Its Time to Move On

After more than 28 years of serving the NMR Facility at the University of Ottawa, I have decided to retire at the end of 2020.  My position will be posted this fall.  What a fantastic job it is managing an NMR Facility!  I have learned so much about NMR over the years and have taken great pleasure in helping so many students.  I have particularly enjoyed writing this BLOG which was originally intended to answer the frequent questions asked of me by uOttawa students.  To my surprise and delight, it has been useful to so many more people than just those at the University of Ottawa.  Thank you for reading it.

It is now time for me to move on.  Much less of this:

 ....and much more of this:

Cheers.

NMR and Food Chemistry - Rhubarb

Many gardeners here in Canada are currently harvesting their rhubarb to make pies and other desserts.  Like cranberries, rhubarb is very sour due to the presence of organic acids.  The leaves of the rhubarb plant contain so much oxalic acid that they are toxic.  The stalks of the plant contain less oxalic acid and are not toxic (although I would argue that they taste awful).  Oxalic acid is used in many cleaning products, so if you are like me and hate the taste of rhubarb, you can still make use of it as a cleaning agent.  If you have a very dirty stainless steel pot, you can clean it by simply stewing rhubarb in it.  The figure below shows the ¹H and ¹³C NMR spectra of an aqueous extract of rhubarb stocks.  The major constituents are organic acids with smaller quantities of sugars.  Some of the ¹³C signals from oxalic, malic and citric acids are labelled.

NMR and the Liquid-Gas Interface

Most NMR spectra are recorded for liquid or solid phase samples.  Many chemists have not even considered measuring NMR spectra of gas phase samples.  Such spectra are indeed possible to record and the information available from such spectra has been studied and reviewed in detail.*  In our first high school science classes we learn that molecules in the gas phase diffuse much more quickly than those in the liquid phase and that there is an equilibrium between the liquid and gas phases.  These two elementary concepts can be demonstrated nicely with ¹H NMR spectroscopy.

A suitable sample was prepared by putting 1-2 µL of acetone in a standard 5 mm NMR tube.  A greased rubber plug was then forced into the tube such that it resided about 6 cm  above the bottom of the NMR tube.  This was done to limit the volume over which the vapour could diffuse to that of the active volume of the probe coil.  The tube was then sealed with a torch to prevent the loss of sample.  The sample contained a small amount of liquid in the bottom of the NMR tube and a mixture of acetone vapour and air above the liquid.  A sketch of the sample is shown in the figure below.

The 600 MHz ¹H data were collected in a cryoprobe at 298 K without a ²H lock.  The magnet was shimmed using the ¹H FID. The ¹H spectrum has two resonances, a broad one at ~2.2 ppm (Δν_1/2 = 30 Hz) and a narrower one (Δν_1/2 = 4 Hz) at ~3.8 ppm due to liquid and gaseous acetone, respectively.  The large 30 Hz line width for the liquid resonance is due to the magnetic susceptibility discontinuity boundary between the droplet of liquid with the glass and vapour interfaces.  There may also be broadening as the droplet resides near the edge of the homogeneous region of the magnetic field.  A DOSY spectrum, acquired with δ = 0.5 msec and Δ = 4.9 msec, illustrates the vastly different molecular diffusion rates between the liquid and gaseous phases of acetone.  An EXSY spectrum, acquired with a 2 second mixing time, clearly shows exchange peaks between the liquid and the gas phases, illustrating the liquid-gas equilibrium.

* C.J. Jameson. Chem. Rev. 91, 1375-95 (1991).

12C/13C Isotope Effects on 1H T1 Relaxation Times

What is the ¹H T₁ relaxation time of chloroform?  It seems like a simple enough question, but the answer is not so simple.  The relaxation rate for any proton is the sum of relaxation rates resulting from several different mechanisms (eg. homonuclear dipolar coupling, heteronuclear dipolar coupling, chemical shielding anisotropy, spin rotation etc...).  Each of these mechanisms of relaxation depends on dynamic effects and the extent to which those processes occur at the Larmor frequency.   Often, in proton-rich organic compounds, ¹H T₁ relaxation is dominated by the homonuclear dipolar coupling interaction.  For chloroform, with only a single proton, there can be no intra-molecular homonuclear ¹H dipolar interaction and the ¹H relaxation rate must depend on other mechanisms.  One of these mechanisms is the result of the heteronuclear dipolar coupling interaction.  For the ¹³C isotopologue of chloroform, one would expect a significant heteronuclear dipolar interaction between the directly bound ¹H and ¹³C.   This interaction is absent in the ¹²C isotopologue and one would therefore expect the T₁ relaxation time of ¹³CHCl₃ to be much shorter than that of ¹²CHCl₃.  This is illustrated in the figure below.

The ¹H T₁ relaxation times for both ¹²CHCl₃ and ¹³CHCl₃ were measured with the inversion recovery method for a degassed, dilute (1%) sample of chloroform in acetone-d6.  The inversion recovery delay was varied from from  1 sec. to 300 sec.  The recycle delay was 300 sec.  Relaxation is much more efficient for ¹³CHCl₃ compared to ¹²CHCl₃.  The T₁ for ¹³CHCl₃ is only 46% that of ¹²CHCl₃, indicating the significance of the heteronuclear ¹H - ¹³C dipolar coupling interaction as a relaxation mechanism.       

1H T1ρ Edited 13C CPMAS Spectra - Pharmaceutical Analysis

Active pharmaceutical ingredients (API's) are often mixed with other compounds (excipients) used to dilute, stabilize, sweeten, color, flavour, bind, coat (etc...) the medication.  Often the API is a crystalline compound while the excipients are noncrystalline or amorphous.  When analyzing pharmaceutical pills by solid-state ¹³C CPMAS NMR, one often wants to observe the active crystalline API and not the amorphous non-active excipients.  This can sometimes be accomplished by collecting spectra with differing recycle delays, as the excipients often have shorter ¹H T₁ relaxation times than the crystalline API.  An example of this is given here.  Crystalline API compounds often have much longer ¹H T_1ρ's than the amorphous excipient compounds.  One can modify the simple CPMAS pulse sequence to discriminate against the excipients with short ¹H T_1ρ's by introducing an additional ¹H spin locking pulse before the cross polarization during which the transverse ¹H magnetization of the excpient decays to zero while that of the API decays to a much smaller extent.  After the ¹H spin locking pulse, cross polarization is applied, transferring polarization from the remaining API ¹H magnetization to the API ¹³C.  This modification to the CPMAS experiment is shown in the figure below.

