You are on page 1of 27

View Online

The Use of N.M.R. Spectroscopy in the Structure Determination of Natural Products : One-Dimensional Methods
1. H. Sadler Department of Chemistry, University of Edinburgh, West Mains Road, Edinburgh EH9 3J.J

1 2 2.1 2.2 2.3 2.4


Downloaded by Instituto Gulbenkian Ciencia (IGC) on 14 July 2011 Published on 01 January 1988 on http://pubs.rsc.org | doi:10.1039/NP9880500101

3 3.1 3.2 4 4.1 4.2 5 5.1 5.1.1 5.1.2 5.2 5.2.1 5.2.2 6 6.1 6.2 6.3 7 7.1 7.1.1 7.1.2 7.1.3 7.2 8 9

Introduction Basic Concepts The Single-Excitation-Pulse Experiment The Multi-pulse Experiment The Spin-Echo Sequence A Polarization-Transfer Sequence : Sensitivity Enhancement Methods for Distinguishing between Carbon- 13 Resonances in Methyl, Methylene, and Methine Groups and Quaternary Carbon- 13 Resonances Methods Excluding Polarization Transfer Methods Utilizing Polarization Transfer Methods for the Simplification of Fully ProtonCoupled Carbon- 13 Spectra Separate-Multiplet Spectra Multiplet Spectra by Editing Heteronuclear Chemical-Shift-Correlation Methods One-Bond Correlations Sequences Employing a Selective Proton Pulse Sequences Employing Only Non-selective Proton Pulses Long- Range Correlations Correlation through Coupling Correlation through Space : Difference Spectroscopy Methods for the Direct Correlation of Carbon- 13 Resonances The Suppression of Strong Signals from Uncoupled Carbon- 13 Nuclei Sensitivity Enhancement and Spectral Simplification Selective Carbon-Carbon Correlation The Location and Correlation of Proton Resonances Homonuclear Double Resonance Difference Spectroscopy Decoupling Difference Spectroscopy Selective Population Transfer Difference Spectroscopy Nuclear Overhauser Effect Difference Spectroscopy Simplification of Proton Spectra by Multiple-Quantum Filtration Reviews and Books References

spectra of low-abundance nuclei (in particular, those of carbon- 13) became routinely available. Further advances in electronics have resulted in the appearance over the past five years of a generation of spectrometers that are capable of obtaining what appear at first sight to be a bewildering variety of spectra. Some of these look like normal spectra, i.e. the position of each peak is determined by a single frequency. These are referred to as one-dimensional spectra. Other spectra look like contour maps or occasionally like views of lunar landscapes. These are known as two-dimensional spectra since each resonance peak is determined by two frequencies. Many of the new techniques that are involved in producing both one- and two-dimensional spectra are frequently described in the literature as routine. In the hands of an experienced operator many of them are - in the same as an appendectomy is a routine operation to a qualified surgeon. One slip and chaos may result. It is particularly important that the spectrometer is correctly set up to carry out such experiments, which are frequently very sensitive to small maladjustments of the instrument. Equally important is the choice of experimental parameters. Errors of either kind unfortunately produce misleading spectra rather than no spectrum at all. Valid interpretation, particularly of two-dimensional spectra, also usually requires considerably more than minimal experience. This review is concerned with methods for obtaining onedimensional spectra. A review of methods for obtaining twodimensional spectra will be published at a later date.

1 Introduction As far as the chemist who is not an n.m.r. spectroscopist is concerned, the development of n.m.r. spectroscopy over the past 40 years may appear to be quantized. After the initial experiments of the research groups of Bloch and Purcel12in the mid and late nineteen forties, the commercial availability of routine CW (continuous-wave) proton spectrometers in the early sixties provided a new non-destructive and quantitative view of the hydrogen location in many molecules and opened up many previously inaccessible areas of organic and inorganic chemistry. The recognition3 that a spectrum that had been obtained by Fourier transformation of a radiofrequency-pulseinduced emission signal was in many ways equivalent to a conventionally obtained absorption n.m.r. spectrum, coupled with development in the electronic computer industry, led to the commercial availability in the early seventies of the FT spectrometer, with its vastly more efficient use of time. Thus
101

2 Basic Concepts The simplest method for obtaining an n.m.r. spectrum by the pulse-Fourier-transform technique involves irradiating the sample with a single radiofrequency pulse, of a few microseconds duration. This is normally short enough to excite all of the nuclei of a given magnetic isotope and results in an emission signal from the excited nuclei. The signal is known as the free induction decay (FID). The duration of this decay for spin-: nuclei in most non-polymeric molecules is in the range 0.2-5 s and the decay signal is collected immediately after the pulse has been applied. Fourier transformation of this decay (or, more frequently, of an accumulation of a set of successively obtained and combined decays) yields the n.m.r. spectrum. Apart from the saving of time that this method provides when compared with the earlier frequency-sweep (or field-sweep) method, an important additional advantage is that the spinexcitation and signal-detection parts of the experiment are separate in time. This allows excitation to be carried out in such a manner that the FID signals contain less information than usual, and transformation of these can yield a simpler spectrum in which only certain signals appear. In these experiments the single excitation pulse of the simple experiment is replaced by a sequence of two or more pulses, separated by fixed delays of a few milliseconds; these are commonly referred to as onedimensional pulse sequences. Systematic variation of one of the delays within a multi-pulse sequence yields a set of FID signals which, if they are subjected to a procedure that involves two Fourier transformations, generate a two-dimensional spectrum. This type of sequence is referred to as a two-dimensional pulse sequence. Although it is not intended in this review to explain how all
5-2

View Online

102

NATURAL PRODUCT REPORTS, 1988

of the pulse sequences that are presented achieve their results, it is useful to be able to follow how some of the simplest sequences work, particularly since these are the building blocks from which many apparently very complex sequences are composed. In many cases the operation of the sequences can be explained in terms of a simple vector model. This will be described first for the standard single-excitation-pulse n.m.r. experiment and then extended to show how two of the most basic pulse sequences produce their results. This section of the discussion is restricted to nuclei of spin one-half.

2.1 The Single-Excitation-Pulse Experiment Any particular line in an n.m.r. spectrum arises from the difference between the populations of two energy levels. For each line, the equilibrium excess of nuclei (with spins oriented a)in the lower energy level is represented by a vector M,, which is aligned in the direction of the external field Bo. The radiofrequency radiation behaves as another magnetic vector B,, of constant size, which rotates in a plane perpendicular to the direction of B,. To make it easier to visualize how signals are distinguished by the spectrometer and how sequences do their job it has become customary to view the system as if one were an observer sitting on the vector B,, which represents the applied radiofrequency radiation. More formally, a set of three mutually perpendicular axes is defined such that the x and y axes rotate, at the frequency of the pulse, about the z axis, which coincides with the direction of the external field B, [Figure l(a)]. This set of axes is frequently referred to as the rotating frame o r e f e r e n ~ eThe radiofrequency vector B, is f .~ therefore stationary in this frame, and its application along, say, the x axis causes all of the magnetization vectors M (one for each line in the spectrum) to be tipped through an angle 8 (known as the flip angle) towards the y axis until the pulse ceases [Figure 1 (b)]. It is easier at this point to think of each of these shifted vectors as being composed of two components. These are (i) a longitudinal component M, (= M cos 0),in the direction of the field, which is less than M , and which reflects the reduced population between the two levels, and (ii) a transverse component M y (= M sin 0),which is detected as the n.m.r. signal. The size of Mg, and thus of the n.m.r signal, is determined by the pulse strength B, (usually several gauss) and its duration, which is usually the experimental parameter that is adjusted. Clearly, a pulse length corresponding to O = 360" would leave the system unchanged. Pulse lengths corresponding to O = 90" and 180" are particularly significant and are commonly used in multi-pulse experiments. A 90" (or n/2) pulse results in a maximum value for M y [Figure l(e)] and hence a maximum signal. There is no residual longitudinal magnetization M y , indicating that the levels are equally populated. (This is not the same as saturation, in which there is no magnetization in any direction.) A 180"(or n) pulse results in no transverse magnetization [Figure I (c)], and therefore no n.m.r. signal, but inverts the populations of the energy levels, i.e. it changes the orientation of every nucleus. After a pulse has been applied, the system returns towards the equilibrium state : the magnetization vectors eventually resume their positions along the z axis and the n.m.r. signals are obtained by recording the decay of the transverse components of M that have been created by applying the pulse. For a 180" pulse the return path is directly along the z axis [Figure l(d)]. This is relatively slow and may take several seconds, depending upon the longitudinal (spin-lattice) relaxation times (T,) of the nuclei. For all other pulse angles the return path that is taken by M is dependent on the difference between the frequency of the pulse and the resonance frequency of the nucleus. It is this difference in behaviour that enables resonances at different frequencies to be distinguished, If the resonance frequency of the nucleus is the same as the pulse frequency then, after a 90" pulse, the magnetization vector M moves back across the y z plane from the y axis to the z axis.

Downloaded by Instituto Gulbenkian Ciencia (IGC) on 14 July 2011 Published on 01 January 1988 on http://pubs.rsc.org | doi:10.1039/NP9880500101

I vr f

3 Vr f

(k)

Figure 1 The single-excitation-pulse n.m.r. experiment. (a) The magnetization vector M a t equilibrium (= M,) in the rotating frame; (b) the position of vector M immediately after a 0" pulse has been applied along the x axis of the rotating frame; (c) the position of vector M immediately after a 180' pulse; (d) the relaxation path taken by vector M after a 180" pulse; (e) the position of vector M immediately after a 90" pulse; (f and g) the relaxation paths taken by the vector M after a 90" pulse where the corresponding spin-resonance frequencies either (0 equal or (8) differ from the pulse frequency; (h and i) the decay of the signal along the y axis of the rotating frame, corresponding to situations (0 and (g); (i and k) the positions of spectral lines after Fourier transformation of (h) and (i).

Thus M, grows from zero to the original value M, and the transverse component Mu decays to zero. If both relaxation processes occur at the same rate, the tip of the M vector moves in a circular path [Figure 1 (f)]. A detector monitoring changes in magnetization along the y axis would observe M y decaying exponentially with time [Figure l(h)]. If the resonance

View Online

NATURAL PRODUCT REPORTS, 1988-1.

