Nuclear Magnetic Resonance Most of you should have learned about NMR in your organic course. Just to underscore that learning, let me say that NMR is arguably the best technique we have for characterizing of chemical species. In the work you do in this course, you will often need to take several kinds of spectra on a single sample, and these notes are provided to help you do that, and more importantly, to analyze the resulting spectra properly. In general, we are interested in seven principal spectra types: 1H (or proton); 13C (or carbon), cosy, hsqc, hmbc, noesy, DEPT. Many more experiments are available, but we will concentrate on these. A. Simple proton spectra (1H nmr), of the sort that were discussed in your organic course. These are so-called one dimensional (1D) spectra that display separate signals resulting from each type of proton in the sample solution under investigation. 1) Solvents used for this usually are chosen so as to have no protons, either by having none to start with (CCl 4 ) or having protons substituted by deuterium (e.g., CDCl 3, (CD 3 ) 2 C=O, CD 3 CN, etc.). a) Note that deuterated solvents will have a trace of singly protonated molecules (e.g., CHCl 3, CD 3 C(=O)CD 2 H, CD 2 HCN, etc), which provide an internal standard to the spectra you acquire. Tables are available (one is in this manual) that provide the ppm position of the signals of such molecules in the proton spectrum. b) Note that if the proton you detect is on the same carbon as one or more deuterium atoms, then coupling will be observed between that proton and that (those) deuterium atom(s). Because of the nature of deuterium (spin = 1, whereas proton has spin = 1 / 2 ), each such coupling results in two extra lines in the absorption observed, and the relative intensities of such lines is different from what you'd expect from proton-proton coupling considerations. Note that the same tables generally show the "multiplicity" of such absorptions as well as their positions. b) Note also that there will almost always be a little water present in such solvents, and though most of that water is expected to be fully deuterated (i.e., D 2 O), a trace of that water will be in the form of HOD, and that water will show a signal in the spectrum. The same tables often show the ppm position of this water signal, and it is important to note that this position is variable, depending on the solvent used. 2) Solutions to be examined are prepared with 10 to 100 mg of sample in 300 to 500 mg of solvent, and the resulting solutions are transferred to an nmr tube, in our case, a 5 mm tube. Use 10-20 mg of sample for proton or cosy; use 80-100 mg of sample for carbon, hsqc, hmbc, noesy, or DEPT. a) Note that it is not mass of solvent but volume of resulting solution that matters in the end. The spectrometer has a "sweet spot" space, where the magnetic field is focused and which the detector system is monitoring, and it is necessary that the solution volume be sufficient to make that space be a bit more than filled. "A bit more than filled" is required because the sample is spinning, and that creates a small vortex (a whirlpool). The bottom of the vortex must always be above the "sweet spot" space, and the curvature of the closed end of the nmr tube must be below it.
b) If you use more solvent than is sufficient to fill this volume, then you have made the sample unnecessarily dilute, and this makes the resulting spectra less informative than they could be. c) Most deuterated organic solvents have about the same density, but note that chloroform-d and methylene chloride-d 2 are significantly heavier than solvents like acetone-d 6, acetonitrile-d 3, methanol-d 4, or dimethyl formamide-d 7. Heavier solvents will require a bit more mass to achieve the same solution volume. 3) Position (chemical shift), intensity (integration), and pattern (coupling) are important factors to consider in relating an observed spectrum to a candidate sample structure. These factors should already be familiar to you. Correlation charts relating structure to chemical shift are available. a) Something important to note is this: Most every sample subjected to NMR analysis came from a synthetic process, usually involving a work-up procedure of some kind. In almost every case, organic solvents of various kinds were in contact with the sample, either during synthesis or during workup, or both. It is very important to know (1) what solvents these were for any sample you analyze, (b) what signals such solvents would have given, and then (c) whether such signals were observed in your spectrum. You do not want to mistake a residual solvent signal for part of your sample signal. b) It is wise to attempt to predict the spectrum you expect to see from the sample you believe you have. It is equally wise not to work too hard to force an observed spectrum to fit an expected pattern. If there are discrepancies, you do not want to miss them, since such discrepancies might be telling you you've not got what you think you have, or that you've missed a feature of its structure in your prediction. B. Carbon spectra (13C nmr), display a signal from each carbon atom present in the sample solution, again in a one dimensional (1D) manner. 