University of Ottawa NMR Facility Web Site

Please feel free to make suggestions for future posts by emailing Glenn Facey.



Wednesday, October 31, 2007

T1 Measurements and Estimation

T1 relaxation time measurements are usually done with a simple 180 -tau -90, inversion recovery pulse sequence (see figure). Tau is varied from a small value to a large value and a nonlinear regression is carried out to fit the best T1 value.These measurements can be very time consuming. One can get a reasonable estimate of the T1 much more quickly. Follow these simple steps:

1. Call up the pulse sequence "t1ir1d" (Bruker) or "s2pul" (Varian).
2. Set p1 and p2 (Bruker) or PW and P1 (Varian) to the 90 degree and 180 degree pulses , respectively.
Set the recycle delay, d1 (Bruker and Varian) to something you believe is much longer than the T1.
3. Set tau to a very small value (3 microseconds for example). Tau is d7 on a Bruker spectrometer or d2 on a Varian spectrometer.
4. Collect a spectrum and phase it such that all peaks are negative (one scan is often enough for protons). Store the phase correction.
5. Repeat step 3. increasing d7 (Bruker) or d2 (Varian) until the peak of interest is nulled. If the peak is negative, tau is too short. If it is positive, tau is too long.
6. The T1 of the peak of interest is the tau value for the null divided by the natural log of 2.

59 comments:

kjmarjun said...

Its a good comprehensive explanation for T1 measurement.

Luiz F. Pinto said...

Hello!
Very nice blog. I'm following you.
I'd like to ask you a simple question: Why do most people prefer to measure T1 instead of T2?
Thank you

Luiz

Glenn Facey said...

Hi Luiz,

Thank you for your comment question.

T1 and T2 are DIFFERENT parameters. Depending on what problem is being addressed, it may be necessary to meaure T1 rather than T2. Although they are often equal in solution (under extreme narrowing conditions), it is always true that T1 is greater than or equal to T2.

T1's go through a minimum when the product of the resonance frequency and the correlation time is approximately equal to 1. This minimum can be observed by measuring T1's as a function of temperature. The same minimum does not exist for T2.

Glenn

Anonymous said...

Glen,

Thanks for posting this. You saved me from doing the measurement the "long way". For my experiments, an approximate value is all I need.

Cheers

Anonymous said...

Thank you for this excellent post, it saved me a lot of time.
I use is to check T1 for 31P compounds on a Bruker instrument using Topspin v.2.1. Is there a way of modifying the experiment to include 1H decoupling as well?

Glenn Facey said...

Anonymous,

On a Bruker spectrometer, you must modify the pulse program to include decoupling.

Glenn

Anonymous said...

Hi,

would finding the null condition be valid in the situation where you have long very 90 and 180 pulses (due to poor Q-factor at high temperature). Fitting the data is unsuccessful since significant magnetization is lost at the beginning since T2 effects begin.

I find that I can fit the data if I add a constant offset, but I get a different answer than when I double check by using the null condition as you had suggested in the post.

Any thoughts about how to get around this issue?

thank you.

M.S.

Glenn Facey said...

Anonymous,

From your comment, it sounds as if the T1 is of the same order of magnitude as the pulses. In this case it would be very difficult to measure an accurate T1. I think that the single measurement of the null would be less precise than a proper fit to an inversion recovery curve.

Glenn

monk said...

Hello,
Very informative and practical blog: thanks for been around...

Let me please ask you the question about T1:
I learned that T1 measurements are concentration and temperature dependent.. While I can explain why it's T dependent, but what about concentration? Tnanks in advance

Glenn Facey said...

Monk,

Thanks for the comment and question. There are several mechanisms for T1 relaxation however they all depend on the product of the resonance frequency a correlation time for molecular motions. Relaxation is most efficient (i,e, T1 minimum) when this product is close to 1. As the concentration changes perhaps the viscosity changes as well altering the motional correlation time. Another explanation could be that there are more inter-molecular dipolar interactions at higher concentrations. In any event, the concentration dependence on T1 is much less than the temperature dependence.

Glenn

kjmarjun said...

I agree with this answer. More over, at extremely low temperatures, one can expect multiple minima and/or nonexponential behavior, when the correlation times pertaining to Rotational quantum tunneling effects in certain systems.

monk said...