The figure below illustrates this technique applied to generic acetylsalicylic acid (ASA) tablets.

The spectrum in the bottom panel is a ¹³C CPMAS spectrum of a crushed generic ASA tablet with a 60 second recycle time and a 2 msec contact time.  One can see the nine ¹³C resonances of the ASA plus the broad excipient signals between 50 ppm and 110 ppm (highlighted in yellow).  The spectrum in the middle panel was acquired under identical conditions with an additional 10 watt, 12 msec ¹H spin locking pulse applied prior to cross polarization.  Clearly, the broader signals of the excipients (with short ¹H T_1ρ) are suppressed leaving only the resonances of the ASA.  The top panel is a weighted difference of the two spectra showing only the signals of the excipients.  Although a pharmaceutical example is used here, the technique is applicable generally to any mixture of solids with different ¹H T_1ρ's.

Where is my INEPT signal? - Proton Exchange Issues

INEPT and DEPT sequences are routinely used to enhance the NMR signals for low γ nuclides such as ¹⁵N or ¹³C.  The enhancement relies on polarization transfer between the protons J-coupled and the low γ nuclide.  The pulse sequences incorporate delays based on the reciprocal of the J-coupling constant between the protons and the low γ nuclide.  In the case of ¹⁵N INEPT, the enhancement for each scan can be as much as γ_H/γ_N (~ 10) compared to that from a conventional one-pulse sequence with inverse gated decoupling.  Furthermore, the recycle delay for the INEPT sequence depends on the the ¹H T₁ relaxation time rather than that of ¹⁵N.  ¹H T₁'s are typically an order of magnitude (or more) less than those of ¹⁵N so the recycle delays required for ¹⁵N INEPT spectra are at least ten (and possibly 100 times) shorter than those required for one-pulse data collection.  These two factors mean that the true time saving for a ¹⁵N INEPT measurement compared to a one-pulse ¹⁵N measurement can be on the order of 100 - 1000 times.  There are, however cases where ¹⁵N INEPT signals are attenuated or entirely nonexistent.  Attenuated ¹⁵N INEPT signals are observed when the protons (with short T₁) coupled to ¹⁵N exchange with those of water (longer T₁) on a time scale of seconds.*  The problem arises because of saturation transfer during the inverse gated decoupling used during the acquisition time. The partially saturated protons are unable to transfer as much polarization to the ¹⁵N as they would were they fully polarized. The problem can be reduced if a recycle delay much greater than the T₁ relaxation time of the water protons is employed.  If the protons bound to ¹⁵N undergo exchange with other labile protons at a rate fast with respect to the ¹H-¹⁵N J coupling interaction, polarization transfer from ¹H to ¹⁵N is not possible and a ¹⁵N INEPT signal cannot be observed.  This is demonstrated in the figure below.

Concentrated solutions of the methyl ester of anthranilic acid and anthranilic acid were prepared in DMSO-d₆.  The ¹⁵N NMR data were collected on a 600 MHz instrument with a cryoprobe.  The left-hand panel of the figure compares the ¹⁵N one-pulse spectrum with inverse gated decoupling (bottom) to the INEPT spectrum (top) for the methyl ester.  The spectra were collected with the same number of scans. For the methyl ester, the ¹⁵N bound protons do not exchange with any other labile protons.  The enhancement in the ¹⁵N INEPT spectrum is clear.  Similar spectra for anthranilic acid are shown on the right-hand side of the figure.  In anthranilic acid, the ¹⁵N bound -NH₂ protons undergo intramolecular exchange with the acid proton at a rate fast with respect to the one-bond ¹⁵N-¹H coupling constant (~90 Hz).  As a result, polarization transfer is not possible and no INEPT signal is observed.  The same is true for the meta- and para- isomers (data not shown). 

Thank you to Jin Hong for sharing her experience with collecting ¹⁵N INEPT data for anthranilic acid and Mojmir Suchy for kindly providing the samples.

* G.D. Henry and B.D. Sykes, J. Magn. Reson. B, 102, 193 (1993).

1H J-Resolved Spectroscopy to Evaluate 1H-1H and 1H-19F Coupling Constants

2D ¹H J-RESolved spectroscopy (JRES) is able to separate the ¹H chemical shift and J coupling interactions in the F2 and F1 domains of the 2D data, respectively. The F2 projection represents the pure-shift ¹H decoupled ¹H NMR spectrum while the individual F1 slices at each chemical shift reveal the ¹H - ¹H J coupling for each resonance.  When this technique is applied to a spin system with both homonuclear ¹H-¹H coupling and heteronuclear coupling, it has the ability to provide both the homonuclear and heteronuclear coupling constants.  This is demonstrated in the figure below for 2,3-difluoro pyridine which has both ¹H-¹H and ¹H-¹⁹F coupling.

The top trace in the figure is the ¹H NMR spectrum showing the complex resonances due to both the homonuclear and heteronuclear coupling.  The 2D JRES spectrum is highlighted in grey.  The ¹H-¹H coupling is shown in the F1 slices which were summed to produce the blue, red and green vertical traces in the figure for ¹H resonances A, C and B, respectively.  These traces are identical to the resonances in the separately collected ¹H spectrum with ¹⁹F decoupling shown in the bottom trace of the figure.  The F2 projection of the JRES spectrum is shown in the trace directly on top of the 2D spectrum, colour coded in yellow.  The F2 projection represents the ¹H decoupled ¹H spectrum showing only the ¹H- ¹⁹F coupling.  It can be compared to the separately collected PSYCHE pure-shift ¹H spectrum, colour coded in orange which is very nearly identical.  Clearly this very simple, often overlooked, technique can provide a great deal of both homonuclear and heteronuclear coupling information. 