H. SADLER

103

Downloaded by Instituto Gulbenkian Ciencia (IGC) on 14 July 2011 Published on 01 January 1988 on http://pubs.rsc.org | doi:10.1039/NP9880500101

frequency of the nucleus is greater than the pulse frequency, then, after a 90" pulse, the vector M also moves clockwise as it returns to the z axis. Therefore, as M, grows to M,, the transverse component (M,J rotates in the xy plane whilst decaying to zero. This rotation is called free precession, and the free-precession frequency of M,, is equal to the difference between the frequencies of resonance and the pulse. If both relaxation rates are equal, the tip of the vector M moves in a spiral path on the surface of a hemisphere [Figure l(g)]. A detector monitoring changes in magnetization along the y axis (My)observes a signal in the form of an exponentially decaying cosine wave [Figure 1(i)]. Since relaxation is relatively slow, magnetization vectors corresponding to signals more than a few hertz away from the pulse frequency precess many times before they return to the z axis. Where several nuclear spins of different resonance frequencies, are present they are distinguished by their precession frequencies, and the FID signal that is obtained consists of a set of exponentially decaying cosine waves, one for each line in the spectrum. Provided the acquisition of signal is started immediately after the pulse, the spectrum that is obtainable by Fourier transformation can be 'phased' so that all of the peaks appear in the pure absorption mode. 2.2 The Multi-pulse Experiment If there is a short delay ( 7 ) of a few milliseconds between applying the pulse and acquiring the signal it is unlikely that it will be possible to phase the spectrum correctly. Free precession occurs during the delay, so that the transverse magnetization vectors have moved relative to each other by the time that acquisition is started. The result in the final spectrum will be that, if a resonance line at the same frequency as the pulse is a pure absorption signal, any line whose magnetization vector has moved to the - y axis will appear inverted and lines whose magnetization vectors are at other positions in the xy plane will appear badly phased (Figure 2). This apparently inconvenient effect is turned to advantage in multi-pulse experiments. The precession frequency of a vector corresponding to one line in a multiplet is the sum of a contribution from the chemical shift (8) and a contribution from the coupling constant (4. By applying an appropriate sequence of pulses, separated by delays which are frequently inversely related to the coupling constants, it is possible to remove the precession due to the chemical shift (&precession) whilst using that due to the coupling (J-precession) to enhance, to remove, or to invert certain resonances and to obtain properly phased spectra. To avoid loss of signal intensity, the total duration of the pulse sequence must be kept short compared with the nuclear relaxation times. Therefore, where possible, couplings in excess of 50 Hz (i.e. principally one-bond couplings) are employed. In all except the simplest sequences the spectrometer must normally be capable of being programmed to apply both observing and decoupling pulses along any of the x, y , -x, and - y directions and also of detecting signals in these directions. The direction along which the pulse is applied (or the signal is detected) is often referred to as the phase of the pulse (or of the detector). Most currently produced spectrometers have this facility to program pulse and detector phases, and on the most

recent spectrometers it is possible to specify phases in increments of 5" in the xy plane. On older spectrometers, however, it is not possible to specify the decoupler phase or independently to specify the observe transmitter and detector phases. This limits (rather than excludes) pulse-sequence work. In many experiments the phases of the pulses and of the detector are systematically changed through a set pattern from one scan to the next; this process is known as phase ~ y c l i n g .This may ~ be a refinement to remove artifacts arising from imperfect or mis-set pulses or from non-orthogonal directions of phases. In some sequences, particularly those involving multiple-quantum transitions, phase cycling is a necessary part of the experiment to eliminate strong unwanted signals. A simple form of phase cycling, the CYCLOPS6 procedure, (in single-pulse-excitation experiments) is commonly employed automatically on many spectrometers to reduce the residual images that are caused by the quadrature detection system. In the remainder of this section an attempt is made to show how two simple one-dimensional pulse sequences achieve their results. The discussion is restricted to the observation of nuclei that are present in low abundance and where the situation is not complicated by the presence of homonuclear coupling. 2.3 The Spin-Echo Sequence Where there is a delay between the excitation pulse and the acquisition of the signal, a properly phaseable spectrum is obtainable by applying a 180" pulse, in the middle of the delay period, to the nucleus that is being observed. This is illustrated (Figure 3) for the carbon- 13 magnetization vectors that correspond to the two lines of a 13C-H doublet where both lines are at frequencies greater than the pulse frequency. The vector Macorresponding to the higher-frequency line arises from those carbon- 13 nuclei whose attached protons are orientated a and the vector M8 corresponding to the lowerfrequency line from carbon- I3 nuclei whose attached protons are orientated p. During the first delay period 7 following the 90" pulse, the vectors Ma and Ms move clockwise and separate in the xy plane, as represented by instant (ii) in Figure 3. Application of a carbon-13 180" pulse rotates each vector through 180", over the surface of a cone, to new positions in the xy plane [Figure 3(iii)]. The vectors continue to move clockwise, at their own frequencies, and since the slower-moving (i.e. MB)is now ahead, the two vectors converge and coincide along the - y axis at the end of the second delay period 7 . This sequence is known as the spin-echo sequence' and the 180" pulse is referred to as a refocussing pulse. Acquisition of the decay signal at the point where the vectors have re-aligned leads to a spectrum which can be properly phased. Additionally, any line-broadening that has been introduced during the period before the 180" pulse was applied as a result of inhomogeneity of the magnetic field is removed during the period between applying the 180" pulse and acquiring data. Provided that the total time (27) between applying the first pulse and acquiring the signal is sufficiently short, this sequence gives a spectrum identical with that from a simple experiment in which a single 90" pulse is applied. Although, so far, there has been no gain over a single-pulse experiment, this spin-echo sequence is probably the most

Figure 2 The effect on the spectrum of introducing a delay ( 7 ) of a few milliseconds between a 90" excitation pulse and the start of the acquisition of signal. Signals (b), (c), and (d) are, respectively, 45", 90", and 180" out of phase with signal (a).

View Online

104

NATURAL PRODUCT REPORTS, 1988

ii

iii

z
I

ii

Ii I

iv

Downloaded by Instituto Gulbenkian Ciencia (IGC) on 14 July 2011 Published on 01 January 1988 on http://pubs.rsc.org | doi:10.1039/NP9880500101

(c 1

Figure 3 The spin-echo pulse sequence. (a) A diagrammatic representation of the sequence; (b) the positions of magnetizationvectors in the rotating frame at instants (it-(iv) of the sequence; (c) views of the xy plane corresponding to (b).

180
'H
i

Broad - Ban d Decouple

Figure 4 A sequence to distinguish carbon-I3 resonances of CH and CH, groups from those from C and CH, groups. The behaviour of the carbon magnetization vectors is shown for CH and CH, groups.

important sub-sequence and is found as a component of nearly all of the more complicated sequences. A slightly modified version* provides a method in which resonances from carbon-13 nuclei of methyl and methine groups are inverted with respect to those from methylene groups and from quaternary carbon-13 nuclei. This is effected by applying a proton 180" pulse simultaneous with the carbon 180" pulse, by using a delay 7 of (24-' seconds, and by employing broadband proton decoupling whilst acquiring the FID signal. This sequence relies on the similarity in magnitude of most onebond C-H coupling constants, which are assumed (for this

discussion) to be identical. The behaviours of CH and CH, groups are considered separately (Figure 4). As previously, the two carbon- 13 magnetization vectors for a CH group separate after the 90" excitation pulse [Figure 4(a)] until, after a period of (24-' seconds, they lie at 180" to each other [Figure 4(b)] and also at 90" to a vector representing the chemical-shift frequency (i.e. the position of the magnetization vector, had proton decoupling been employed). At this point the carbon 180" pulse rotates the vectors to new positions [Figure 4(c)] in the xy plane. If no further pulses were applied the vectors would continue towards the - y axis and coincide there.

View Online

NATURAL PRODUCT REPORTS, 1988-1.

H. SADLER

105

However, the application of a proton 180" pulse immediately following or simultaneous with the carbon 180" pulse changes the orientation of every proton. Thus carbon nuclei that had formerly been bonded to protons that were orientated a are now bonded to protons that are orientated p, and their corresponding magnetization vectors become the slowermoving ones. Similarly, the magnetization vectors of carbon nuclei that are bonded to protons that had formerly been orientated p but which are now orientated a become the fastermoving ones [Figure 4(d)]. Thus the direction of J-precession of these vectors is reversed by the proton 180" pulse, and the vectors now return towards the + y axis, ultimately coinciding there [Figure 4(e)] after a further period of (24-' seconds. It can similarly be deduced that the same behaviour is expected for the four carbon- I3 magnetization vectors for a CH, group. However, this is not so for CH, groups and quaternary carbon atoms. The centre line of a CH, triplet arises from carbon- I3 nuclei whose attached protons are orientated in opposite directions, the high-frequency line arises where both attached protons are orientated a, and the low-frequency line arises where both attached protons are orientated p. At the end [Figure 4(b)] of the first period of (24-l seconds following the application of the carbon 90" pulse the magnetization vectors M , and Maa (corresponding to the outer lines) have moved by 180" with respect to the vector M , (corresponding to the central line), which shows only &-precession. As before, the carbon 180" pulse moves the vectors to new positions (c) and the proton 180"pulse changes the orientation of every proton (d). However, this last pulse results in no net change for the Mapvector and so this continues to the --y direction; although the direction of J-precession of the other two vectors is reversed, their coincident position at this time also results in no net effect, and these also continue towards the --y axis, where all three vectors coincide [Figure 4(e)] at the end of the second period of (24-' seconds. The signals will be inverted compared with those from CH and CH, groups. The precession of the single vector from a quaternary carbon atom is similarly not affected by the proton 180" pulse and lies along the - y axis at the start of signal acquisition. The application of broad-band proton decoupling during acquisition of the FID signal causes the multiplets to collapse, as would be expected since J-precession is removed. A typical spectrum that was obtained in this way (using an average value for lJCH 135 Hz) is shown in Figure 5. of Spectra showing resonances from only CH, groups and quaternary carbon nuclei may be obtained by combining both sequences described above, i.e. by applying the proton 180" pulse on alternate scans.* In this case the signals from CH and CH, groups are alternately positive and negative, and cancel one another. If alternate FID signals are also subtracted rather than added, a spectrum is obtained in which only CH, and CH groups can be seen. In all of these experiments there must be a pre-excitation delay, whose magnitude is governed by the relaxation times of the carbon nuclei, between scans to ensure that M, can adequately recover.

160

140

120

100

80
p.p.m.

60

40

20

Figure 5 The carbon-13 n.m.r. spectrum of linalool, obtained by the sequence of Figure 4. Signals from C and CH, groups appear inverted.

Downloaded by Instituto Gulbenkian Ciencia (IGC) on 14 July 2011 Published on 01 January 1988 on http://pubs.rsc.org | doi:10.1039/NP9880500101

13

d d R = + 5 A w 2

CH

(C

1
I II

l3C

Figure 6 (a) The energy-level populations of a CH system at equilibrium ; (b) the corresponding proton magnetization vectors, where the subscripts refer to the orientation of the coupled carbon13nuclei ;(c) the normal carbon- 13 excitation sequence and spectrum.

2.4 A Polarization-Transfer Sequence : Sensitivity Enhancement Another process that is employed in many pulse sequences is increasing the differences between populations of the energy levels that are responsible for the transitions, to improve the signal intensities. At normal temperatures the population difference is proportional to the gyromagnetic ratio ( y ) of the nucleus. If a nucleus that has a lower value of y, e.g. carbon-13 or nitrogen- 15, is spin-spin-coupled to a nucleus with a high y (frequently protons) it is possible to alter the average difference in populations for the low-y nucleus to that of the high-y nucleus. This is known as polarization transfer and will result for in an intensity gain of yhigh/ylow the low-y nucleus. Thus proton coupling may be used to increase intensities of carbon signals by a factor of four and those of nitrogen by a factor of

ten. This is only a small improvement for carbon-13 since the intensity increase due to the nuclear Overhauser effect (NOE)9that can be obtained during proton decoupling is threefold. However, the negative gyromagnetic ratio for nitrogen- 15 can result in capricious results, including zero signal, where NOE's are involved. Polarization transfer is therefore a far more reliable and effective way of improving the sensitivity when observing this nucleus. A further advantage is that, where polarization transfer is utilized, the pre-excitation delay is governed by the relaxation rates of the protons and not of the observed carbon- 13 or nitrogen- 15 nuclei. This allows an additional saving in time for both nuclei. In the simple situation of a sample such as 13CHC1,, where a single carbon-13 nucleus is coupled to one proton, an increase in the sensitivity in the carbon-13 spectrum is obtainable by inverting the populations of either pair of energy levels corresponding to one of the proton signals. The four energy levels, together with their equilibrium deviations from the average population, are shown in Figure 6(a). The proton

View Online

106
(b)

NATURAL PRODUCT REPORTS, 1988

2
ns-5,!