Thank you Glenn! That's explain a lot.

Mike said...

T1s vary as a function of concentration in a few different ways. For instance, paramagnetic ions will shorten T1 in aqueous solution such that R1=c*r1, where c is the concentration and r1 is the longitudinal relaxivity. In the case of a solute, say water, in a solvent, say D2O, T1 will vary in a less convenient fashion. As Glenn pointed out, the viscosity changes, which of course affects correlation times. I also agree with Gleen that, depending on the solvent, the strength and types of couplings which lead to relaxation change, as interactions between solute spins will increase with concentration.

Anonymous said...

Thanks for the post! Great info.

My question is rather naive.

What is the advantage of using inversion recovery versus saturation (for spin 1/2) recovery? I guess I am not sure which is more appropriate to use under certain experimental conditions.

Thank you,
Jonas

Glenn Facey said...

Jonas,

Thank you for your question. Inversion recovery is used most often because it is the simplest. One can use it to measure the T1's for multiple peaks in the spectrum and the magnetization is followed from all the way from -Z to +z for inversion recovery rather than just 0 to +z in the case of saturation recovery.

Glenn

Anonymous said...

Hi,

Thanks for the response!

Sometimes I notice that I have a situation where I cannot fully invert the signal (even though 90,180 pulse durations are okay). Is this because T1 is relaxing too quickly and starts to relax within the deadtimes? Not sure if you even run into this situation.

Thanks for the helpful blog!
Jonas

Glenn Facey said...

Jonas,

If your 90 and 180 degree pulses are OK and you do not see an inverted signal in the inversion recovery spectra, then your choice of delays is too long with respect to the T1.

Glenn

David Bennett said...

Hi,
I'm trying to do T2-correction of STEAM (and PRESS) spectra by acquiring localized spectra at different TE values. I'm not quite sure of the correct way to do phase correction on the Bruker system. Here's what I do now: for each TE step I am using "ft", then "apk" on the first scan, then "ft" and "pk" on the remaining scans in that iteration of TE (4 iterations at each TE). For the next TE I repeat that process (for a total of 5 TEs). I assume this is ok but would be grateful to hear back any comments to the contrary. Thank you

Anonymous said...

Hi,
Thanks for posting this. My comment is with regards to T1 undergoing a minimum. Does the correlation time depend on frequency? For example, if I know T1 or T2 at 10 MHz and the correlation time, but want to know T1 or T2 at 300 MHz can I use the same correlation time? I am assuming the temperature is the same.

Thanks,
Warren

Glenn Facey said...

Dear Warren,

Thanks for the question. The correlation time does not depend on frequency. The T1 minimum occurs when the product of the Larmor frequency (omega) and the correlation time (tau) is approximately = 1. The T1 minimum for a resonance changes for different fields because the Larmor frequency changes.

Glenn

Anonymous said...

Hi,

Thanks for the response! So basically, if I know T1 or T2 at a particular frequency or field strength, then I can convert T1 or T2 to its corresponding value at any other field strength? Is that reasonable?

Warren

Glenn Facey said...

Hi Warren,

Yes, that is reasonable providing the mechanism(s) of relaxation are well known and the spectral density functions are precicely known.

Glenn

Anonymous said...

Dear Glenn
I run T1ir1D and T1ir experiments on the same samples to check the compliance of results and the outcome was that the T1s form 2D is about 1.5/2 fold longer than T1 from T1ir1D.
The longer T1s are equal to those reported in the certificate of NMR analysis of the samples.

Do you have an explanation to this outcome? I found none.

Greetings

Anna

Glenn Facey said...

Anna,
For your t1ir1d measurements, are you dividing the tau(null) value by ln(2)? This would account for a factor of 0.693.

Glenn

Anonymous said...

Hi Glenn,
yes I calculated T1 as Tnull*1.443.

Anna

Glenn Facey said...

Anna,
Could it be that the 90 and 180 deg pulses are not calibrated well? If the pulses are calibrated well then one of the parameters in the three parameter fit used for the 2D t1ir measurement should be 2.00. If this parameter is much different than 2.00, you might expect the T1's measured from the t1ir1d experiment to be off.

How broad are the resonances? Are the T1's very short?

Glenn

Chunqing Zhao said...