13C-13C Connectivity via 1H-13C 1,1-ADEQUATE

One of the most valuable pieces of information one could obtain in elucidating the structure of a small organic molecule is carbon-carbon connectivity information.  This information can sometimes be indirectly deduced from HMBC and/or H2BC data with reasonable sensitivity.  The same information can be determined directly, albeit with dramatically less sensitivity, using the ¹³C INADEQUATE technique.  Another option for obtaining carbon-carbon connectivity information is the 1,1-ADEQUATE technique (Adequate sensitivity DoublE QUAnTum spEctroscopy).  This method is proton detected and relies on a 1-bond INEPT transfer between ¹H and ¹³C.  One-bond ¹³C-¹³C double quantum coherence between the carbon bound to the proton used for the initial INEPT transfer and adjacent carbons is allowed to evolve in much the same way as in the INADEQUATE technique.  Magnetization is transferred back to single quantum coherence for proton detection.  The 2D NMR data show correlations between the proton resonances and the double quantum frequencies between the carbon attached to the proton and those carbons bound to that carbon.  The carbon-carbon connectivity information is provided in the double quantum carbon frequencies.  One drawback to the 1,1-ADEQUATE technique is that connectivity cannot be established between two quaternary carbon atoms not attached to protonated carbons.  Connectivity information between a quaternary carbon bound to a protonated carbon can however be established.  The sensitivity advantage of the 1,1-ADEQUATE technique compared to the ¹³C INADEQUATE technique arises from ¹H rather than ¹³C detection and that the recycle delay depends on the proton T₁'s rather than the ¹³C T₁'s.  Here is an example of how one could use the 1,1-ADEQUATE technique with other methods to unambiguously assign the structure of a small organic molecule.  The edited HSQC spectrum of the unknown molecule with separately acquired ¹H and ¹³C NMR spectra as projections is shown in the figure below.

 The ¹³C spectrum provides all of the ¹³C frequencies, while the edited HSQC signals provide the ¹H-¹³C one-bond connectivity and multiplicities for each protonated carbon.  Note that the carbon frequencies could also be determined from a high resolution HMBC spectrum if insufficient material is available for a direct ¹³C measurement.  From the carbon frequencies, one can determine all of the double quantum frequencies as shown in the table below, taking into account the ¹³C offset frequency expressed in ppm, 'o1p'.

Those  highlighted in pink are those that are present in the 1,1-ADEQUATE spectrum which is shown below.

The spectrum was acquired on a concentrated sample at 600 MHz with a cryoprobe using the standard 'adeq11etgpsp' Bruker pulse program .  The total data collection time was less than 1 hour.  The carbon-carbon connectivity is labelled on the spectrum based on the double quantum frequencies using the numbering scheme from the ¹³C spectrum presented as the projection on the edited HSQC spectrum above.  From these connectivities, the structure of the compound can unambiguously be assigned to limonene.

11B Double Quantum - Single Quantum Correlation Spectroscopy

The 2D ¹³C INADEQUATE method provides double quantum - single quantum (DQ/SQ) correlations and enables one to determine the carbon - carbon skeleton of small organic molecules.  The method is quite insensitive for ¹³C since the natural abundance of ¹³C is only 1.1% and the chance of having two adjacent ¹³C nuclei is only 1 in 8264.  For spins other than ¹³C, for which the natural abundance is high, one expects the sensitivity of DQ/SQ correlation spectroscopy to be much higher.  ¹¹B has a natural abundance of 80.42% and the chance of having two adjacent ¹¹B nuclei in compounds with boron-boron bonds is 1 in 1.55.  ¹¹B NMR spectra are often sufficiently broad due to efficient quadrupolar relaxation such that homonuclear ¹¹B - ¹¹B J coupling is unresolved however the T₂'s are long enough to allow the collection of  2D ¹¹ B COSY and 2D DQ/SQ data.  The figure below shows the 2D ¹¹B DQ/SQ correlation spectrum of ortho-carborane.  The ¹¹B bonding connectivity can be determined easily from the spectrum.

2D 13C INADEQUATE

The 2D ¹³C INADEQUATE (Incredibe Natural Abundance DoublE QUAmtum Transfer Experiment) is undoubtedly one of the most definitive, yet under-used NMR techniques able to assign chemical structures of small organic molecules.  It gives a connectivity map for all carbon atoms in the molecule.  The reason that it is so under-used is that it relies on one bond ¹³C - ¹³C coupling therefore, adjacent carbon atoms must both be of the ¹³C isotope.  Since ¹³C is only 1.1% naturally abundant, the chance of having two adjacent ¹³C atoms is approximately 1 in 8300, reducing the sensitivity of the measurement drastically.  As a result, 2D INADEQUATE spectra can only be run on very concentrated samples.  The sensitivity afforded by high magnetic field strengths and cryogenically cooled probes has certainly made these measurements more accessible than they have been in the past and they may be within reach when sample quantity and solubility are not a problem.  The 2D ¹³C INADEQUATE spectrum of ~450 mg of limonene in benzene-d6 was acquired on a Bruker TCI H/C/N cryoprobe at 600 MHz and is shown in the figure below.

The spectrum was acquired with the gradient version of the INADEQUATE pulse sequence using a shaped refocusing pulse (Bruker pulse program inadgpqfsp).  It was acquired in 11.8 hours with 64 scans for each of 128 increments using a 5 second recycle delay. The proton decoupled ¹³C spectrum is in the horizontal F2 domain.  The spectrum is interpreted by locating the vertical cross peaks for each ¹³C resonance.  Each cross peak has a partner peak along the horizontal axis.  The partner peak lies on the same vertical axis as the carbon atom bonded to the initial carbon.  This is shown in the figure for C1 which is bonded to C2, C3 and C4.  The entire carbon skeleton of the molecule can be traced unambiguously in this manner to provide a complete assignment.  The same sample was run under the same conditions in 45 minutes with only 4 scans.  A comparison of the signal-to-noise ratio for both data sets is shown in the figure below.

It is clear that usable 2D INADEQUATE data can be acquired in less than an hour for extremely concentrated samples at high field with a cryoprobe.     