3
n= + 3 ~
lH

+1

OA

Downloaded by Instituto Gulbenkian Ciencia (IGC) on 14 July 2011 Published on 01 January 1988 on http://pubs.rsc.org | doi:10.1039/NP9880500101

rn
on 3

l -66

Figure 7 (a) The energy-level populations of a CH system after inversion of populations of the aa and ap levels; (b) the corresponding proton magnetization vectors ; (c) a sequence for selective proton inversion followed by carbon- 13 excitation, and the resulting spectrum.

ti

Figure 8 The INEPT pulse sequence, to invert the populations, for example, of all of the higher-frequency carbon-13 satellites in the proton spectrum. The subscripts f indicate that the final 90" proton pulse alternates between the y and --y directions, and the signals that are detected are alternately added to and subtracted from the accumulated FID, on successive scans. The vector diagrams represent the behaviour of the proton magnetization vectors, where the subscripts refer to the orientation of the coupled carbon- 13 nuclei.

magnetization is represented in Figure 6(b) by two vectors M, and Ma, lying on the + z axis. The vector M, corresponds to the excess a-orientated protons that are coupled to aorientated carbon-13 nuclei, and gives rise to transition 3; Ma corresponds to the excess or-orientated protons coupled to P-orientated carbon-13 nuclei, and gives rise to transition 4. In this situation a carbon 90" pulse gives a normal spectrum, as seen in Figure 6(c). If this carbon excitation pulse is preceded by a selective proton 180" pulse, applied on the high-frequency line in the proton spectrum, the populations of the aa and a/3 levels are inverted [Figure 7(a)]. A selective pulse, sometimes called a 'soft' pulse, is one that is of relatively low power and which is sufficiently long (several milliseconds) to affect only a region of a few hertz. The proton magnetization vectors M, and Ma now lie along the - z and + z directions [see Figure 7(b)]; in other

words, those excess a protons whose coupled carbon-13 nuclei were also orientated a are now orientated p. This redistribution of populations is reflected in the carbon-13 spectrum, in which the high-frequency line is inverted and the intensities of lines increase, on average, by a factor of four. This selectivepopulation-inversion (SPI) techniquelo relies on a knowledge of the precise position of the carbon-13 satellites in the normal proton spectrum, and may be used satisfactorily in simple cases. It is used particularly for the observation of isolated nitrogen- 15 nuclei. In the general situation of carbon-13 where there are many resonances, the simultaneous selective pulsing of all highfrequency carbon- 13 satellites in the proton spectrum would be impossible. However, this result may be effected by the INEPT (Insensitive Nuclei Enhancement by Polarization Transfer) technique.l' This again relies on the similarity of magnitude of

View Online

NATURAL PRODUCT REPORTS, 1988-1.

H. SADLER

107

Downloaded by Instituto Gulbenkian Ciencia (IGC) on 14 July 2011 Published on 01 January 1988 on http://pubs.rsc.org | doi:10.1039/NP9880500101

140

120

100

80
p.p. m.

GO

40

20

Figure 9 The fully proton-coupled carbon- I 3 n.m.r. spectra of linalool at 90 MHz. (a) by the INEPT sequence; resonances for quaternary carbons are absent, centre lines of triplets are missing, and the line intensities of the quartet are equal. The sensitivity is much greater than is obtained (b) in the single-pulse-excitation method (without nuclear Overhauser enhancement) for the same experimental time.

all one-bond C-H couplings but uses non-selective (i.e. 'normal') pulses to invert all of the Mu vectors, regardless of their frequencies. In this sequence (Figure 8) the directions (phases) in which the pulses are applied are important and are given as subscripts. The proton magnetization vectors Maand Mpare first transferred [Figure 8 (a)] to the xy plane by the first proton 90; pulse. After a period of (44-' seconds they have precessed to positions that are separated by 90", as shown in Figure 8(b). The proton 180," pulse then rotates these around the x axis and the carbon- 13 180" pulse reverses their direction of travel [Figure 8(c)]. In this respect this sequence resembles the modified spin-echo sequence. After a further period of (44-' seconds the vectors have moved to positions along the - x and + x directions. At this point [Figure 8(d)] a proton 90; pulse, applied along the y axis, rotates the vector Ma to the z direction and the vector Mu to the - 2 direction. This will have been achieved for all such vectors, regardless of their frequencies. Excitation of the carbon- 13 nuclei with a 90" pulse at this point leads to a spectrum in which the high-frequency line of every CH doublet is inverted, and with the same gain in intensity. Had the final proton 90" pulse been applied along the - y direction, the vector M8 would have been inverted, resulting in a carbon spectrum in which the low-frequency lines of each CH doublet had been inverted. Subtraction of this spectrum from the former results in equal intensities for each line. In practice this effect is obtained by alternating the phase ( + y ) of the final proton 90" pulse and alternately adding and subtracting successive FID signals. It can be shown that CH, triplets appear with relative intensities -1, 0, and + 1 and CH, quartets with relative intensities - 1, - 1, + 1, and + 1. A

typical result is shown in Figure 9. As it stands, this sequence does not allow proton decoupling while the signal is being acquired since the multiplet components are opposite in phase, and would cancel each other, yielding no signals. The resolution of this problem is considered in Section 3.2 and the methods that are described there may be used to obtain sensitivityenhanced broad-band proton-decoupled spectra of heteronuclei that are coupled to protons.

3 Methods for Distinguishing between Carbon-13 Resonances in Methyl, Methylene, and Methine Groups and Quaternary Carbon-13 Resonances
Since the advent of routine pulse Fourier-Transform n.m.r. spectroscopy, the classification of carbon- 13 resonances has been made by the single-frequency off-resonance protondecoupling technique12 in which long-range proton<arbon couplings are removed and one-bond couplings are markedly reduced. This method suffers from several disadvantages. The experiment requires two to five times as long as is needed for a broad-band proton-decoupled carbon- 13 n.m.r. spectrum and assignments are ambiguous or even impossible where there is severe overlap of adjacent multiplets or where coupled patterns result from strong homonuclear proton couplings. Several multi-pulse methods are available which do allow unambiguous distinctions in all cases, and the spectra are frequently obtained more rapidly. These methods fall into two classes, according to whether or not polarization transfer is involved.

View Online

108
Broad-Band Decouple
t

NATURAL PRODUCT REPORTS, 1988

'H

'"nib

................
C
/

*&--

l3C

/ CH, /

/
Sequence 3A
0

0
/
180
1 r

'H

BB

1 J

Sequence 3B APT; GASPE.

Downloaded by Instituto Gulbenkian Ciencia (IGC) on 14 July 2011 Published on 01 January 1988 on http://pubs.rsc.org | doi:10.1039/NP9880500101

-10

LJ
l3C

BB
t

1
t

BB

Figure 10 Variation of intensities of carbon-I3 signals of CH,, CH,, and CH groups and of quaternary carbon with the delay 7 in sequences 3A to 3C and with the angle 0 in Sequence 3D. Variations in intensity ( I ) are given by Z = I,; ZCH = Z cos 0; , , ICHp I,, cos28 ; and Zca, = 10c0s38,where 8 = 180J7". =

Sequence 3C APT; GASPE.

3.1 Methods Excluding Polarization Transfer Sequence 3A, with the specific value of (24-' seconds for the delay 7 as discussed above (Section 2.4), goes some way towards classifying carbon resonances.l3 The same result, i.e. resonances from CH, and CH groups appearing inverted with respect to those from CH, groups and quaternary carbons, is obtained by using Sequence 3B or Sequence 3C, which are known as GASPE1415(GAtedSpin-Echo) or APT16 (Attached Proton Test) sequences. In these the proton 180" pulse of Sequence 3A is omitted and replaced by broad-band proton decoupling, which is applied during one of the two delay periods 7 . This has the advantage that the spectrometer does not require the capability to apply pulses on the decoupler channel, although it must be able to switch the decoupler both on and off independent of data acquisition. A theoretical disadvantage is that the duration of these sequences is twice that of Sequence 3A, but in practice this is not a problem in these applications. Indeed, it is likely that Sequence 3B and Sequence 3C will give significantly better spectra than Sequence 3A under comparable conditions since Sequence 3A is dependent on the homogeneity of the proton 180" pulse. In all of Sequences 3A to 3C the maximum peak intensities are only obtained for proton-bearing carbon atoms if the delay 7 is equal to J-l. The variation of signal intensities with 7 is shown in Figure 10, and arises because the individual magnetization vectors of any particular proton- bearing carbon, although symmetrically disposed about the y axis, do not coincide unless 7 is an integral number of periods J-l. This variation can be used to distinguish between signals from C and from CH, groups but less easily between signals from CH and CH, groups. Problems arise, however, from the significant for differences in lJCH protons that are coupled to sp3 (ca 125 Hz), to sp2 (ca 160 Hz), and to sp (ca 250 Hz) carbon. Thus a delay 7 of 8 ms, corresponding to 'JqHfor non-strained saturated carbon atoms, yields negative signals at about 60 % of full intensity for aromatic, alkene, and cyclopropyl CH groups but positive signals for acetylenic CH groups, not to be confused with signals from C or from CH, groups. A delay of 5 4 m s shows signals from CH, and saturated CH, groups with markedly lower intensities than those from saturated CH groups and from quaternary carbon. A delay T of 3 ms effectively nulls aromatic CH signals, leaves small residual signals for CH, and saturated CH, groups, and shows aliphatic CH signals at about 50% of full intensity.

2 00

150

100

50

p.p.m.

Figure 11 Carbon-I3 n.m.r. spectra of carvone, obtained at 90 MHz by the APT sequence (Sequence 3C) with different values for the delay 7 . (a) 7 = 0 ms, all resonances are positive; (b) T = 3.0 ms; (c) 7 = 3.5 ms, principally quaternary resonances; (d) 7 = 5.5 ms, CH and alkene CH, resonances are largest, and CH and CH, signals are inverted; (e) 7 = 8.0 ms, sp2 CH and CH, resonances are smaller, CH and CH, signals are inverted. The intensities of quaternary resonances remained the same for all values of 7 but are low in absolute terms, due to a relatively short (3 s) pre-excitation delay.

View Online

NATURAL PRODUCT REPORTS. 1988-1.

H. SADLER

109

A fair spectrum of resonances from quaternary carbon atoms is obtained by using a mean value for J of 143 Hz (i.e., 7 = 3.5 ms). Signals from CH, and CH, groups are negligible, quaternary carbons give signals of full intensity, and most CH groups give significant ( < 20%) signals, these changing from positive to negative as 'JcH increases. Acetylenic CH signals have full negative intensities. Typical spectra are shown in Figure 11. To some extent it is possible to present subspectra, showing resonances of only one category of carbon, by adding and/or subtracting the spectra that have been obtained at different values of 7 in appropriate proportions. This process is frequently referred to as ' subspectral editing'. For example, a subspectrum of CH signals (of similar lJCH), containing only small signals ( < 10 Yo)from other carbon nuclei, is ~ b t a i n e d by subtracting '~ a spectrum that has been obtained with 7 = 0.4J-' from one that has been obtained with 7 = 0.6J-l, although the intensity of CH signals is only ca one-third of the full value. A variety" of other editing procedures have been described but these will not be discussed, since a more satisfactory and faster method is provided by the DEPT sequence (Sequence 3G; see Section 3.2). In an alternative sequence (Sequence 3D), known18 as SEMUT (Subspectral Editing by a Multiple-quantum Trap, variations in signal intensity are obtained by fixing the interpulse delays at (24-' and selecting different values for the angle 6 of the proton pulse. This pulse transfers part of the signal intensity into unobservable multi-quantum transitions (usually called multiple-quantum coherences), and the variation of signal intensities with 8 parallels the variation of intensity versus 7 that is observed for Sequences 3A to 3C (as shown in Figure 10) such that 6 is equivalent to 180J~". This sequence requires the spectrometer to be capable of applying pulses to the decoupler channel, but a phase-shifting network (although always an advantage) is not necessary. This sequence is reported to be less sensitive to variations in J than Sequence 3B or Sequence 3C, and it is suggested that spectral editing is best performed by obtaining four spectra, with 6 = 0" (I), 6 = 60" (II), 6 = 120" (111), and 6 = 180" (IV), and collecting data for twice as many scans for spectra (11) and (111) as for spectra (I) and (IV). Combining these, according to the Table below, yields acceptable subspectra provided the spread of values of J does not exceed 20 YO : CH, CH, CH C subspectrum subspectrum subspectrum subspectrum