Dear Glenn:

when the t1 was calculated from Agilent software, triplet signal has three T1 values, corresponding to each peak. How to explain their differences? How to report T1 value in publication?

Thanks,

Nelson

Glenn Facey said...

Chunqing,

You should look into exactly how the Agilent software calculates the T1. Clearly, if your triplet arises from a single proton signal all members of the triplet should have the same T1. I suggest that you calculate the T1 based on the integral of the entire triplet.

Glenn

Chunqing Zhao said...

Thanks, Glenn

Anonymous said...

Hi Glenn thank you for your suggestions.
I calibrated the 90 and 180 pulses. Signals are not broad resonances - I used a 0.1% ethylbenzene in CDCl3 for further experiments-.
I used the Bruker uxnmrt1 fitting function.

Do you think possible that D1 and AQ times in the T1ir1D pulse program account for signal relaxation and affects results?
I noticed that running T1ir1D with 1 scan only and no ds results in T1much closer to that resulting from t1ir.
Have you ever double checked T1 from 1D and 2D pulse programs on the same sample?

Thank you

Anna

Glenn Facey said...

Anna,

When collecting more than one scan, the recycle delay (d1) you use is any T1 measurement is extremely important. It must be set to at least 5 times the longest T1 you intend to measure in your sample. If you are using the degassed 0.1% standard ethylbenzene sample, the T1's are likely to be quite long.

I have never compared the results from T1ir and T1ir1D but I would not expect there to be a discrepancy.

Glenn

V.I.L.F. said...

Hi,

I was wondering what to do when different peaks have different Tau. Do you simply choose the longest one?

Great blog by the way.

Regards,

Glenn Facey said...

V.I.L.F.

Every resonance is likely to have a different T1 therefore the tau values for the null will be different for every peak. In order to estimate the T1 for a particular resonance, you must find the tau value for the null of that particular peak. If you are doing this estimate to determine what recycle delay to use for a standard measurement then it should be based on the resonance with the longest T1.

Glenn

Julian Haigh said...

Hi Glenn, I have been the recipient of your NMR wisdom previously on another blog post but I need more.

I understand the T1 exponential equation needs the factor of "2" to account for the fact that the inversion recovery pulse sequence puts M onto the -z axis, but, how then is T1 still related to the magic 63% value? If I set Tau to be equal to T1 (i.e. 1/1 = 1) and Mo as 100, I do the algebra (perhaps incorrectly) and do not get 63%, but I do when i use the form of the equation without the "2" factor...any thoughts? Thanks, Julian.

Glenn Facey said...

Hi Julian.
Thank you for the question. When a 90 degree pulse is applied 63% of the magnetization is on the Z axis after 1 T1. If the magnetization starts on the -z axis (i.e. after an inversion pulse). It must relax to the +z axis. After T1 one expects the magnetization to recover by 63%. That is -1 + 2(.63) = 0.26 where -1 is the magnetization of the initial state immediately after a 18) degree pulse. The magnetization must relax from -z to +Z (i.e. 1- (-1) = 2. The magnetization after T1 is therefore (63% of 2) plus the initial state of -1.

Glenn

Anonymous said...

Dear Glenn,

Could you please help me? I did t1ir/cpmg measurements of H2O:D2O 5:1 and got some repeatable very strange results (t1<t2) and t2 about 400ms...... I checked a lot all my parameters, my lists and everything. I usually get sensible results on other experiments, including D2O:H2O 99:1, using the same water and same D2O bottle. The viscosity is changing so I am expecting different results by changing proportions, but not t1<t2. I do not understand.

Thank you,
Best regards,
Charlotte

Glenn Facey said...

Hi Charlotte,
T2 obviously cannot be longer than T1. If you have used the same methods successfully in the past to measure T1 and T2, then the problem you are having may be the result of radiation damping because of the very strong water signal. See this link.

http://u-of-o-nmr-facility.blogspot.ca/2007/10/width-of-your-water-line-radiation.html

Do you find that the 180 degree pulses do not provide null signals?

Glenn

Anonymous said...

Dear Glenn,

Thank you so much for your answer, it helped a lot. Yes this is what is happening apparently, a friend of mine told me that there are some sequences that correct this effect, particularly when working at high fields.

Thank you very much, I was not aware of this phenomenon.