1) Normally the spectrum is recorded in proton-decoupled mode, which means that any coupling between carbons and attached protons is eliminated, and the signals are almost always observed as singlets. a) Note that this mode does not eliminate deuterium-carbon coupling, so that deuterated carbons will show signals that reflect this coupling. 2) Carbons with attached protons (one or more) tend to give much more intense signals than carbons free of attached protons (so-called quarternary or quat or ipso carbons). There is not a strict relationship between number of carbons af a given type and intensity of signal, so integration is not otherwise reliable. 3) The first two points, taken together, tell you that position is the most important factor to consider, and that intensity is of only limited utility. 4) However, you should note that 13C nmr spectra are inherently much less intense than proton spectra, with the result that acquiring a carbon spectrum can take a very long time. a) The reasons for this include (1) the fact that almost all the hydrogens in a normal sample are protons, but only about 1% of the carbons are 13 C atoms, and only the 13 C's give an nmr signal. the 12 C
atoms are "nmr silent"; (2) the intrinsic signal strength from a proton is on the order of 100 times more intense than the signal from a 13 C atom. b) The result is that it might take acquiring and averaging thousands of 13C scans (from one sample) to achieve the same the signal-to-noise ratios as would be observed in a single 1H scan of that same sample. And acquiring thousands of scans to average requires much more time. Conclusion: acquiring carbon spectra takes a long time, compared to acquiring proton spectra. 4) Again, correlation charts are available relating structure to chemical shift. 5) You should note that most NMR solvents contain carbon atoms, and therefore solvent signals are present in most of these spectra. You need to know where the carbons in your solvent absorb, so you will not misinterpret such signals as if they came from your sample. Please notice that most NMR solvents have no protons attached to carbon, so the solvent signals are not huge, as they would be if they were protonated. However notice also that most NMR solvents are deuterated, and that deuterium atoms produce a coupling pattern for the nuclei with which they interact. In particular, this means that solvent signals in 13C NMR are often multiplets, since the deuterium coupling is not eliminated in the experiment (as was the proton (1H) coupling). Each deuterium atom produces two new peaks in the original 13C signal, so CDCl3 would appear as a 1:1:1 triplet while CD3CN would show a weak singlet for the CN-carbon and a 7 line pattern for the CD3 carbon. The same tables show both position and multiplicity for carbon signals in the various NMR solvents. C. Proton-proton correlation spectra (cosy), which display in a two dimensional (2D) manner the coupling interactions between signals in proton spectra. 1) Generally a cosy spectrum will appear as a square or a rectangle. The spectrum will show the proton spectrum of the sample along both the horizontal and the vertical axes, and spots in the middle indicate that a correlation exists between two signals. 2) Every signal correlates with itself, so a characteristic of these spectra is a series of spots on the diagonal, one spot for each peak in the original 1D spectrum. The important spots, then, are the offdiagonal ones, and these should appear symmetrically about the diagonal. 3) The spots you see are really cross sections of islands of intensity standing on what might be called the correlation plane. It is therefore necessary to cut at several altitudes to see the pattern of these islands of intensity as they are. One sets this level by choosing a floor level and a ceiling level, though the process of choosing those levels may not reflect those names. 4) This technique shows through-the-bonds interactions between protons. Other correlations could be shown, but they would not be cosy spectra. D. Heteroatomic correlation spectra (1H 13C Hetcor, or just "Hetcor") display, again in a two dimensional (2D) manner, the correlation between the signals of carbon atoms in the 13C spectrum of a sample solution and the 1H spectrum signals of protons to which those carbons are directly attached. Modern instruments, such as our 300 and 500 Hz Bruker instruments, normally collect this spectrum in a different way than the classic Hetcor method, so the spectra we acquire have a different name: hsqc.