widely from that selected for 7, and this is referred to as 'J crosstalk'. In this sequence an additional group of proton pulses (90~-r-180~-~-90~, known as a 'purging sandwich ') is , introduced before the 6 pulse. The three delays T ~ T,, and 7, take different values, to accommodate a range of values of J. These delays are conveniently selected from the maximum and minimum values of J as follows: 7y1 = 2(Jm,,+O.O7AJ), 7i1= (Jmin J,,,), + and 7i1 = 2(Jmi,- 0.07A4, where A J = By using delays of T~ = 3.86 ms, 7, = 3.17 ms, and 7, = 2.7 ms, coupling ranges of 120-160 Hz for CH, groups, of 120-180Hz for CH, groups, and of 125-190Hz for CH groups can be tolerated with less than 2 % crosstalk. This sequence, however, does require a phase-shifting network for the proton decoupler and extensive phase cycling. For the direct generation of a 'quaternary carbon atoms only' spectrum, the use20v21 Sequence 3D, with 6 = 90, is of probably the most satisfactory. Errors from imperfect carbon 180" pulses can be removed by cycling the phase of the 19pulse in steps of 90", accompanied by simultaneous shifts by 180" in the phase of the detector. When used across the normal spectral region of carbon (i.e. 0-220 p.p.m.), a compromise value of J o f ca 135 Hz may lead to substantial positive signals ( d 8 "/O) for aromatic and alkene CH groups. It is frequently useful to use a delay 7 that is based on a value of J that is appropriate to the region of the spectrum of greatest interest and to set 0 slightly greater than 90, to ensure that any residual CH or CH, signals are negative. A typical spectrum is shown in Figure 12. It has been proposed that a reduction in intensity of the unwanted signals might be achieved by setting the period before the proton 90" pulse to a value of (24-' that corresponds to a compromise value of the delay 7 for aliphatic groups and the following period to a value corresponding to a compromise 7 for aromatic groups. In this modification the carbon 180" pulse must be set at the centre of the total delay period, and not coincident with the proton 90" pulse. A somewhat involved sequence that has been proposed22 in the hope that it would eliminate signals from proton-bearing carbon atoms and show only the signals of quaternary carbon atoms has been shownz3 to give signals of reduced intensity for some CH, groups also. In this discussion it has been assumed that sufficient time elapses between one acquisition of data and the next so that the normal equilibrium population distribution of the energy levels is re-established. This will be governed by the longest carbon relaxation time in the sample. However, a delay of at least 5 s is needed between the scans for most medium-sized molecules otherwise signals from quaternary carbon atoms will be severely reduced. In a single-pulse experiment a reduction in the preexcitation delay is coupled with a smaller pulse angle. In these multi-pulse experiments the pre-excitation delay can be reduced by using a greater value than 90" (e.g. 155") for the first carbon pulse. The distinction between CH, CH,, CH,, CHD, CH,D, CHD,, CD, CD,, and CD, groups by spin-echo sequences has been However, since these methods are more appropriate to analyses in isotopic labelling studies, they will not be examined here.*
Jmax

- Jmin.

Downloaded by Instituto Gulbenkian Ciencia (IGC) on 14 July 2011 Published on 01 January 1988 on http://pubs.rsc.org | doi:10.1039/NP9880500101

= = = =

(I) - (IV) - SII) + +(HI) (I) + (IV) - (11) - (111) (11) - (111) - 81)+ i(IV) (11) + (111) - 51)-8IV)

A modification,19 known as SEMUT GL (Sequence 3E), markedly reduces signals of the wrong category in subspectra. These frequently arise with signals whose values of J differ

'H
l3C

86

Sequence 3D SEMUT.

'H

Tl

I8 3

3.2 Methods Utilizing Polarization Transfer The sequences that are described in this section rely on onebond proton-carbon couplings not only to distinguish between signals from CH,, CH,, and CH groups but also to produce the signals ; therefore quaternary carbon atoms, since they have no directly bonded protons, are not observable. In principle, longrange couplings could be used to observe quaternary carbon atoms; however, these couplings are relatively small and show
* See J. C. Vederas, Nat. Prod. Rep., 1987, 4, 277 for a discussion of applications of n.m.r. to isotopic labelling studies.

Sequence

3E SEMUT GL.

View Online

110

NATURAL PRODUCT REPORTS, 1988

Downloaded by Instituto Gulbenkian Ciencia (IGC) on 14 July 2011 Published on 01 January 1988 on http://pubs.rsc.org | doi:10.1039/NP9880500101

( a1

140

120

100

80
p.p.m.

60

40

20

Figure 12 Carbon-13 n.m.r. spectra of lanosterol at 90 MHz. (a) A normal proton-decoupied spectrum; (b) a spectrum of quaternary carbons (a), obtained by the SEMUT sequence (Sequence 3D), with 8 = 90" and using a pre-excitation delay of 10 s.

Sequence 3F Refocussed INEPT

(7

1/2J).

such wide variations that in practice their use is severely limited. The INEPT sequence'' (see Section 2) alone will result in no signals at all if broad-band proton decoupling is applied during the acquisition period since the positive and negative parts of the multiplets cancel each other. This can be overcome by adding a further delay, A , to the INEPT sequence; 180" pulses are applied to both nuclei, at the centre of this delay before the acquisition of the FID, with broad- band proton decoupling. This allows the components of multiplets to regain the same phase. The sequence (Sequence 3F) is known25as the refocussed INEPT sequence. The optimum delay 7 is (24-1 for all CH, groups since each proton is coupled to only one carbon nucleus. However, signal intensities for CH, CH,, and CH, groups vary differently with the delay d since the rates of J-precession are different. This variation is shown in Figure 13. Approximately equal intensities are obtained for CH, CH,, and CH, signals if d is ca 0.3.J-'; a value of d of ca 0.7J-' will show positive signals for CH and CH, groups and negative signals for CH, groups. Only signals from CH groups should be obtained if d is equal to (24-', but, since the intensities of signals from CH, groups show their maximum rate of change at this point, slight variations in values of J from that selected for d lead to significant signals from CH, groups, and Sequence 3F (like Sequences 3A to 3C) is very susceptible to intensity errors due to differences in values of J . A more accurate CH spectrum is provided26 by the EPT

Figure 13 Variation of intensities of carbon-13 signals of CH,, CH,, and CH groups with the delay d in the refocussed INEPT sequence (Sequence 3F) and with the pulse 8 in the DEPT sequence (Sequence 3G). Variations in intensity (I) are given by ICH= I, sinf?; ICH = I, sin 28; and I C H ,= 31, (sin 38+ sin 8)/4, where 8 = 180JAO.

(Exclusive Polarization Transfer) sequence [Sequence 3G (0 = 90")], which is a special case of the more general DEPTZ7 (Distortionless Enhancement by Polarization Transfer) sequence [Sequence 3G (8 is variable)]. This sequence has fewer pulses than the refocussed INEPT sequence and effects polarization transfer in a different way. Except for the special case of 8 = 90", this cannot be described by using the simple vector pictures that have been presented earlier since multiplequantum effects are involved. In this sequence the separation

View Online

NATURAL PRODUCT REPORTS, 1988-1.

H. SADLER

111

Sequence 3G DEPT ; EPT (0 = 90").

Ac 0

Downloaded by Instituto Gulbenkian Ciencia (IGC) on 14 July 2011 Published on 01 January 1988 on http://pubs.rsc.org | doi:10.1039/NP9880500101

rN

1
I
I

80

60
P.Pm

40

20

Figure 15 The carbon-13 n.m.r. spectra of Figure 14, edited to show separate subspectra for CH, CH,, and CH, resonances.

80

60
p.p.m.

40

20

Figure 14 Carbon- 13 n.m.r. spectra of a triterpene derivative (40 mg in 0.4ml CDCl,) at 90 MHz. Spectrum (a) is a normal protondecoupled spectrum (2000 scans), showing all resonances for sp3 carbon; spectra (b)+d) were obtained by using the DEPT sequence (Sequence 3G) (320 scans each; pre-excitation delay of 3 s) and show the proton-bearing carbons. In spectrum (b) 0 = 45" and all resonances are positive; in (c) 8 = 90, and only CH resonances are visible; in (d) 0 = 135", and whereas CH and CH, resonances are positive, the CH, resonances are negative.

shows CH resonances only; if 0 is ca 130" the spectrum shows CH and CH, groups as positive signals and CH, groups as negative signals. Typical spectra are shown in Figure 14. Features arising from values of J that differ widely from that chosen for the calculation of a suitable inter-pulse delay include the following. Acetylenic CH groups may give very small or zero signals. Alkenemethylene and methylenedioxy groups appear with up to 50% of full intensity in a 'CH group only' (0 = 90") spectrum but are readily identified (as negative signals, of reduced intensity) in the other spectrum (0 = 130"). These signals also lie in the relatively empty region between the aliphatic and the aromatic resonances. Should editing be considered necessary, the DEPT sequence provides more accurate subspectra than the refocussed INEPT sequence. Three spectra are obtained21[using values of 0 of 45" (I), 90" (11), and 135" (111), with twice as many scans being acquired for spectrum (11)] and these are combined as follows: CH, subspectrum CH, CH

between the pulses remains fixed at (24-l and signal intensities are varied by altering the value of the third (0) proton pulse. This variation of intensity (see Figure 13) with 0 parallels that obtained by changing d in Sequence 3F, and there is a general correspondence28between 0 in DEPT and 18OJd" in Sequence 3F. A comparison of the DEPT and refocussed INEPT techniques forms a significant part of a review,' of pulsed methods for polarization transfer in carbon- 13 n.m.r. spectroscopy. Although the total duration of the DEPT sequence is greater, this is of little significance in practice for carbon-13 spectra, and the DEPT sequence is preferred since, among other reasons, it tolerates greater differences in values of J. Residual signals from CH, and CH, groups that might be obtained by using a compromise value for J of 135 Hz are significantly less in an EPT spectrum than for a refocussed INEPT spectrum with the equivalent value of A . In most cases the CH,, CH,, and CH groups are readily distinguished by visual examination of two DEPT spectra. The spectrum that is obtained by using a compromise value for J o f 135 Hz, a preexcitation delay of 5 seconds, and with a value of 0 equal to 90"

= (I) + (111) -(II)/d2 subspectrum = (I) - (111) subspectrum = (11)

In practice these proportions may-need to be adjusted slightly to give the best results, due to imperfections in both the timing and the homogeneity of pulses. Such imperfections may also result in small CH, signals in a CH subspectrum. Unwanted signals arising from a variation of 10 % in J constitute less than 7 % . In particular, CH, subspectra show no residual CH or CH, signals from this source of error; CH, subspectra show no residual CH signals and CH subspectra show less than 1 YOof residual CH, signals. The largest errors arise from CH, signals in a CH subspectrum and for CH, signals in a CH, subspectrum. Negative CH, signals (arising from methoxy-groups or acetyl groups) in a CH subspectrum are usually indicative of too short a pre-excitation delay, but this is a diagnostically useful error. The variations assume that the 90" pulse and the 180" pulse for both nuclei are accurately set. Edited DEPT spectra are shown in Figure 15.