Fast 2D Data Collection - NOAH and NUS

One always strives to collect high quality 2D NMR data in a short period of time.  This is particularly important for samples of limited stability or perhaps for monitoring chemical reactions.  High magnetic fields and cryogenically cooled NMR probes have allowed for a higher signal-to-noise-ratio for a given quantity of sample, thereby reducing data collection time as a fewer number of scans are required.  Gradient enhanced 2D NMR data collection gained widespread use in the 1990s.  This represented a tremendous time saving as multi-step phase cycles required for coherence selection could be reduced or eliminated as they were replaced by pulsed field gradients.  Some pulse sequences which required 16 scans per increment to accommodate the necessary phase cycle could be run with a single scan for every increment with the use of pulsed field gradients, thus reducing the data collection time by a factor of 16.  Now, 2D data collection with coherence selection via pulsed field gradients is considered "conventional".  More recently, Non-Uniform Sampling (NUS) was introduced.  Data collection with this technique samples only a limited number of increments in the t1 domain.  The unsampled increments are calculated based on the sampled increments prior to Fourier transformation. The data collection time is reduced in accordance with the number of increments not sampled.  Recently, Kupce and Claridge^1,2 have developed a technique where multiple 2D methods are concatenated in a single super pulse sequence employing a single relaxation delay. They have called the technique NOAH (NMR by Ordered Acquisition using ¹H detection) The time saving of the NOAH technique compared to individually collected 2D spectra results from waiting a single relaxation delay for all experiments rather than a single relaxation delay for each separately acquired spectrum. The data for each spectrum is acquired in separate memory blocks which are separated after data collection allowing the data for each 2D method to be processed individually.  Very recently, both NUS and NOAH have been used together to further reduce data collection times³.  A comparison of the time saving is shown in the figure below for a sample of sucrose in DMSO-d6 collected on a Bruker AVANCE III HD 600 NMR spectrometer equipped with a cryoprobe.

All spectra were collected with 2 scans and a 1 second recycle time.  Individually, both NOAH and NUS offer a significant time saving but when used together they permit very fast, high quality data collection.  A COSY, edited HSQC and HMBC can be collected in a total time of only 4 minutes and 8 seconds.  Other ultra-fast techniques have been developed by others where an entire 2D spectrum is collected in less than one second.

1. Eriks Kupce and Tim D. W. Claridge.  Chem. Commun. 54, 7139 (2018).
2. Eriks Kupce and Tim D. W. Claridge. Angew. Chem. Int. Ed., 56, 11779 (2017).
3. Maksim Mayzel, Tim D. W. Claridge and Ēriks Kupce. Bruker User Library (2018).

Dying iPhones and Liquid Helium Fills

As a habit I do not expose my iPhone to the large stray magnetic fields of high-field or unshielded NMR magnets.  I do however feel safe carrying it near low-field shielded magnets with 5 Gauss stray fields within the croyostat of the magnet. That is - until lately.  Last year, after topping up the liquid helium on a 300 MHz shielded magnet in a fairly small room, I noticed that my iPhone 8 had become completely unresponsive.  The only stimulus it appeared to respond to was gravity.  As it was under warranty, I sent it back to Apple.  After a week or so, they sent it back to me with a note saying that there was nothing wrong with it.  I found this very strange and did not make a connection between the helium fill and the problem with the phone.  I had done many helium fills in the past while carrying an older iPhone 5.  Approximately 9 months later, my iPhone 8 suffered a similar problem again after filling the same shielded 300 MHz magnet with liquid helium.  This time, I took it to a local Apple Store while it was dead.  The technician examined it, ran it through a software protocol, confirmed it was dead and issued me a new phone as it was in its last weeks of its warranty.  Shortly after this, I read about problems others have had with iPhones and Apple watches around helium gas and finally made the connection between the problem I was having and my helium fills. Some Apple iPhones (apparently, iPhone 6 and higher) will completely die when exposed to helium gas. As if in the spirit of Easter, however, they will resurrect themselves after the helium has dissipated from the phone and the battery has been allowed to discharge.  The problem is that in newer iPhones, Apple has swapped out a quartz oscillator, used in older versions of the phone, with a microelectromechanical systems chip (MEMS) which is sensitive to the presence of helium gas. This sensitivity is indeed mentioned in the User Guide of the iPhone.

Exposing iPhone to environments having high concentrations of industrial chemicals, including near evaporating liquified gasses such as helium, may damage or impair iPhone functionality. Obey all signs and instructions. 

Android phones apparently do not use MEMS and therefore are not vulnerable to the problem. Several weeks ago, I absent-mindedly entered a room with my iPhone 8 while a magnet was being filled with liquid helium.  Again, the same thing happened.  This time, I allowed the phone to sit for a week after which I was able to charge it.  After charging, it worked well with no loss of data.

Warning: If you see that a magnet is being filled with liquid helium, Do not enter the room with an iPhone 6 or higher.

Comparison of Broadband Decoupling Schemes

Many NMR measurements such as HSQC or HMQC rely on broadband X nucleus decoupling (X = ¹³C, ¹⁵N, ³¹P .... etc.)..  Broadband decoupling schemes, using conventional rectangular pulses (e.g GARP) require fairly high power levels leading to undesired sample heating.  They are also limited in their effective decoupling bandwidth.  Adiabatic decoupling schemes use shaped adiabatic pulses and have become more and more common over the last couple of decades due, in large part, to the flexibility of modern NMR instruments to generate shaped pulses.  Adiabatic decoupling schemes (e.g. WURST) use much less power than those using conventional rectangular pulses. thereby reducing or eliminating problems associated with sample heating.  Due to the lower power requirements and increased effectiveness over wider frequency ranges, adiabatic decoupling schemes are ideally suited for X nucleus decoupling at higher field strengths. The figure below shows 500 MHz ¹H [³¹P] NMR spectra measured with inverse gated decoupling for the P-CH₃ methyl resonance of dimethyl methylphosphonate  The single scan spectra were collected in a pseudo-2D fashion, as a function of the decoupler offset frequency from -256 ppm to +256 ppm from the ³¹P resonance frequency in 1 ppm steps.  The acquisition time and recycle time for each FID were 2 sec and 4 sec, respectively.   In the left-hand panel, broadband GARP decoupling was employed at a power of 1.22 W  (60 µsec 90° pulses).  In the right-hand panel, WURST decoupling was used at a peak power of 0.755 W (2 ms WURST pulses, bandwidth = 250 ppm).