The advantage of hsqc is that the spectrum is collected with much more sensitivity and therefore in much less time. (For those who want more of the details, the acquisition involves "inverse detection", by which the carbon correlation is obtained through perturbations in the proton spectrum arising from the coupling between protons and the carbons to which they are attached, and since the proton spectrum is so intense, this method provides the proton-carbon correlation spectrum much more quickly.) 1) Again, as a 2D spectrum, an hsqc spectrum will show 1D spectra along its axes, usually carbon on the horizontal and proton on the vertical. 2) Since the two spectra correlated are very different, there is no diagonal of spots, and there is no necessary symmetry to the arrangement of spots. 3) Spots should only appear to mark a carbon that has at least one proton attached directly to it. Quat carbons and carbons bearing only deuterium atoms should not be marked. a) This last point says that though you should see solvent peaks in the 13C spectrum, no spots should appear for those absorptions out in the 2D area. 4) The observant student might note that acquiring an hsqc spectrum is typically much faster than acquiring a carbon spectrum, and much of the carbon spectrum is inherently contained within the hsqc spectrum. Specifically, the hsqc spectrum contains the approximate positions of every carbon signal for carbons that have at least one proton attached. This is a very valuable thing for the investigator to know, since sometimes you don't have enough sample to get a good carbon spectrum. E. The hmbc spectrum is a companion to the hsqc. It is about as easy to acquire, but it provides different information. The hmbc spectrum shows the correlation between proton signals and the signals from the 2 nd and 3 rd carbons away from the carbon on which that proton sits. 1) The "mbc" in the abbreviation stands for "multiple bond correlation". The spectrum acquisition takes advantage of the fact that protons couple both with the carbon on which they sit and with carbons further away. Those coupling constants are usually quite different, though, so the acquisition suppresses the effects of what should be the single bond coupling and magnifies the effects of what should be the two and three bond coupling, to bring out the signals displayed. 2) Note that if the coupling constants are not what they are expected to be, then the spectrum may not show peaks you'd expect to see, or, conversely, might show peaks you'd expect not to see. In particular, you don't always see all the peaks you might predict for the compound structure you have, and at least occasionally, signals showing correlation with the fourth carbon away may be seen. 3) Sometimes you see pairs of extra peaks, one on either side (horizontally) of where the hsqc spectrum had a spot; these spots, which typically do not correlate to any proton signal, result from the single-bond correlation "bleeding through" (my term) their separation, in Hz, measures the one-bond 1 H- 13 C coupling constant for that proton/carbon set, which is typically in the neighborhood of 140 Hz. (For comparison, the 2 and three bond coupling constants are reported to be in the range of 5-20 Hz.) This phenomenon tends to be seen more often with protons that give very strong and/or sharp singlets, such as methyl groups or isolated CH groups.
F. The noesy spectrum can be thought of as a varient on the cosy spectrum. Instead of disclosing coupling connections through the bonds (cosy), the noesy spectrum discloses through-space correlation. That is, protons that are physically close to each other through space will generate a signal in this 2D spectrum. 1) Since this is a 2D spectrum involving proton-proton correlation, you should see a proton spectrum across two sides, horizontal and vertical. 2) And since this is proton-proton correlation, you will again see a diagonal of spots for each signal in the proton spectrum, just as you did in the cosy spectrum. The nice thing about most noesy spectra, though, is that the diagonal spots are usually or mostly negative peaks, while the off-diagonal spots, which reflect the through-space correlation you want to see, are usually positive signals. Thus those reviewing such spectra often look at "positive-only" versions. 3) The correlation generated by the noesy spectrum takes time to develop, and thus a critical factor in the acquisition is "mixing time". Spectra can look quite different from short to longer mixing times during acquisition, so it is often the case that noesy spectra are acquired at two or more such mixing times and compared. 4) Many noesy spectra will show "ridges" of intensity vertical or horizontal bands of intensity that extend out from exceptionally strong peaks along the diagonal. These can obscure regions in which offdiagonal spots might be found. Changing mixing times can help here as well the ridges may be less intense with some mixing times than they are with others. You must be on guard not to mistake a ridgeline intensity for a true off-diagonal spot. The difference is usually noted by authentic signals showing "bull's eye" sets of concentric circles that vary smoothly with different "floor/ceiling" combinations. Signals in ridge lines tend to be irregular. On the Bruker instruments we use, ridge lines are usually first seen in vertical bands, and since authentic signals come in pairs (off diagonal spots), you can usually see at least one spot of the two that is not obscured by a ridge line. G. The DEPT spectrum tells you how many protons are attached to the carbons giving signals in the carbon spectrum. 1) The DEPT experiment involves acquiring three new 13C spectra (in addition to the standard 13C spectrum). These three new spectra look, at first glance, like normal 13C spectra, but the acquisition procedures generate some differences. a) The first spectrum shows only signals from carbons that have at least one H attached (CH, CH 2, CH 3 ) thus "quat" carbons will not appear. b) The second spectrum shows only signals from carbons that have one H attached thus quats, CH 2 and CH 3 carbons will not appear. c) The third spectrum shows signals from CH and CH 3 carbons positive and signals from CH 2 carbons negative.