View Online

112

NATURAL PRODUCT REPORTS, 1988

A full discussion2' of sources of errors in DEPT spectra has been given, along with full experimental details for measuring the pulse times and for checking the accuracy of the phase shifts of the proton decoupler pulses. Both Sequence 3F and Sequence 3G require that spectrometers should be used that are capable of providing phase shifts of 90" on the proton-decoupler channel, and therefore it may not be possible to use these sequences on some older spectrometers. It is, however, possible30 to carry out polarization-transfer experiments with spectrometers which do not have this facility, but in which pulses may be applied from the proton decoupler, by using Sequence 3H. This is a modified refocussed INEPT sequence in which the first pair of 180" pulses on both nuclei is omitted from the centre of the delay 7. This results in an extent of polarization transfer (and therefore signal intensity) which, assuming 7 = (24-l and a fixed value of A , varies with sin{180vo/J), where v is the separation between the proton decoupler frequency and the individual proton resonance frequencies. This sine dependence has led to the acronym SINEPT for Sequence 3H. Signal intensities are maximum where the relevant proton resonance lies such that v = (2n+ 1)J/2 and zero when v = nJ. This dependence of signal intensity is markedly reduced by alternating the proton decoupler frequency between two values, differing by O S J , on successive scans. Alternatively, half of the scans may be acquired at one proton transmitter frequency and the accumulated FID be combined with that from the remainder, which is obtained with the shifted frequency. In this way proton-bearing carbon atoms are observed with at least 70% of their 'INEPT' intensities. The DEPT sequence may be similarly modified30 by omitting the second proton 180" pulse to give the MODEPT sequence (Sequence 31). In this sequence, in which 7 = (24-', signals are at a maximum when v = (2n + 1)J/4 and zero when v = nJ/2 for any specified value of 8.In this instance the variations in signal intensity are reduced by choosing proton decoupler frequencies which differ by 0.25J. A new sequence (Sequence 3J), known as POMMIE (Phase Oscillations to MaxiMIze Editing), has been described,' as an alternative to the DEPT technique. This differs from the previously discussed sequences in that, in addition to the interpulse delays being fixed at (24-', the pulses are all 90" and 180" pulses and intensity variations are achieved by cycling the phase of one additional decoupler pulse. Such experiments can only be carried out on the most recent instruments, which employ digital phase-shifting networks that allow incremental
I

Downloaded by Instituto Gulbenkian Ciencia (IGC) on 14 July 2011 Published on 01 January 1988 on http://pubs.rsc.org | doi:10.1039/NP9880500101

variations in phase (e.g. by 5") rather than the standard 90" and 180" shifts that are employed in most spectrometers. This allows more accurate setting of phases and more extensive phase cycling, for which reason POMMIE is expected to be less prone to errors than DEPT. However, this option is not available to most workers at present. 33 and investigations into Some e~perimental~~' the use of polarization transfer from deuterium to carbon- 13 to distinguish between CD, CD,, and CD, groups have been carried out. Selective enhancement of CD groups is straightforward but the distinction between CD, and CD, is more difficult. The possibility of obtaining separate subspectra of CD, CD,, CD,, CDH, CDH,, and CD,H groups by combining polarization transfer with gated spin-echo sequences has been explored34 but in practice these methods are unlikely to be error-free. All of these methods require a spectrometer that is capable of pulsing and decoupling at the deuterium frequency as well as the proton frequency, and a fourth nucleus would be necessary for locked operation.

4 Methods for the Simplification of Fully Proton-Coupled Carbon-I 3 Spectra


Proton-coupled carbon- 13 spectra of natural products are frequently very difficult, if not impossible, to analyse, due to overlap of multiplets from carbon resonances whose chemical shifts lie in the same region of the spectrum. Complete separation is achievable by two-dimensional methods35 but large amounts of data storage and plenty of material are needed for high resolution, and distortions in multiplet spectra may occur if the proton spectrum is not first-order. One-dimensional simplification methods involve obtaining individual spectra for each multiplet or the careful use of editing techniques, based on the number of protons bonded to the carbon- 13 nuclei. In some cases the positive/negative presentation of the simple INEPT sequence with the absence of the centre section of triplets can provide sufficient separation of multiplets. The additional intensity in the outer sections of quartets can also be valuable where sensitivity is low.

180

'H

BB

l3C

T :

90

A 2

180
Y

Sequence 3H SINEPT

(7

= 1/24.

'ti

BB
1 2J

3c

Sequence 31 MODEPT.

4.1 Separate-MultipletSpectra Two methods have been proposed which allow individual proton-coupled carbon- 13 multiplets to be observed separately. One method (Sequence 4A) relies on selective excitation of a carbon resonance at the pulse frequency whilst employing broad-band proton decoupling and switching off the decoupler during the acquisition of the FID, to restore the proton Selective excitation is achieved by using a train of n short pulses, each of flip angle 8 = 90"/n, spaced 7 seconds apart. This alternately tips the magnetization vectors through an angle 8 and allows them to precess, during the delays, through an angle q5 = 180~7, where v is the separation of the vector from the pulse frequency. Vectors from signals located at the pulse do not precess during the delays, and at the end of the pulse train they have in effect received a 90" pulse. Vectors at intervals of 1/7 Hz from the pulse are also tipped through 90" since they precess by an integral number of revolutions during each delay. Provided enough pulses are employed, magnetization vectors from resonances more than a few hertz away from the pulse never stray far from the z axis of the rotating frame, and significant signals are only obtained from resonances within a narrow band f l/n7 Hz around the pulse frequency and at intervals of 1/7 Hz from it. The behaviour of
'H -

'H
1 2J 1 2J

BB

BB
Q

l3C

Sequence 35 POMMIE; the last proton 90" pulse uses a variable phase angle 9.

W u e n c e 4A DANTE; selective excitation.

View Online

NATURAL PRODUCT REPORTS, 1988-1.

H. SADLER

113

these vectors has led to this sequence acquiring the name DANTE (Delays Alternating with Nutations for Tailored Excitation). A typical experiment may employ a train of thirty pulses of 3", at 2 ms intervals, and to be able to define such small pulses it is frequently necessary to attenuate the power to the carbon- 13 signal transmitter. Suitable selection of the interval 7 allows two carbon-13 multiplets that are far enough apart not to overlap to be excited simultaneously, with a consequent saving in time. If excitation employs the equivalent of a 90" pulse, relatively long pre-excitation delays (of 1&20 seconds) may be needed. An alternative to the DANTE sequence relies on selective acquisition3' of the desired signals rather than selective excitation, and similar spectra are obtained. A single, hard excitation pulse is applied at the chemical-shift frequency of the carbon resonance whose multiplet is to be observed and then, after a short delay, t, the FID is acquired (Sequence 4B). As in DANTE, broad-band proton decoupling is applied at all times except during the acquisition of data. During the total number of scans the delay t is regularly incremented by an amount 6t from an initial value of 6t to a final value t,. Selectivity is obtained by making use of the variations in phase that are undergone during the delay periods by signals that are not coincident with the pulse. Every time the delay is extended by one increment the phase of any off-resonance signal, v Hz from the pulse, alters by an amount 4 = 180v6to,whereas the phase of a signal that is located at the pulse and at multiples of 1/6t Hz from it remains unchanged. Provided enough short increments are employed, the alterations in phase of the signal throughout the experiment result in a reduction of the final offresonance signals to an insignificant level, and only signals from a resonance that is located at the pulse frequency and at integral units of 1/6t Hz from it are observed. The application of this method to part of the carbon- 13 n.m.r. spectrum of P-pinene is shown in Figure 16. Similar results are obtained with DANTE. Severe overlap in the fully protoncoupled spectrum (a) prevents satisfactory analysis of the multiplets. Selective-acquisition spectra [(bt-(f)] show the multiplets clearly, and the power of these methods is illustrated by the multiplets that can be obtained for the two methylene carbons whose centres lie only 10 Hz apart. As a guide for general useage, good selectivity can be obtained if not less than 25 increments are used and if the size of the increments is set such that the final delay is not less than 0.8/Af, where Af is the separation between the closest pair of carbon resonances in the proton-decoupled spectrum. The final delay should be made as short as possible, consistent with the desired selectivity, since if it is too long the final signal intensity will be reduced by relaxation. The excitation pulse may be set at the flip angle that is normally used to obtain fully proton-coupled spectra with an appropriate pre-excitation delay, which should be at least one quarter of the acquisition time to retain the nuclear Overhauser enhancement. It is not necessary to reduce the power of the carbon- 13 transmitter.
4.2 Multiplet Spectra by Editing

described above (Section 4.1). However, simply removing the broad-band proton decoupling during the period while data are being acquired in most of the sequences that are described in Section 3 leads to spectra that cannot be properly phased, and in which the relative intensities of the lines within the multiplets are incorrect. Such effects may be inherent in the technique or dependent on the strength and multiplicity of the proton coupling, and are particularly noticeable if the refocussed INEPT [Figure 17(a)] and SEMUT sequences are used without proton decoupling.

Downloaded by Instituto Gulbenkian Ciencia (IGC) on 14 July 2011 Published on 01 January 1988 on http://pubs.rsc.org | doi:10.1039/NP9880500101

28

26

24 p.p.m.

22

20

Figure 16 Carbon-13 n.m.r. spectra of P-pinene at 90 MHz. Spectrum (a) is a fully proton-coupled spectrum; spectra (b)--(f) are selectivemultiplet-acquisition spectra, obtained by Sequence 4B (conditions : 5280 pulses, 22" pulse, pre-excitation delay 0.23 s, acquisition time 1.O s); spectrum (g) is a broad-band proton-decoupled spectrum (320 pulses).

In Section 3 it was described how spin-echo and polarizationtransfer sequences could be used to give separate protondecoupled subspectra for CH, CH,, and CH, groups. In many cases a similar simplification of fully proton-coupled spectra reduces the overlap of signals to an acceptable level. Clearly, obtaining several carbon- 13 multiplets in a single spectrum saves time compared with the individual observation methods
I

60
p.p.m.

40

20

l3C

- L J b
varied

Se<luence 4B Selective acquisition.

Figure 17 Carbon-13 n.m.r. spectra of menthol, obtained at 90 MHz by (a) the refocussed INEPT sequence (Sequence 3F), using A = 2.2 ms, showing intensity and phase distortions, and (b) the INEPT+ sequence, largely eliminating the distortions in (a); (c) is the fully proton-coupled spectrum.

View Online

I I4

NATURAL PRODUCT REPORTS, 1988

Sequence 4C DEPT GL+; the final proton pulse (----) is applied on alternate scans only; extensive phase cycling is required.

Sequence 5A SHECOR; the selective proton 180" pulse (---) applied on alternate scans only.

is

Downloaded by Instituto Gulbenkian Ciencia (IGC) on 14 July 2011 Published on 01 January 1988 on http://pubs.rsc.org | doi:10.1039/NP9880500101

The DEPT sequence without proton decoup1ing2'. is far less prone to these spectral distortions but is not entirely free from them. In all of these sequences, however, it is frequently impossible to achieve completely clear edited subspectra due to variations in the magnitude of the one-bond proton-carbon couplings. To overcome these problems, several modifications to these sequences have been proposed. The addition, for example, of a single proton 90" pulse to the APT and to the refocussed INEPT sequences immediately before data are acquired, to give the APT+ and INEPT+ sequence^,^^.^^ results in distortionless coupled spectra [Figure 17(b)]. However, these sequences often lead to poorly edited subspectra. Probably the most satisfactory modification for proton-bearing carbons is that39 described as the DEPT GL' sequence (Sequence 4C), in which the delays 71,, and 73are set according to the range of 7' J,, values that is required (as described under SEMUT GL; see Section 3.1). In Sequence 4C the final proton 180" pulse is applied on alternate scans only. Satisfactory editing is obtained by using values of 38", 90", and 142" for the 0 pulse and by combining these spectra as described previously (Section 2). Fully coupled spectra in which only quaternary carbons are seen are best obtained by using the analogous SEMUT GL+ sequence.