Clearly, the data collected with WURST decoupling, at lower power, have a much larger decoupling range (250 ppm) compared to the data collected with GARP decoupling (98 ppm).  Furthermore, while the GARP data were collected, the sample temperature increased and had to be compensated for by the variable temperature unit.  No such temperature increase was observed while collecting  the WURST data.  It is also interesting to note that, in the case of GARP decoupling, distorted line shapes are observed just outside of the decoupling range, while for WURST decoupling, the spectra are fully coupled just outside of the decoupling range with a very sharp transition between being fully coupled and fully decoupled.  For broadband decoupling, WURST is best!   

NMR and the Taste of Christmas - Gingerbread

The holiday season is full of delicious treats.  Aside from rum spiked eggnog, candy canes, fruitcake, shortbread and cranberry sauce, one of my favorite Christmas treats is gingerbread.  Whether you enjoy biting the limbs off a gingerbread man or munching on the roof of a gingerbread house, you cannot escape the wonderful aroma and flavor of ginger and cinnamon.  These fragrant spices can be easily examined by NMR spectroscopy.  The figure below shows the 600 MHz ¹H NMR spectra of CDCl₃ extracts of ground ginger (top) and ground cinnamon (bottom).

The main constituents of these extracts are 6-gingerol and cinnamaldehyde for the ginger and cinnamon extracts, respectively.  Think of these compounds and their NMR spectra while you bite the head off your next gingerbread man in front of your beautifully decorated Christmas tree.  Merry Christmas!  

NOAH - Faster 2D Data Collection

NMR users typically run ¹H, ¹³C, COSY, HSQC, HMBC and NOESY spectra to elucidate the structures of small molecules. Even with cryogenically cooled probes and pulsed field gradient accelerated methods, collecting 2D spectra can be quite time consuming. For concentrated samples, each 2D experiment will typically take minutes to tens of minutes to collect. Much of this time is the result of waiting for T₁ relaxation in each of the experiments. Recently, Kupce and Claridge^1,2 have developed a technique using standard NMR hardware where multiple 2D methods are concatenated in a single super pulse sequence employing a single relaxation delay. They have called the technique NOAH (NMR by Ordered Acquisition using ¹H detection) The time saving of the NOAH technique compared to individually collected 2D spectra results from waiting a single relaxation delay for all experiments rather than a single relaxation delay for each separately acquired spectrum. The data for each spectrum is acquired in separate memory blocks which are separated after data collection allowing the data for each 2D method to be processed individually. The data can also be processed in automation. The authors have kindly made this method accessible to all Bruker users through the Bruker User Library which contains pulse sequences, parameter sets, automation scripts and detailed instructions. The left-hand panel of the figure below shows the 600 MHz HMBC, Ed-HSQC and COSY spectra obtained from the NOAH-3 BSC (HMBC, HSQC, COSY) pulse sequence for sucrose in DMSO-d6.  The right-hand panel shows separately acquired 2D data sets for comparison.

The NOAH spectra were obtained from the raw concatenated data with the automation script provided. The high quality NOAH-3 data using 2 scans, 256 increments and a 2 second relaxation delay, took only 24 minutes to acquire in comparison to the separately acquired 2D spectra obtained with similar parameters, which took a total of 59 minutes to acquire. This represents a time saving of 35 minutes or 59%. It should also be noted that the data from the NOAH-3 BSC sequence is of comparable quality to that of the individually collected spectra.

1. Eriks Kupce and Tim D. W. Claridge.  Chem. Commun. 54, 7139 (2018).
2. Eriks Kupce and Tim D. W. Claridge. Angew. Chem. Int. Ed., 56, 11779 (2017).

Pure-Shift HSQC

Pure-shift NMR has become more and more common over the last few years.  A special issue of Magnetic Resonance in Chemistry has recently been dedicated to developments in these methods.  Pure-shift NMR methods offer simplified proton NMR spectra free of ¹H - ¹H coupling.  These methods have been extended to proton detected 2D NMR measurements, yielding 2D data sets with higher proton resolution compared to conventional 2D measurements.  The NMR Methodology Group at the University of Manchester has been a primary contributor to this technique and has kindly shared their efforts on-line.  The figure below compares the 600 MHz partial Pure-Shift HSQC spectrum of sucrose in DMSO-d6 to a more conventional HSQC spectrum acquired under similar conditions.  The projections on the spectra are independently collected high resolution ¹H NMR spectra.  Clearly, the  Pure-Shift HSQC data have higher ¹H resolution than the more conventional HSQC.  What may not be so obvious from the figure is that the sensitivity is also improved in the Pure-Shift HSQC.  The gain in signal-to-noise-ratio depends strongly on the degree of coupling collapsed.  For some signals in this spectrum, an improvement in the signal-to-noise-ratio as high as 72% was observed.

The the top and middle panels of the figure below show the 1D ¹H projections of the Pure-Shift HSQC and HSQC data from the above figure, respectively.  The bottom panel is the conventional high resolution ¹H spectrum for comparison.

Clearly, the Pure-Shift HSQC proton projection offers much improved resolution.   

13C Satellite Observation with 1D 1H - 13C HSQC

Two dimensional ¹H - ¹³C or  ¹H - ¹⁵N HSQC spectra are typically used to obtain one-bond heteronuclear correlation information between ¹H and ¹³C or ¹H and ¹⁵N.  The data are proton detected and typically employ pulsed field gradients for coherence selection.  The technique uses an incremented delay for chemical shift correlation and essentially discards the very intense signals from protons bound to ¹²C or ¹⁴N while enhancing the remaining doublet signals for protons bound to ¹³C or ¹⁵N.  Usually the ¹H FIDs are collected with ¹³C or ¹⁵N decoupling to collapse the doublets into singlets in the F2 domain leaving a single correlation between proton signals in F2 at the ¹³C or ¹⁵N chemical shifts in F1 of the nuclei to which the protons are bound.   One can use a 1D version of this experiment (without the incremented delay for chemical shift correlation) to collect ¹H NMR spectra exclusively of the protons bound to ¹³C or ¹⁵N.  If used without ¹³C or ¹⁵N decoupling, it allows easy observation of the ¹³C or ¹⁵N satellites in the absence of the very intense signals due to protons bound to ¹²C or ¹⁴N and allows the precise measurement of one-bond  ¹H - ¹³C  or  ¹H - ¹⁵N coupling constants.  The top panel of the figure below shows the 300 MHz 1D  ¹H - ¹³C HSQC spectrum of ethyl acetate.  Clearly the ¹³C satellites are observed and the  ¹H - ¹²C signals are suppressed.  The ¹H spectrum of ethyl acetate is shown in the bottom panel of the figure for comparison.