It should be obvious that a comparison of the original 13C spectrum and the three DEPT spectra should normally allow the experimenter to identify each signal in the original spectrum as arising from a quat, a CH, a CH 2 or a CH 3, carbon. 2) It is possible to add and/or subtract pairs or sets of these four spectra and generate new spectra in which only CH or only CH 2 or only CH 3 signals appear. While we have that capability, the standard DEPT spectra supply all you need to analyze a normal carbon spectrum properly. Additionally, any errors in the addition and subtraction process could leave you with ambiguous results, which examination of the original data should allow you to overcome. 3) You might note that we have available an experiment on the Bruker instruments, called hsqc-ed, or "edited hsqc". The output spectrum from this experiment has the spots from carbons with an even number of protons attached (2) with a different phase from the spots that arise from carbons with an odd number of protons attached (1 or 3). This would provide the equivalent of the third DEPT spectrum within the hsqc output. We have not had a lot of experience with this experiment, but it would give results much more rapidly than would a DEPT experiment combined with a regular hsqc experiment. 3) Note that there is also an experiment available called APT (stands for "attached proton test"), which might be of interest here. In the APT experiment, you get a spectrum that is almost the same as the third DEPT spectrum. The APT output is a single carbon spectrum, in which carbon signals from carbons with 0 or 2 protons attached appear with opposite phase to the carbon signals from carbons with 1 or 3 protons attached (CH 0 and CH 2 down; CH and CH 3 up). This experiment is sometimes called the "poor man's DEPT", since it gives much of the information that the DEPT experiment provides, but in about one quarter the time. 4) Note that the same problems exist for DEPT spectra acquisition as exist for any 13C acquisition, in terms of the time it takes to get a decent signal-to-noise ratio in the resulting spectrum. Now it should be obvious that for an organic compound, this combination of techniques is very powerful, showing as it does the number and kind of carbons, the number and positions of associated protons, and their through-the-bonds and through space associations. It should be equally obvious that the same techniques may tell you less about a metal complex, since there are significant features in the metal complexes that do not include carbons or protons in their substructures. You should therefore plan to obtain as much of this information as seems potentially helpful on each compound you make that has protons or carbons (or both) in it. In this course, you will potentially be competing for time on the larger NMR instruments with students from Analysis II and P.Chem. as well as with students from various research groups. There will be a smaller (60 MHz) instrument available in the 317 laboratory area on which you can obtain preliminary proton spectra. This will allow you to check for purity and general spectrum appearance before submitting samples for running on the larger instruments. The operation of the smaller instrument is supposed to be roughly the same as the operation of the 90 MHz instrument you had access to in your organic course. In order to use the larger instrument efficiently, you will prepare samples for nmr analysis and submit them to your instructor. Your instructor will then see to acquiring the spectra you want during times outside of lab. For your information, we have the capacity to run, unattended, several spectra on each of several samples (up to 24 samples at a time), with only a short set-up time at the front end required to get
things started. We do these multiple runs in the evenings or on weekends, so as to minimize conflicts with other users. The best way to view the spectra resulting from the larger instruments is to look at the spectra on a computer, where you can manipulate and alter viewpoints "on the fly". We have a site license for the spectral work-up program, NUTS, and viewing the spectra on a computer running NUTS is recommended. While there is a bit of a learning curve involved with using NUTS, many of you will have some facility with NUTS already, since that is the software you used in organic to work up and look at your nmr spectra in that course. A little additional knowledge is needed, so a handout is provided to help you get started, and we will demonstrate this in class or in lab early in the term. We will be trying out a new method of getting spectra to you, which involves moving the spectra files from the Bruker instrument to a web server. We will let you know more about this during the term.