Best regards,
Charlotte

Ambrish Kumar said...

Dear Glenn,

I would be grateful if you could help me in NMR experiment setup. My project entails the study of water proton relaxation in the presence of different MRI contrast agents. I have been trying to setup a standard T1 and T2 relaxation experiment on 90% H2O+10%D2O. I am not much familiar with the NMR pulse program. Initially, I used the inverse recovery pulse on 99%D2O+1%H2O and the value for T1 is much higher close to 15s due to water exchange with deuterium and I can not use 90% H2O because of radiation dumping. So if I use this pulse program on 99%D2O+1%H2O for T1 calculation it will take lots of time to acquire the data. Therefore, I am looking for the pulse program where I can get the normal value of water proton T1, 3-4s on 90%H2O.Then I came across some of the papers where they have used saturation recovery (selective presaturation by continuous wave, CW) pulse to overcome radiation dumping and the value for T1 is close to 3-4s.

I just want you to have a look at my pulse program below and suggest me if any changes are required.

T1_Saturation recovery:

;zgpr
;avance-version (06/11/09)
;1D sequence with f1 presaturation
;
;$CLASS=HighRes
;$DIM=1D
;$TYPE=
;$SUBTYPE=
;$COMMENT=


#include
#include
#include


"d12=20u"
"acqt0=-p1*2/3.1416-4u"


1 ze
30m st0
2 3m
3 3m
4 30m

"TAU=vd-4u-d12-p16-d16"
d12 pl9:f1
d1 cw:f1 ph29
4u do:f1 UNBLKGRAD
d12 pl1:f1
p16:gp1
d16
TAU
p1 ph1
4u BLKGRAD
goscnp ph31
3m st ivd
lo to 3 times nbl
3m
lo to 4 times ns

30m mc #0 to 4 F1QF()

; go=2 ph31
; 30m mc #0 to 2 F0(zd)
exit


ph1=0 2 1 3
ph29=0
ph31=0 2 1 3


;pl1 : f1 channel - power level for pulse (default)
;pl9 : f1 channel - power level for presaturation
;p1 : f1 channel - 90 degree high power pulse
;d1 : relaxation delay; 1-5 * T1
;d12: delay for power switching [20 usec]
;NS: 1 * n, total number of scans: NS * TD0



;$Id: zgpr,v 1.7.10.2 2006/11/10 11:04:15 ber Exp $

Glenn Facey said...

Ambrish,

I was unable to compile your pulse program with my version of TOPSPIN. It reported syntax errors. I have used the following program whch uses a train of pulses for saturation. It is a simple modification from the "satrect1" which should be in your pulse program library. The only modification is the removal of decoupling. Find the modified program below.

Glenn



;satrect1
;
;TS 3 / 08.09.2011
;
;saturation recovery T1 experiment
;pulse program is for 2D acquisition
;establish suitable vd list
;uses mc syntax, so that it can be used in 1D mode for set-up of
;saturation pulse train
;
;
;Avance III version
;parameters:
;pl1 : power level
;p1 : 90 degree pulse
;d1 : recycle delay
;d20 : delay in saturation pulse train
;l20 : number of pulses in saturation pulse train, 0 if undesired
;vdlist : list containing tau delays
;FnMODE: QF
;
;
;$CLASS=Solids
;$DIM=pseudo 2D
;$TYPE=direct excitation
;$SUBTYPE=T1/T2
;$COMMENT=saturation recovery T1 experiment


"acqt0=-(1u*+p1/2)"

1 ze
2 10m
d1
3 d20
(p1 pl1 ph4):f1
lo to 3 times l20
vd ;recovery delay
(p1 pl1 ph1):f1
1u
go=2 ph31
10m mc #0 to 2 F1QF(ivd)
HaltAcqu, 1m
exit

ph1= 0 0 2 2 1 1 3 3
ph4= 0
ph31= 0 0 2 2 1 1 3 3



;$Id: satrect1nodec,v 1.4.2.1 2014/02/11 09:04:53 ber Exp $

ambrish said...

Thanks Glenn for the reply.I really appreciate your help.
As I am not much familiar with NMR pulse program, I wanted to know what parameters other than p1 and O1 I need to optimize before running your modified pulse program on 90%H2O+10%D2O? What optimal value I should use for d2O delay?