5 Heteronuclear Chemical-Shift-Correlation Methods


5.1 One-Bond Correlations The correlation of carbon-13 resonances with those of their bonded protons has formerly been obtainedg0from a series of single-frequency proton-decoupled carbon- 13 n.m.r. spectra. Assignment may, however, be ambiguous or impossible in this way in crowded regions of the carbon-13 spectrum and also where the proton spectrum shows strong homonuclear coupling. Elegant proton-carbon correlations can be obtained with a very high degree of certainty by using two-dimensional methods,41 but these frequently need large data matrices and are very time-consuming if only small quantities of material are available or if high resolution is necessary. In many instances a relatively small number of one-dimensional spectra suffice, and represent a considerable saving in time. A variety of onedimensional sequences for C-H correlation are available. All involve selective polarization transfer from specific proton(s), so that the signal from the corresponding coupled carbon nucleus appears with increased intensity. These methods fall into one of two classes, according to whether or not they employ selective (soft) pulses to initiate the polarization transfer process. Clearly, such methods are applicable to the correlation of protons with nuclei other than carbon, but the problems are rarely so complex and can frequently be solved by conventional decoupling experiments. It is important in all experiments that involve selective irradiation of protons whilst other nuclei are being observed that the same probe is used to determine the proton spectrum as will be used for the main experiment. If a switchable probe is not available, satisfactory proton spectra can usually be obtained by using the proton-decoupler coils. In the case of long accumulations it is also necessary to prevent evaporation of solvent since this may result in a shift of the frequencies of the proton resonances.

5.1.I Sequences Employing a Selective Proton Pulse All of these sequences use a selective proton 180" pulse, applied at one of the carbon-13 satellites of a particular proton, to identify the corresponding carbon resonance. In the simplest sequence (Sequence 5A), known as SHECORg2 (Selective HEteronuclear CORrelation), the initial selective proton pulse is followed by a gated spin-echo sequence; the FID is acquired with broad-band decoupling. No phase cycling is necessary and Fourier transformation yields a spectrum in which there is an enhanced fully proton-decoupled signal for the corresponding carbon resonance. Typical durations for the selective proton 180" pulse are 25-50 ms and pre-excitation delays of 2-3 s are used. In practice, non-enhanced signals are removed by applying the selective pulse on alternate scans and alternately adding and subtracting successive FID signals. This has the same effect as subtracting a normal spectrum from a selectively enhanced one. Enhancements for CH, CH,, and CH, signals vary with the delay 7 . In particular, signals are obtained only from CH groups if 7 = (24-'. Nearly even enhancements are given for all proton-bearing carbons if 7 = (44-I; alternatively, if 7 = 3/4J, CH, signals are inverted with respect to CH and CH, signals. Use of this value for the delay 7 is not recommended since signals are inverted (for all values of 7 ) if a low-frequency rather than a high-frequency satellite is irradiated. Where the proton spectrum is well resolved and clearly assignable, a few carefully chosen irradiation positions should suffice to give a correlation of the two spectra. However, in more complex cases the high-frequency carbon-13 satellite of one proton resonance may be very close to the low-frequency satellite of another. In such cases it is preferable to increase the frequency of the selective pulse in steps of 10 Hz across the proton region and to present the data as a stack of one-dimensional spectra. Spectra of the methylene and methyl carbons in linalool that have been obtained in this way are shown in Figure 18. Each carbon resonance shows a maximum positive intensity at the frequency of the high-frequency carbon satellite of the corresponding proton resonance and a maximum negative intensity at the frequency of the corresponding low-frequency carbon satellite. The corresponding proton resonance lies at the average of these two frequencies. Thus the carbon resonances of 6 = 17.7 and S = 25.7 correlate with the alkene methylproton resonances at S = 1.54 and 6 = I .62 respectively and the carbon resonance at S = 42.6 correlates with the methyleneproton multiplet that is centred around S = 1.5. Although the proton region that is examined covers only one satellite from each of the extreme proton resonances, there is no ambiguity in the correlation of the remaining two carbon resonances, that at S = 23.1 correlating with the methylene-proton multiplet centred around S = 2.0 and that at S = 27.7 correlating with the methyl resonance at 6 = 1.21. Additional low-intensity peaks may occasionally arise from long-range couplings. Several variants of Sequence 5A have been proposed which result in roughly a two-fold increase in signal-to-noise ratio.43 The seleaive pulse is applied on every scan and consecutive FID signals are all added. Additionally, a carbon 180" pulse is applied at (i) on alternate scans or a carbon 90" pulse is applied at (i) or at (ii), alternating the phase of this pulse between + x and - x on successive scans. The nulling of un-enhanced resonances is more sensitive to imperfections in the pulses here than in the unmodified sequence.

View Online

NATURAL PRODUCT REPORTS, 1988-1.

H. SADLER

115

Downloaded by Instituto Gulbenkian Ciencia (IGC) on 14 July 2011 Published on 01 January 1988 on http://pubs.rsc.org | doi:10.1039/NP9880500101

45

40

35

30
p.p.m. 13c

25

20

15

Figure 18 A stacked plot of carbon-I3 n.m.r. spectra of CH, and CH, resonances of linalool, obtained at 90 MHz by the SHECOR technique with selective irradiation of protons at intervals of 10 Hz over the range 6 = 1.2 to 6 = 2.0. The power of the proton pulse was 45 db below 0.2 W, its at duration was 59 ms, and the pre-excitation delay was 3 s. Frequencies of correlated protons and carbon nuclei lie ( 0 ) the average of the proton frequencies corresponding to the maximum positive and the maximum negative intensities of the carbon resonances.

l3C
Sequence 5B SEPT; the first hard proton pulse (----) alternate scans only.
is applied on

Sequence 5C SDEPT.

BB

v
Sequence 5D Selective suppression.

A related sequence (Sequence 5B), known as SEPT44 (Selective inEPT), is used in a similar fashion to SHECOR. Un-enhanced signals are removed by applying a normal (i.e. non-selective) proton 180" pulse, simultaneously with the carbon 90" pulse, on alternate scans; successive FID signals are alternately added and subtracted. If a carbon 90" purging pulse is applied at (ii), to remove out-of-phase signals, then phase cycling must be employed. The variation of signal enhancements with 7 follows that described for SHECOR, and the same care is required for complex proton spectra. No direct comparison of the two sequences has been reported. However, SEPT (and therefore presumably SHECOR) has been reported to suffer some of the disadvantages of INEPT, namely error signals arising from variations in J . In an attempt to overcome this, Sequence 5C, known as SDEPT (Selective DEPT), has been 45 although it requires an extensive phase-cycling procedure if optimum results are to be attained. Signal intensities display the same dependence on 8 as in the DEPT

sequence and show maxima where the frequencies of the selective pulses coincide with carbon- 13 satellite. However, the signal is not inverted if the selective pulse is transferred from one satellite to the other, and a value of 0 of 135" will invert CH, signals with respect to CH and to CH, signals in a series of experiments in which the frequency of the selective pulse is increased in uniform increments over a region of the proton spectrum. One possible experimental drawback to Sequences 5B and 5C is the need for very rapid switching of the power level of the decoupler from the selective pulse to the first nonselective pulse, and the spectrometer may need to be modified to enable this. In situations where the positions of the proton resonances and corresponding satellites can be clearly identified it is possible to use a technique (Sequence 5D) in which the . ~ corresponding carbon signals can be s u p p r e ~ s e dA~refocussed INEPT sequence is preceded by a modified SPT sequence, made up of a selective proton 180" pulse sandwiched between two normal carbon- 13 90" pulses. This effectively equalizes the populations of the four levels of the selected CH system and thus no signals are possible for this carbon. The magnetization remains at equilibrium for all other protons and thus resonances for all other carbons are observed. In this way it has been possible to correlate the six resonances of methyl protons and the corresponding carbon resonances in the cyclic depsipeptide v a l i n ~ m y c i n , and the method has also been successfully ~~ applied to the proton and nitrogen resonances of the NH groups in the same m01ecule.~~ Although no DEPT version of this method has been reported, there appears to be no reason why this should not be a viable method.
5.1.2 Sequences Employing Only Non-selective Proton Pulses Since the use of non-selective proton pulses in a standard DEPT or refocussed INEPT sequence leads to equivalent enhancements for all carbon resonances, regardless of the frequency of the proton pulse, these sequences must be modified in some way, ideally to provide an enhancement only for the carbon resonance corresponding to the proton resonance that coincides with the frequency of the proton pulse. If this ideal is not attainable, and more than one carbon resonance is obtained, then hopefully the intensity of the carbon might be related in some simple way to the separation of the corresponding proton resonance frequencies and the proton pulse frequencies. This latter situation is considered first since the method does not require phase cycling of the decoupler. Variation of the

View Online

116

NATURAL PRODUCT REPORTS, 1988

h0
3

Wuence 5E SINEPT-2 (T+ 7 = I /24.

'H

EB

7:

= 1.8ms

I
C

v = 1.336

Downloaded by Instituto Gulbenkian Ciencia (IGC) on 14 July 2011 Published on 01 January 1988 on http://pubs.rsc.org | doi:10.1039/NP9880500101

Sequence 5F CHORTLE (T+7 = 1/24; extensive phase cycling is required.


S

intensity of carbon resonances with the frequency of the proton pulse has been briefly considered above in the SINEPT sequence (Section 3.2), where signal intensities of zero for carbon are obtained for corresponding proton resonances which lie at intervals of 'JCH from the frequencies of proton pulses. Sequence 5E, which is known4' as SINEPT-2, is more flexible in that positions yielding zero or maximum intensity can be placed at any desired separation from the frequency of the proton pulse by adjusting the period 7 . Sequence 5E is a refocussed INEPT sequence in which the first pair of 180" pulses is placed offcentre in the (24-l period during which polarization transfer occurs. Assuming a fixed value for A and that all pulses have the same phase, then the intensities of carbon signals vary according to I,, sin(360vT0), where I,, is the maximum INEPT intensity and v is the separation of the corresponding proton resonance from the pulse frequency. The delay d is selected as in the refocussed INEPT sequence. Carbon signals of zero intensity arise where corresponding proton resonances lie at the frequency of the transmitter and at intervals of (27)-l Hz on either side of that frequency. Other carbon resonances appear with positive or negative intensities up to the maximum, according to the position of their corresponding protons. By superimposing a sine curve of appropriate wavelength (i.e. 1/ 7 ) on the proton spectrum it is frequently possible to select suitable values for the frequency of the proton pulse and for the delay 7 to allow correlation of proton and carbon- 13 resonances by qualitative inspection of the spectra, provided the proton spectrum is well resolved and not too complex. A qualitative approach is unlikely to be satisfactory for many proton-carbon correlations, and a more rigorous method has been Sequence 5F, known as CHORTLE (Carbon-Hydrogen correlation from One-dimensional polaRization-Transfer spectra by LEast-squares analysis), differs from SINEPT-2 primarily in the extensive cycling of all of the transmitter and receiver phases, so that any errors that have been caused by phase or pulse imperfections are removed and peak intensities can be used reliably. Additionally, the last proton 180" pulse is replaced by gating the decoupler and half of the accumulations are carried out with the period 7 placed at the start of the initial (24-' period. Two spectra are obtained for each value of 7 . Where the two proton 90" pulses have the same phase (or differ by 180") the intensities show the sine dependence as in SINEPT-2, and the spectrum is referred to as the 'sine spectrum'. If, however, the phases of the proton pulses differ by 90" (or by 270"), the intensities show a cosine dependence and the spectrum is referred to as a 'cosine spectrum'. Thus, in its simplest form, 'sine' and 'cosine' spectra are obtained for one value of 7 (e.g. the inverse of the spectral width) and the intensity ratio !JIG is determined for each carbon resonance. The frequency Y of the corresponding proton resonance relative to the pulse frequency is given by lJIC tan (360~7") v = [tan-' (1,/1,)]/360~". If the carbon is = or bonded to two non-isochronous nuclei, the frequency v is the

T = 1.2ms v = 1.33 6

(f

Y
I I

I.