Glycine as a 13C CPMAS Setup Sample

Glycine is an excellent setup compound for ¹³C CPMAS NMR measurements.  Its utility in this regard has been described in detail.^1,2  It can easily be observed in one scan and has reasonably short ¹H T₁'s, allowing it to be used for ¹H 90° pulse calibration and to setup the Hartmann Hahn matching condition.  The width of the methylene carbon signal can be conveniently used to evaluate the proton decoupling efficiency.  The width and shape of the carbonyl signal are very sensitive to the angle at which the sample is spun and can be used to set the magic angle with a reasonably high degree of precision.  In addition the carbonyl resonance is sharp and can be used as a secondary standard for chemical shift calibration.  If used as a secondary chemical shift standard, one must be aware that glycine has three polymorphic forms, each with different carbonyl chemical shifts.  The polymorphic form is not generally displayed on the reagent bottle and different suppliers may provide different polymorphs or mixtures of polymorphs.  It is therefore important to know which polymorph is being used to calibrate the chemical shift scale.   The α and γ polymorphs are the most common and stable, while the β polymorph is less stable and easily converted over time to the α polymorph.  Furthermore, the β polymorph has a very short ¹H T_1ρ at room temperature and therefore difficult to observe with typical millisecond CP contact times.  The γ polymorph can be converted to the α polymorph at 165°C.  The chemical shifts for the carbonyl resonances for the α and γ polymorphs are 176.5 ppm and 174.6 ppm, respectively.¹ The chemical shift of the β polymorph is between that of the α and γ polymorphs however, it is not usually observed.  The figure below shows the carbonyl region of the ¹³C CPMAS spectrum of three samples of glycine: pure α, pure γ and a mixture of the α and γ polymorphs.

If a single carbonyl resonance is observed for a sample of glycine using typical millisecond CP contact times, one can determine if it is the α or γ polymorph by measuring its chemical shift with respect to another chemical shift standard.  Alternatively, since the ¹H T_1ρ characteristics for the α and γ polymorphs are quite different from one another at room temperature, the authors of reference 1 report that a CPMAS spectrum collected with a 20 msec contact time will show almost no signal for the carbonyl carbon of the γ polymorph.  The carbonyl signal of the α polymorph, on the other hand, will be only slightly attenuated compared to a CPMAS spectrum measured with a 1 msec contact time.

1.  M.J. Potrzebowski, P. Tekely, Y. Dusausoy. Solid State Nuclear Magnetic Resonance. 11, 253 (1998).
2. R. E. Taylor. Concepts in Magnetic Resonance. 22A, 1 (2004).

Information-Rich 13C Satellites

Seemingly simple NMR spectra often contain much more information than one might think.  For example, the ¹H NMR spectrum of 1,4-dioxane is primarily a singlet from which one obtains only an isotropic ¹H chemical shift value.  There is however much more information available in the spectrum which is often not recognized or used.  The ¹H NMR spectrum of a naturally occurring sample of 1,4-dioxane is the weighted sum of the ¹H spectra of all possible isotopomers.  It is the dominant tetra-¹²C isotopomer that gives rise to the singlet but since ¹³C (spin I = 1/2) is 1.1% naturally abundant, one expects to observe also the mono-¹³C isotopomer.  The di-, tri- and tetra-¹³C isotopomers are very rare and can be neglected.  The symmetry in the mono-¹³C isotopomer is lost compared to the tetra-¹²C isotopomer and one obtains a complex second-order spectrum, part of which can be represented by an AA'BB'X spin system.  The spectrum of the AA'BB'X spin system depends on many more parameters than just the isotropic ¹H chemical shift.   This is illustrated in the figure below.

The bottom panel of the figure is the measured 300 MHz ¹H NMR spectrum of 1,4-dioxane with an exaggerated vertical scale to accentuate the ¹³C satellites resulting from the protons color coded in pink in the mono-¹³C isotopomer.  The large central region of the spectrum is the result of all the protons color coded in yellow from both the tetra-¹²C and mono-¹³C isotopomers.   A simulation of this second-order spectrum was calculated from the parameters below and is shown in the top panel of the figure. 

Any isotope shifts in the ¹H frequencies due to ¹³C vs ¹²C bonding were neglected in the simulation.  The fit of the simulation to the ¹³C satellites is particularly sensitive to ¹J_C-Ha,  ¹J_C-Hb,  ³J_Ha-Hc,  ³J_Ha-Hd, ³J_Hb-Hc and ³J_Hb-Hd and much less sensitive to  ²J_C-Hc,  ²J_C-Hd,  ²J_Ha-Hb and  ²J_Hc-Hd.  A fit of the simulation to the experimental spectrum produces estimates for all of the coupling constants in the AA'BB'X spin system - much more information than a single ¹H isotropic chemical shift!

Distortions due to Lock Saturation

The amplitude of the ²H lock signal provides information for an electronic feedback circuit which continuously corrects the magnetic field strength (by way of a B₀ shim) to compensate for environmental instability.   A poor ²H lock signal will provide unreliable input for the feedback circuit and B₀ compensation will be erratic.  This leads to undesirable effects in NMR spectra.  For example, noisy lock signals will lead to undesirable noise at the base of the observed NMR peaks.  If one uses too much lock power, the ²H lock signal gets saturated and the lock amplitude is unstable.  A saturated ²H lock will lead to problems in the NMR spectrum since the input to the B₀ compensation feedback circuit is unstable.  This is demonstrated in the figure below.

When one scan is collected, there are spectral distortions at the base of the NMR resonances.  When 16 scans are collected these artifacts average to produce a general broadening at the base of the NMR resonances.  Be careful not to saturate the ²H lock.