Actually I have also tried other pulse program given below but it gives 0.1s for T1 which is much less than expected value 3s. I have tried to optimize the p1, d2, o1, pl9 but some how water peak is saturation very fast (within 1s) and T1 value is close to 0.1-0.2s.

;satrecpr.ret
;avance-version (00/02/07)
;T1 measurement using saturation recovery


#include
"d11=30m"
"d12=20u"

1 ze
2 d1
d12 pl9:f1
d2 cw:f1 ph29
4u do:f1
d12 pl1:f1
vd
p1 ph1
go=2 ph31
d11 wr #0 if #0 ivd
lo to 1 times td1
exit

ph1=0 2 2 0 1 3 3 1
ph3=0
ph29=0
ph31=0 2 2 0 1 3 3 1



;pl1 : f1 channel - power level for pulse (default)
;p1 : f1 channel - high power pulse
;d1 : relaxation delay; 1-5 * T1

;$Id: zg,v 1.6 2000/05/08 11:41:13 eng Exp $


I would be grateful if you could give me some suggestions.

Thanks
Ambrish Kumar

Glenn Facey said...

Ambrish,

Make sure the pulse is calibrated on an appropriate sample that does not suffer from radiation damping. I think that any of the saturation recovery methods should work because there is no strong signal at the beginning of the variable delay. There is however strong signal during detection so you will have to use the integrals in the analysis rather than the intensities. See these references.
Chem Phys Lett 222, 417 (1994)
Prog Nuc Mag Res Spec 68, 41 (2013)

Glenn

ambrish said...

Dear Glenn,

For calculating T1 relaxation time, I am using a standard inversion recovery with a gradient during the recovery delay (gpz1=5%) and this should remove the influence of radiation damping (http://www.ncbi.nlm.nih.gov/pubmed/16264247). My lab has 400MHz NMR equipped with z- gradient BBO probe. I ran this pulse program with d1=15s and gzp1=5%. I am encountering one problem that while running modified IR pulse program, the lock gets shifted at the end of the experiment. In the beginning, the lock makes beeping sound one or two times but it reaches to bottom at the end. However, data after processing give reliable T1 value close to 2-3s. I want to know that change in the lock would affect the accuracy of relaxation time. Please tell me what other parameters except gpz1 should I adjust so that lock does not shift. I am also scared that would it affect the probe?

Thanks and I really appreciate your help

Following is the pulse program:

;t1ir
;avance-version
;T1 measurement using inversion recovery


#include
#include


"p2=p1*2"
"d11=30m"


1 ze
2 d1 UNBLKGRAD
p2 ph1
100u gron1
vd
100u groff
3u BLKGRAD
p1 ph2
go=2 ph31
d11 wr #0 if #0 ivd
lo to 1 times td1
exit


ph1=0 2
ph2=0 0 2 2 1 1 3 3
ph31=0 0 2 2 1 1 3 3


;pl1 : f1 channel - power level for pulse (default)
;p1 : f1 channel - 90 degree high power pulse
;p2 : f1 channel - 180 degree high power pulse
;d1 : relaxation delay; 1-5 * T1
;d11: delay for disk I/O [30 msec]
;vd : variable delay, taken from vd-list
;l4: l4 = number of experiments = number of delays in vd-list
;NS: 8 * n
;DS: 4
;td1: number of experiments

;define VDLIST

;this pulse program produces a ser-file (PARMOD = 2D)

Glenn Facey said...

Ambrish,

I am not sure why you are having lock problems. You may try incorporating the following 'include' statements:
#include
#include
#include

You may try using a fixed gradient pulse during the recovery delay rather than simply turning on the gradient for the entire recovery delay.

You can also just run the experiment unlocked.

Glenn

Glenn Facey said...

Ambrish,

I see that in my response to your comment that the include statements weren't reproduced properly due to the HTML meaning of the angled brackets. they are here:

#include Avance.incl
#include Grad.incl
#include Delay.incl

Glenn

ambrish said...

Hi Glenn,

Thanks for the reply.
You said in the previous post that I should use a fixed gradient pulse during the recovery delay rather than simply turn on the gradient for the entire recovery delay. For that what changes do I have to make in the pulse program and which delay values should I change?