T = 1.8ms

50

40

30

20

10

P.P.m

Figure 19 Carbon- 13 n.m.r. spectra of camphor at 90 MHz. Spectrum (a) is the proton-broad-band-decoupledspectrum, with assignments : (b)-(g) were obtained by using the CHORTLE pulse sequence. Locations of proton pulses, delays 7,and spectrum types ['sine' (s) or 'cosine' (c)] are as indicated.

Sequence 5G Direct correlation.

average of that for the two protons. It is clearly more desirable to obtain data with several different values of 7 and/or frequencies of the proton pulse to provide a check. Additionally, non-linear least-squares analysis of the intensities will yield chemical shifts of protons to within a few hertz. This degree of precision, however, requires that any systematic errors should be minimized and entails holding the pre-excitation delay, (24-', and d constant within a series of spectra together with extensive phase cycling. This procedure allows the positions of both protons in a non-equivalent pair to be located. Good results can be obtained from four pairs of spectra, such as those shown for camphor in Figure 19. Spectra that show only a single carbon resonance, correlating with a proton resonance that is located at the frequency of the proton pulse, may be obtainedg9 with Sequence 5G. This sequence also resembles SINEPT except that the first delay period t is varied during the course of the experiment and the phase of the second proton 90" pulse differs from that of the first by 90". The extent of polarization transfer depends on both the value of 'JcHand cos(360vt0), where v is the separation

View Online

NATURAL PRODUCT REPORTS, 1988-1.

H. SADLER

I17

3-exo - H

The sequence in which the polarization- transfer period is varied provides the clearest visual correlations, as a series of essentially single-line spectra. However, a knowledge of the positions of the proton resonances is necessary. 5.2 Long-Range Correlations 5.2.1 Correlation through Coupling Two- or three-bond proton-carbon correlations can frequently be obtained from carbon-13 n.m.r. spectra by selective irradiation of specific proton resonances at low power, since the coupling constants are small (3-1 5 Hz). This method is regularly used for the assignment of resonances of quaternary carbon nuclei and is also valuable,50though less frequently used, for the assignment of resonances of protonated carbons. The same technique can obviously be applied for nitrogen resonances in polyaza-compounds and for silicon resonances in multisilyl ethers. However, in carbon spectra, the resonances of some non-quaternary carbons may show marked changes in intensity and in multiplicity if one of the corresponding carbon-13 satellites in the proton spectrum lies near a proton that is being irradiated. This problem can be markedly attenuated by employing51 proton broad-band irradiation during the pre-excitation delay and by switching to single-frequency decoupling during excitation and acquisition of the FID signal. The principal drawback of the experiment for quaternary carbon nuclei is its greater duration, necessitated by the comparatively long relaxation times of carbon nuclei. The situation is even worse for nitrogen- 15 and silicon-29 nuclei, where incomplete nuclear Overhauser effects can result in there being no signals. A marked improvement should be obtained in many cases by employing a modified refocussed INEPT sequence [Sequence SH, known as INAPT (Insensitive Nuclei Assigned by Polarization Transfer)] in which all of the proton pulses are selective and are applied to a specific proton resonance, and thereby52 using long-range couplings for polarization transfer. Since these couplings are much smaller than one-bond couplings, inter-pulse delays are significantly longer and some relaxation may occur while the pulse sequence is being applied. Thus signal enhancements of two- or threefold are observed for carbon- 13 resonances rather than the expected four-fold. Since the lengths of proton pulses are now relatively long (10-15 ms), this must be taken into account when calculating the delays A , and A,, for which values are otherwise set as discussed previously. Since the strength of the selective proton pulses (expressed in hertz) is much smaller than the one-bond coupling constant, no signal is obtained from the directly bonded carbon- 13 nucleus. If the chosen proton resonance coincides with a carbon-13 side-band of another then a small additional signal may be observed. The spectrum thus only shows resonances from carbon- 13 nuclei which have long-range couplings [of ca (2A)-'] with the resonance at the frequency of the proton pulse. The wide variations in J mean that the delay A cannot be optimized for all likely values of J . In practice this delay is rarely set longer than 50 ms, to avoid too much signal being lost by relaxation. The main restriction is that proton resonances which are coupled to that selected must be greater than 30 Hz away, otherwise multiple-quantum effects will result in further losses of signal intensity. Protons that are close to but which are not coupled to the chosen proton will not affect the polarization transfer. The method has also been used for the observation and assignment of nitrogen- 15 resonances in small pep tide^.^^

(g)

4-H

(f)

5-exo-H

Downloaded by Instituto Gulbenkian Ciencia (IGC) on 14 July 2011 Published on 01 January 1988 on http://pubs.rsc.org | doi:10.1039/NP9880500101

(d 1
1

anti-Me - H

(C
1

1-Me-H

(b)
A

syn - M - H e

50

40

30
P P.m

20

10

Figure 20 Carbon- 13 n.m.r. spectra of camphor at 90 MHz. Spectrum (a) is the proton-broad-band-decoupled spectrum; spectra (b)--(h) were obtained by using the variable-delay sequence 5G and each show one principal carbon resonance. Corresponding locations of proton pulses are as indicated.

between the resonance frequency of the proton that is responsible for polarization transfer and the frequency of the proton pulse. The sequence requires that the delay t be varied from one scan to the next in such a way that, over a large number of scans, polarization transfer is maximum for protons that are located at the pulse frequency and zero for those elsewhere. It has been shown that this is effected by increasing the value o f t in regular increments or by randomly varying the value of t between K / J and ( K + l)/J, where K is an integer. The selectivity increases with increasing K , but more artefacts are generated and the experiment becomes more susceptible to pulse and phase errors. In practice, these problems are overcome by using equal numbers of scans for values of K equal to 0, 1, 2, and 3 and subtracting the FIDs for the even values of K from those for the odd values of K . In each set of scans the value of t is incremented or chosen at random between the limits defined by K . This gives a proton selectivity of ca20 Hz. Protoncorrelated carbon- 13 spectra of camphor that have been obtained in this way are shown in Figure 20. Of these three sequences in which only non-selective pulses are employed, SINEPT-2 is the only one which may be used on spectrometers where phase shifting in the decoupling (proton) channel is not possible. However, proton pulse frequencies must be selected with great care to obtain unambiguous results. The CHORTLE sequence has the advantages that relatively few proton irradiation positions are necessary for a complete correlation and it does not require knowledge of the proton spectrum. However, extensive phase-shifting is necessary if reliable quantitative intensity measurements are to be obtained.

Sequence 5 INAPT; all proton pulses are selective. H

View Online

118

NATURAL PRODUCT REPORTS, 1988

In principle, an analogous selective DEPT is also possible. However, an additional pulse interval of (2.9- increases the opportunity for relaxation, and therefore signals of lower intensity are likely to be obtained.

5.2.2 Correlation through Space : Difference Spectroscopy A method for correlating the resonances of protons and carbon-13 nuclei that are not bonded to each other and which is just beginning to receive the attention it deserves makes use of the heteronuclear Overhauser effjxt. Briefly, if a proton is sufficiently close (usually within 3 A) to a nearby nucleus A then irradiation and saturation of the proton resonance leads to an increase in intensity of the resonance of A. The ratio (0) of the intensities of the resonance of A with and without irradiation of the proton resonance is defined as the nuclear Overhauser enhancement, NOE. For molecules of molecular weight less than about 1000 this enhancement does not exceed 1 +(-yH/2yA).Thus carbon intensities may be increased by any figure up to a factor of 3, and this is frequently found for proton-bearing carbons in a broad- band proton-decoupled carbon- 13 n.m.r. spectrum. Selective low-power (e.g. yB,/2n of ca 2-3 Hz) irradiation of a proton generates nuclear Overhauser enhancements on the resonances of only a few carbon nuclei that are two (or sometimes more) bonds away. Combining this technique with difference spectroscopy5* provides a powerful alternative to selective proton decoupling or selective polarization transfer for identifying or for assigning resonances, particularly (but not only) of quaternary carbons. In difference spectroscopy a control spectrum, which has been obtained with the irradiation point located in a blank region of the proton spectrum, is subtracted from a spectrum where a specific proton has been irradiated, so that only the effects of irradiation are visible in the difference spectrum and unaffected signals are missing. Apart from the position of irradiation, all experimental parameters must be identical for both spectra. The data-handling systems of most modern spectrometers are capable of performing subtractions of this kind, and the increases in intensity (which are usually quoted as a percentage of the intensities in the control spectrum) may be measured accurately. The earlier procedure of direct integration of the individual spectra is too unreliable to give accurate values, whereas difference spectra can (in favourable cases) detect Several procedures have increases in intensity of less than 1 YO. been used for these experiments; of t h e ~ e ,that described ~~,~~ below is probably the most satisfactory. A series of FIDs is obtained, one for each position of irradiation and one for the control. Selective low-power irradiation of one proton is carried out during a long (10-20 seconds) pre-excitation delay, as shown in Sequence 51, and the carbon-13 spectrum is obtained with broad-band proton decoupling at the usual power levels. The long delay is to ensure a maximum selective NOE for the nucleus that is being studied and minimum NOE at other sites which might otherwise arise from the broad-band decoupling. To average out variations due to spectrometer instability or to minor fluctuations in temperature during a long accumulation, it is preferable to use a small number of scans (e.g. 32 or 64) for each FID and to cycle through all of the irradiation points several times, co-adding the corresponding decays. It is usually worthwhile setting up all likely irradiation points in a single session, since observing a number of no-result irradiations usually wastes less time than several extra runs, each with its control spectrum. The occasional no-

Downloaded by Instituto Gulbenkian Ciencia (IGC) on 14 July 2011 Published on 01 January 1988 on http://pubs.rsc.org | doi:10.1039/NP9880500101

result irradiation is always comforting in that it confirms the reliability of the method. Difference spectra may be obtained either by subtracting the control FID from each of the others in turn and transforming the difference FIDs or by transforming all of the FIDs and subtracting the control spectrum from each of the other spectra. All transforms should be carried out on an absolute scaling basis. Enhancements are obtained by direct integration of the difference spectra and of the control spectrum. Some spectra of this type are shown in Figure 21 that have been obtained for andibenin B by irradiating the previously assigned resonances of methyl protons. The level of power that is selected for the irradiation ensures that directly bonded nuclei are not affected. In each case the quaternary carbon is clearly identified in the difference spectrum. Smaller responses are also seen from C-4 and C-5 if the 5- and 4a-methyl protons, respectively, are irradiated, due to some power leaking into one proton resonance whilst the other is being irradiated, since these resonances are only 12 Hz apart. (The proton spectrum is shown in Figure 28). Where an irradiation position also corresponds with (or lies close to) one of the one-bond carbon-I3 satellites of another proton resonance, a strong response of the corresponding carbon is also obtained, due to selective-population-transfer effects. Thus irradiation of the 10-methyl protons [see Figure 21(c)] also gives responses from the 3-, the 5-, and the 4a-methyl carbons, and responses from the 10-methyl carbon are obtained when the 3- and 5-methyl protons are irradiated [Figure 21 (b) and (d)]. A response from the 4a-methyl carbon is also obtained (c)
0

10
I

3-Me r

10-Me

(b)

80

60

40

20

s e l e c t i v e c o n t i n u o u s irradiation

BB

Ir

Sequence 51 Heteronuclear NOE measurement.

p.p.m. Figure 21 Carbon-13 n.m.r. spectra of andibenin B at 90 MHz. Spectrum (a) is the proton-broad-band-decoupledspectrum; spectra (b)--(f) are NOE difference spectra with irradiation at the proton resonances corresponding to (b) 3-methyl, (c) 10-methyl, (d) 5methyl, (e) 4a-methyl, and (f) 4P-methyl groups. Conditions : 256 pulses, 30 pulse, pre-excitation irradiation period 10 s, acquisition time 0.4 s.