The limitations of 19F GARP Decoupling

In a previous post, it was shown that distorted line shapes are obtained for resonances in broadband decoupled NMR spectra when the resonances of the decoupled nuclide are outside of the effective decoupling bandwidth.  This can be a particularly difficult problem when observing ¹H NMR spectra with ¹⁹F decoupling.  ¹⁹F has a large chemicals shift range so, if there are multiple widely spaced ¹⁹F resonances, it will be difficult or impossible to decouple all ¹⁹F sites at once, particularly at higher magnetic field strengths.  If one is not aware of this problem, data misinterpretation may be an issue as distorted line shapes will  lead incorrect splittings used to measure coupling constants.  The figure below illustrates this problem.  The top three panels of the figure show the 300 MHz ¹H[¹⁹F] NMR spectra for the three ¹H resonances of 1,2-difluoropyridine as a function of the ¹⁹F decoupler offset.  The GARP decoupling scheme was used with 90° pulses of 80 µsec.  The decoupler offsets, depicted in the bottom panel of the figure, were varied in 5 ppm increments.

Of the 11 decoupler offsets used, only offset 6 (at -116 ppm) effectively decoupled both ¹⁹F sites.  Varying the decoupler offset by only ± 5 ppm leads to distorted line shapes, which are particularly pronounced for the H₃ resonance.  These distorted line shapes could easily lead to data misinterpretation and erroneous coupling constants.  In this case, the ¹⁹F decoupling bandwidth is 55 ppm.  Since the chemical shift difference between the two  ¹⁹F resonances is 52 ppm, one is able to obtain a fully ¹⁹F decoupled ¹H spectrum with the careful choice of the decoupler offset frequency however, there will be cases where the decoupling bandwidth would not be sufficient to decouple all ¹⁹F resonances in some molecules.  How then can one generally evaluate all of the coupling constants in fluorine containing molecules?  The ¹⁹F-¹⁹F couplings can be evaluated in a ¹⁹F[¹H] spectrum (not shown).  Specific ¹H-¹⁹F coupling constants can be determined by measuring a ¹H PSYCHE spectrum or be collecting ¹H spectra with selective ¹⁹F continuous wave (CW) decoupling  for each of the¹⁹F resonances.  The latter is shown in the figure below.  The bottom panel shows a standard ¹H spectrum.  The middle two panels show the ¹H spectra for each of the ¹⁹F sites decoupled separately using CW decoupling.  The top panel shows the fully ¹⁹F decoupled spectrum.

Using these data, all of the coupling constants can be evaluated and are shown in the figure below.

In conclusion, one must be careful in interpreting ¹H[¹⁹F] spectra and understand the limits of the ¹⁹F decoupling scheme used.

Decoupling Bandwidth and Distorted Line Shapes

Broadband X nucleus decoupling (X = ¹³C, ¹⁵N, ³¹P, ¹¹B, ¹⁹F etc.....) is frequently used in ¹H detected 2D HSQC/HMQC data collection or in standard 1D ¹H spectra to aid in structure assignment.  When broadband decoupling schemes are used, one must keep in mind that they are not infinitely broadbanded.   They have finite bandwidths over which they are effective thus limiting the chemical shift range for the decoupled nuclide.  The effective bandwidth depends on the particular decoupling scheme and the decoupling power used.  If multiple peaks are to be decoupled, one must insure that all peaks are within the decoupler bandwidth.  One can determine experimentally the effective decoupling bandwidth by running a series of ¹H spectra varying the decoupler offset frequency.  Such a measurement is shown in the figure below for the P-CH₃ methyl resonance of dimethyl methylphosphonate. 300 MHz ¹H [³¹P] NMR spectra were collected in a pseudo-2D fashion, incrementing the decoupler offset frequency from 200 ppm to -200 ppm from the ³¹P resonance frequency in 1 ppm steps.  Broadband GARP decoupling was employed with a power of 3125 Hz  (80 µsec 90° pulses).  The pseudo-2D contour plot is shown in the left-hand panel and a stacked plot is shown in the right-hand panel.

One can see that the effective decoupling bandwidth is 15.56 kHz or 128 ppm on a 300 MHz instrument.  When the decoupler offset exceeds ±64 ppm from the ³¹P resonance frequency, one obtains distorted line shapes.  Representative distorted line shapes are shown in the figure below.  The bottom spectrum was collected with no ³¹P decoupling.  The top, fully decoupled spectrum was collected with on-resonance ³¹P GARP decoupling.  The middle spectra, highlighted in pink, are representative distorted spectra outside of the effective decoupling bandwidth.

If one is not aware of the decoupling offset and available bandwidth, one may obtain misleading line shapes subject to misinterpretation.

NMR of Toothpaste

Some common household products contain many NMR active nuclides able to provide information on the identify the major components of the product.  Toothpaste is such an example.  It contains abrasives, surfactants, cleansers, fluoride, sweeteners, foaming agents, flavors, etc.... A survey of some of the NMR active nuclides can reveal the major components.  The figure below shows the ¹⁹F, ³¹P,  ²³Na, ¹³C and ¹H NMR spectra of a D₂O slurry of Crest Complete toothpaste acquired on a 300 MHz spectrometer as well as the ²⁹Si CP/MAS NMR spectrum of a sample of dried Crest Complete toothpaste, collected on a 200 MHz spectrometer.  Except for the ²⁹Si CP/MAS spectrum, which was collected over several hours, all other spectra were collected in a matter of minutes.

The ¹⁹F spectrum is consistent with the fluoride ion which is a well known agent for preventing tooth decay.  The ³¹P spectrum collected with ¹H decoupling shows two major peaks consistent with diphosphate and phosphate anions.  Salts of these anions are used as water retention agents, stabilizers and emulsifiers.  The ²³Na spectrum shows a single peak consistent with sodium cations, balancing the charge for the fluoride, diphosphate and phosphate anions.  The ¹³C and ¹H NMR spectra show one major component consistent with sorbitol, commonly used as a sweetener.  Other minor components are evident in both the aliphatic and aromatic regions of the ¹H and ¹³C spectra.  The ²⁹Si CP/MAS spectrum of the dried toothpaste is consistent with silica, which is used as an abrasive.  The two peaks are due to Q4 (Si(OSi)₄) and Q3 (Si(OH)(0Si)₃) silicon sites.  It should be noted that there are many other components including flavoring agents, coloring agents and preservatives present in concentrations which would require much more time and attention to identify.   