Currently, I have been running that pulse program and lock gets failed at the end of the experiment, however, the T1 value seems reliable and you also mentioned that lock would not affect the relaxation time.I am scared that I might burn the probe because something is wrong and because of that lock gets failed.

I would love to have your suggestions.

Best,
Ambrish

Glenn Facey said...

Ambrish,

To use a gradient pulse, in you sequence replace:

100u gron1
vd
100u groff

with:

p16:gp2
d16
vd

The duration, strength and shape of the gradient pulse must be defined (a 1 msec sine pulse of 5% to 10% should be OK). d16 should be set to 100 usec. Keep in mind that each of the variable delays in the sequence would be (p16 + d16 = 1.1 msec) longer than the delays listed in the vd list.

Glenn

ambrish said...

Hi Glenn,

I have tried fixed gradient pulse during the recovery delay but it does not overcome the radiation damping. I tried gpZ2 (p16=1ms) 5% to 10% with d16 100us but it does not show any change in radiation damping and i am getting very short T1 value for water proton in 90% H2O+10% D2O. However, lock does not get failed and is constant throughout the experiment.

Thanks

Best,
Ambrish

Glenn Facey said...

Ambrish,

You should definitely see the lock dip during the gradient pulse. What is the shape assigned to gpz2?

Glenn

ambrish said...

Hi Glenn,

Thanks for quick response.
I could see lock dip during the gradient pulse in my experiment. I have used SINE.100 assigned to gpz2. I have tried changing gpZ2 from 5% to 10% but still radiation damping is more prominent and as a result I got very short T1 value.Do you think the fixed gradient pulse during the recovery delay is not sufficient to overcome the radiation damping? Could you suggest what other parameter I can change?

Thanks

Best
Ambrish

Glenn Facey said...

Ambrish,

You could try using a a higher power gradient. If this doesn't work, you could revert back to the sequence you were using and run it unlocked.

Glenn

Anonymous said...

Hi Glenn,

Thank you for very informative blog. I have a question regarding the T1 measurement of 13C with and without 1H decoupling at thermal equilibrium in a molecule with 1H nuclei J-coupled to 13C nucleus. Due to the effect of 1H decoupling on 13C-1H cross relaxation, Would there be a difference if I measure T1 time of 13C with and without 1H decoupling? If there is, in which case, T1 relaxation time will be more?
Thank you.

best wishes,
Krish

Glenn Facey said...

Hi Krish,

It is my understanding that the 13C-1H cross relaxation rate constant does not depend on whether or not proton decoupling is employed and therefore one would expect to observe the same 13C T1 in the presence or absence of proton decoupling. The NOE is of course dependent on proton decoupling however the cross relaxation rate constant is not.

Glenn

nmr_chemist said...

Greetings Glenn!

Do you have a recommended software package for identifying the null point?

C

Glenn Facey said...

C.
I don't use any software package to find the null. I run a few spectra with different recovery delays until I find it.
Glenn

Shannon Chow said...

Why the FFT results of some NMR signal were negative? How to decide the sign? I think that FFT results should always be non-negative.

Glenn Facey said...

Shannon,
FFT spectra can have any phase; positive, negative or anywhere in between. See this link.

http://u-of-o-nmr-facility.blogspot.ca/2010/11/phase-of-nmr-spectrum.html

Glenn

govardhan said...

Hi Glenn,

I was wondering about that can we manually calculate T1 and T2 form the spectra. I did record the deuterium NMR spectra for the deuterium labelled ligand in the buffer solution and saw the sharp peaks (Peaks are overlapped). Broad peaks (overlapped) were observed when the deuterium NMR is recorded for deuterium labelled ligand with the protein in buffer solution. My prof asked me to calculate T1 and T2 from both spectra.Could you please guide me how to calculate T1 and T2. I would be greatly appreciated your help.

Thank you
Reddy

Glenn Facey said...

Reddy,
If I understand you correctly, you are asking if a T1 can be estimated from a single spectrum recorded with a one pulse measurement. If the following conditions are true:
1. The magnet is perfectly shimmed
2. All species are in the fast motion limit (extreme narrowing).
3. There are no exchange processes causing line broadening.
4. The data are processed with no apodization function.

then T1=T2 and T2=1/(pi*lw) where lw = the line width at half height.

Glenn