that have a decent geometry to give a decent j and a CH2,

these are all going to be okay typically although I guess we

actually saw one in the constrained 5-membered ring

where you didn't get 1 of your cross peaks

and you might have wondered, but when you start to have

like a CH next to a CH2 or next to a CH, you might want

to think about using it.

So, okay, I'll just write out what I said, but big delays lead

to loss of sensitivity.

More signal to noise problems.

There are tons and tons of flavors of COSY and just

like people develop different synthetic methods, you know,

yet another protecting group.

BJ Cory [phonetic] just has a paper

on a new variant that's very similar to TDBMS [phonetic],

but is a better protecting group and it's similar

to TIPS [phonetic] and so you say, okay, here's another one

in the toolbox and when you're starting out it's

like why do I need another tool in the toolbox

when I barely know how to use the tools I have?

So you can kind of file these away in the sense

that you're not going to be necessarily become an expert

in all of the alphabet soup.

[ Writing on board ]

There's a phase sensitive COSY experiment and what's good

about a phase sensitive COSY experiment it's harder to phase

but the cross peaks show splitting.

[ Writing on board ]

So from that experiment you can extract your js.

[ Writing on board ]

So you can imagine if you had some hideously complicated NMR

experiment and you absolutely wanted to measure your j values.

Let's say we've used j values for determining stereochemistry

so your stereochemistry was dependent on it

and you couldn't get your js by another way,

this might be a nice way to get your j values out of it.

Now there's another experiment that's very popular.

It has never been part

of my personal repertoire although now we're starting

to think about using it,

it's called the double quantum filtered COSY, DQF COSY,

it's a very popular experiment.

I just personally don't have a lot

of good things to say about it.

What it tends to do is reduce digital artifacts associated

with singlets.

[ Writing on board ]

So, for example, if you have a big methyl peak

or a big tert butyl peak in a COSY,

sometimes you get this stripe of T1 noise, this stripe it's

like a cruciform pattern off of that peak

and this can reduce some of that.

It can also reduce crowding around the diagonal.

Let's say helps show cross peaks close to the diagonal.