Appropriate Choice of Presaturation Time

Presaturation is one of the most common methods of solvent suppression.  A long selective low power pulse is applied at the solvent frequency followed by a hard non-selective read pulse (or composite pulse).  Aside from a well-shimmed homogeneous magnet, there are two important parameters required for effective presaturation: saturation power and saturation time.  The selection of saturation power was addressed in a previous post.  With a properly selected saturation power, the appropriate choice for the saturation time depends on the relaxation properties of the solvent and the B₁ field inhomogeneity of the probe.  To determine an appropriate saturation time experimentally, one can run a pseudo 2D saturation pulse sequence like the one in the figure below.

This sequence uses a recycle delay, D1, which is the sum of the incremented presaturation time and a resting delay.  Each FID is Fourier transformed but no Fourier transform is done with respect to the incremented presaturation time.  The result of this sequence for a plant extract dissolved in H₂O/D₂O on a 300 MHz spectrometer using a saturation power of 38.4 Hz (54 dB) and a recycle time, D1, of 5 seconds is shown in the figure below.

 

A partial proton spectrum is displayed on the horizontal axis with the saturation time on the vertical axis, incremented in 50 msec steps.  Several selected 1D spectra are shown on the right with the vertical scale adjusted such that the water signal at 4.7 ppm is at full-scale. Clearly, resonances of the plant extract at 5.08 ppm, 4.51 ppm and 4.48 ppm are independent of the saturation time, whereas the water signal decays as a function of saturation time.  For saturation times between 0 and ~1.3 sec, the intensity of the water signal follows a decaying sinusoidal curve with positive or negative phases depending on the duration of the saturation pulse.  For the saturation power used in this measurement, the 90° pulse is 6.5 msec therefore the trend observed is not the primary ¹H nutation curve but results from sampling the primary nutation curve in 50 msec increments. Since the Nyquist sampling condition is not met with sampling intervals of 50 msec, one observes an aliased nutation curve with a much lower frequency. The overall decay is due to relaxation and B₁ inhomogeneity. After ~1.3 seconds, the water signal is saturated and the data are invariant for longer presatutation times.  These data suggest that the minimum saturation time should be set >1.3 sec.

Field Dependence of a Simple Spin System

With the recent re-emergence of low-field NMR spectrometers at proton frequencies of 40, 60, 80 and 100 MHz, many younger NMR users (who have grown up with high-field spectrometers) are encountering more and more second-order spectra. These spectra are observed when the frequency difference between signals is comparable to the coupling between them.  On a 600 MHz spectrometer, 1 ppm in a ¹H spectrum = 600 Hz while on a 60 MHz spectrometer, 1 ppm in a ¹H spectrum is only 60 Hz.  Unlike frequency differences between signals (in Hz) which depend on the field strength, the coupling between signals (in Hz) is field invariant. Easily interpreted first-order spectra on high-field instruments can be information rich but much more complicated second-order spectra on low-field instruments.  The figure below shows simulated ¹H NMR spectra of a fictitious isolated ethyl group as a function of field strength.  The difference in chemical shift between the -CH₃ and -CH₂- signals is 0.5 ppm and the ³J_H-H  coupling constant is 10 Hz.  The spectra are plotted on a ppm scale on the left and on a Hz scale on the right.  At higher fields, one immediately recognizes the familiar triplet and quartet.  At lower fields, the spectra are much more complicated.  The signals are closer to one another (in Hz) and therefore have more second-order character as the frequency difference between signals becomes comparable to the coupling between them.

How Much Presaturation Power is Needed?

Measuring ¹H NMR spectra of samples in water (or mixtures of H₂O and D₂O) usually requires some form of solvent suppression of which selective presaturation is the most common method.  In this technique, a long low power selective pulse is applied before the high power excitation pulse (or composite pulse).  If too little power is used for presaturation, the water signal will not be sufficiently suppressed.  If too much power is used, one loses intensity of signals close to the water resonance and foregoes the quantitative nature of the NMR data.  The question then arises as to how much power is required for presaturation.  The figure below shows the 300 MHz ¹H NMR spectra of a plant extract in H₂O/D₂O measured as a function of presaturation power.  The data were acquired on a Bruker AVANCE II console with the Bruker zgcppr pulse sequence using a two second presaturation pulse.  The power levels are reported in dB and Hz.

One can see that at higher presaturation powers, one loses intensity for peaks near the water signal.  In this case, where the nearest resonance of interest is 52 Hz from the water signal, one can obtain unattenuated signals with a presaturation field strength of 27 Hz.  This corresponds to 58 dB on this instrument.  

Improved Solvent Suppression with Composite Pulses

A hard rf pulse delivers an rf field to the NMR sample inside of the coil. The rf field is not perfectly homogeneous nor does it end abruptly at the edge of the coil. When a long, selective, low power presaturation pulse is given to suppress a water signal, the water in the coil may be fully saturated whereas water outside of the coil will not be. Furthermore, unsaturated water outside of the coil may be outside of the carefully shimmed region of the magnet and give rise to a broadened residual signal when presaturation is used prior to a hard 90° excitation pulse. One long known* way to avoid this residual broad signal is to use a composite 90° pulse after presaturation.  One of the simplest such pulses is a (90°_x - 90°_y - 90°_-x - 90°_-y) composite pulse which is designed to excite sample inside of the coil but not outside of the coil. As a result, when it is used after a water presaturation pulse, it will not excite the broad signal from the unsaturated water outside of the coil and it will provide a spectrum with better presaturation performance. The figure below shows a small portion of the ¹H spectrum of a plant extract in H₂O/D₂O.

The water signal is highlighted in pink. The left panel shows a conventional spectrum acquired with a 90° pulse. The center panel is a spectrum of the same sample where a two second low power presaturation pulse preceded the 90° hard pulse (Bruker pulse program = zgpr). One can see that most of the water is suppressed from the presaturation pulse however, a broad water signal remains from the water outside of the coil. The spectrum in the right-hand panel is the same as that in the center except that the 90° pulse was replaced with a composite 90° pulse (Bruker pulse program = zgcppr). Clearly, the broad signal from the unsaturated water outside of the coil is essentially gone providing a spectrum with much better water suppression.

*A. Bax.  J. Magn. Res. 65, 142 (1985).