View Online

NATURAL PRODUCT REPORTS. 1988-1.

H. SADLER
A

119

Downloaded by Instituto Gulbenkian Ciencia (IGC) on 14 July 2011 Published on 01 January 1988 on http://pubs.rsc.org | doi:10.1039/NP9880500101

as the 7P-proton multiplet lies partly under the 10-methyl proton resonance. It is important that the irradiation point for the control spectrum is selected carefully since, if this coincides with a one-bond carbon-13 satellite, a negative signal is seen for the corresponding carbon resonance in the difference spectrum. Negative signals may also arise57where a carbon nucleus and an irradiated proton are not sufficiently close to give a direct enhancement of the carbon signal but are both sufficiently close to a second proton for the effect of irradiation to be relayed. Where a particular carbon is thought to be relaxed by two (or more) protons that have different chemical shifts, stronger enhancements may sometimes be by rapidly cycling the irradiating frequency through all of the line positions of all resonances during the pre-excitation delay. This also allows lower irradiation powers to be used, thereby improving selectivity. The use of difference spectroscopy in NOE and other doubleresonance experiments has been thoroughly and carefully reviewed,56and the reader is referred to that source for a much more detailed consideration.

Sequence 6A INADEQUATE : (i) double-quantum excitation of pairs of carbon- 13 nuclei and population inversion of isolated carbon- 13; (ii) signal generation. Broad-band proton decoupling is applied 5 throughout and q and @ are as in Figure 22.
Read-Pulse Phase

Detector Phase

Transformed Signal

6 Methods for the Direct Correlation of Carbon-13 Resonances


Important information concerning the structure and the configuration and conformation of the carbon framework of molecules is provided by the carbon- 13 homonuclear spincoupling Where one-bond couplings between pairs of adjacent carbon atoms are sufficiently different it should, in principle, be possible to deduce carbon-carbon connectivities and to map out the carbon skeleton by measuring and comparing these couplings. The geometrical dependence of longer-range couplings is useful in studies of molecular conformation. The experimental measurement of these couplings presents a number of difficulties. The low natural abundance of carbon- 13 nuclei means that relatively very few molecules possess a pair of carbon-13 nuclei. This is advantageous in that it gives a relatively simple AX or AB pair of doublets, but significant quantities of material are required if the experiment is not to last an unrealistic time. Even if quantity is not a problem, a further difficulty arises with natural-abundance samples. At the centre of each doublet lies a signal that is two-hundred times as strong, from molecules that contain only a single carbon-13 nucleus. This signal may completely cover any long-range coupled doublets, and even one-bond doublets may not be easy to identify, due to the presence of small spinning side-bands or of odd lines that are due to incomplete or inhomogeneous proton decoupling or to the presence of minor impurities. Additionally, couplings to quaternary carbons are frequently similar in magnitude, and the four overlapping doublet signals cannot be readily resolved. Fortunately, methods are now available to alleviate or to overcome these problems and are examined below.

-X

-Y

Figure 22 The effect of cycling the phases of the read pulse and detector for four successive scans, using the INADEQUATE pulse sequence (Sequence 6A). Transformation of the FID corresponding to each combination of phases would give the spectra shown, resulting in cancellation of the large central signal from the uncoupled nucleus. To be effective, very precise and stable phase-shifting is required.

as the generation of a double-quantum coherence,62since it does not produce any observable magnetization. However, the AX system has been perturbed in such a way that if a 90" pulse (usually called the ' read ' pulse) is applied immediately after the sequence it produces a signal for each doublet in which the lines are in antiphase. This four-pulse sequence also produces a dispersion signal for the non-coupled nuclei. The relative 6.1 The Suppression of Strong Signals from Uncoupled phases of the single-quantum dispersion signal and the doubleCarbon-13 Nuclei quantum-derived antiphase doublets depend on the relative phases of the double-quantum excitation pulses, the read pulse, Signals from uncoupled carbon- 13 nuclei may be suppressed by and the detector. In particular, alteration of the phase of the the INADEQUATE60 (Incredible Natural-Abundance read pulse, so that it differs by 90" from the double-quantumDoublE-QUAnTum Excitation) sequence (Sequence 6A). This excitation 90" pulses, causes a change of phase of 90" of the makes use of the special phase properties6' of signals that are signal from the isolated nucleus but a change of 270" in the derived by double-quantum excitation of an AX system such as two carbon- 13 nuclei that are coupled only to each other. The phase of the double-quantum-derived doublets. If, therefore, results that are obtained by using the sequence cannot be the phase (q5) of the read pulse is incremented by 90" and the phase ($) of the detector is incremented by 270" after each scan, explained in terms of simple vector diagrams of the type that whilst the phases of the double-quantum-excitation 90" pulses were used earlier in this review, and the following discussion is remain constant, the phase behaviour of the generated signals purely descriptive. If both nuclei of an AX system are aligned a in the same results in the cancellation of the single-quantum signal and molecule they may be simultaneously excited to the state the accumulation of the double-quantum-derived antiphase (and vice versa) by using the sequence 90~-~-180~-~-90~, doublets. The effect of this phase-cycling procedure is set out in Figure 22, and in principle this should result in a clean where 7 = (2n+ 1)/4JAx. This process is generally referred to

View Online

120

NATURAL PRODUCT REPORTS, 1988

Downloaded by Instituto Gulbenkian Ciencia (IGC) on 14 July 2011 Published on 01 January 1988 on http://pubs.rsc.org | doi:10.1039/NP9880500101

70

60

50
p.p.m.

40

30

20

Figure 23 INADEQUATE carbon-13 n.m.r. spectra of menthol (0.3 ml) C6D, (0.1 ml) at 50 MHz, obtained using Sequence 6A. Individual multiplets are expanded five-fold horizontally, showing the variations in Jcc. The intensities are approximately those expected on the basis of the number of adjacent pairs of carbon- 13 nuclei. Conditions : 8000 scans, pre-excitation delay is 7 s.

spectrum of the coupled nuclei only. Broad-band proton decoupling is applied throughout the entire sequence. In practice, the last two 90" pulses are separated by a very short delay ( A ) of a few microseconds; this delay is necessary for the resetting of the phase of the final read pulse. Complete suppression of the strong singlets, however, requires that highly homogeneous and precisely timed pulses are used and also requires precise and stable phase-shifting networks. Since these requirements are rarely completely satisfied, a residual central signal is often seen. This can sometimes be reduced to an acceptable level by additional phase cycling. Repetition of the four-step cycle, with the phase of the 180" pulse being changed by 180", and repeating this eight-step cycle four times, with all phases being incremented by 90", gives a 32-step cycle. More complex 64-, 128-, and 256-step cycles have also been It has also been shown that the effects that are caused by pulse imperfections can be reduced by replacing the separate pulses of the sequence by composite A composite pulse is a group of two or more differently timed and phased pulses which follow each other immediately, with no delay. The setting of the delay (7) not critical when n = 0, and deviations is in J of up to 40% from the value that is used to calculate 7 (= 1 / 4 4 result in intensities not less than 7 5 % of the optimum. Thus a wide range of one-bond couplings (ca & 3 60 Hz) can be observed by using a setting of T = 6 ms. An INADEQUATE spectrum of menthol (Figure 23) was obtained in this way by using a 32-step phase cycle. Both direct and longer-range couplings may be observed in the same spectrum, by choosing 7 II 50 ms, which corresponds to J = 3-7 Hz for n = 0 and a value of J of ca 35 Hz for n = 3. Where n is odd, the doublets appear to be 180" out-of-phase with those for which n is even or zero. Setting 7 = (2n+ 1)/4J is satisfactory for pairs of carbon nuclei where the frequency separation Av exceeds the coupling constant J by a factor of three or greater. At closer separations this expression does not hold, and outer lines are frequently obscured in the background noise since the spectra are of the AB rather than of the AX type. For strong where A v / J lies between 1.0 and 2.5, it is usually more advantageous to set 7 = 3/4J or 5/4J.

Sequence 6B Refocussed INADEQUATE ; broad-band decoupling is applied throughout; # and @ are as in Figure 22.

If desired, the signals from carbon- 13 pairs may be obtained in-phase by adding a refocussing period, with a 180" pulse at its centre, to give the refocussed INADEQUATE sequence (Sequence 6B). In many cases (particularly in saturated systems) the one-bond proton-carbon couplings are approximately four times the one-bond carbon-carbon couplings. This allows a combination of the INADEQUATE and APT sequences to be used. If, for example, the decoupler is gated off during the second half (1/4J) of the refocussing period, spectra are obtained66in which signals from methyl and methine carbons are inverted with respect to those from methylene and quaternary carbons. Where carbon-carbon couplings significantly exceed this level, further modification of the sequence is required. Members of a more complex group of sequences, known as SEMINA-1 and SEMINA-2 [based on the amalgamation of the SEMUT (see Section 3) and INADEQUATE sequences], provide6' information on the total' proton multiplicity for coupled pairs of carbon- 13 nuclei.
6.2 Sensitivity Enhancement and Spectral Simplification

The sensitivity of the INADEQUATE experiment can be markedly improved by combining the sequence with a protonpolarization- transfer technique such as INEPT or DEPT (Section 3.2). Direct combination of pulse sequences usually results in some redundant pulses, and so some simplification is frequently possible. Sequence 6C is such a simplification, resulting from combining68INADEQUATE with INEPT. To restrict the observable signals to those obtained only by

View Online

NATURAL PRODUCT REPORTS, 1988-1.

H. SADLER
33

121

INADEQUATE experiment with 6' = 90" or an equivalent INEPT-INADEQUATE experiment will reveal signals from CH groups and all of those carbon nuclei that are bonded to them. This provides a useful simplification of spectra and aids the assignments in a complete spectrum. 6.3 Selective Carbon-Carbon Correlation It is frequently not possible to deduce sufficient correlations from an INADEQUATE spectrum to map out the entire carbon skeleton, due to coincidences of coupling constants. The two-dimensional v e r ~ i o 71 ~ ~ ~ in principle, separate all n does, pairs of bonded (i.e. one-bond-coupled) pairs of carbon atoms. However, in complex cases, overlapping signals can make interpretation very difficult. The experiment also requires large amounts of both the compound and storage space for data. In many cases only a few correlations are required above those discernible from the coupling constants. A high degree of selectivity may be obtained, however, by one-dimensional methods which make use of the principle on which the two-dimensional experiment is based. Variation of the period A between the last two 90" pulses of the INADEQUATE sequence results in a variation of the intensity I of the signals from the pairs of carbon-13 nuclei according to cos the expression I = sin (360J~)O (36Ovd)", where v is the double-quantum frequency (which is equal to the sum of the resonance frequencies of the carbon-13 nuclei; both are measured with respect to the pulse frequency). From this expression it follows that (a) the signal intensity is independent of A if u = 0, i.e. if the pulse is placed midway between a pair of coupled resonances, and (b) zero signal is obtained if A = (4u)-l, i.e. where the pulse is placed (8A)-' Hz from the centre of a pair of coupled resonances. These conclusions may be used to confirm (or otherwise) that a pair of resonances arises from a bonded pair of carbon-13 nuclei. In the DOUBTFUL (DOUBle quantum Transition to Find Unresolved Lines) sequence (Sequence 6E), which was originally proposed for proton the time is varied either randomly or by using regular increments throughout the total period during which data are being acquired. With the pulse placed at the centre of a coupled pair of carbon- 13 resonance^,^^ the signal intensities of these nuclei do not vary with A and will accumulate normally. However, a pair of carbon- 13 nuclei whose centre lies away from the pulse has a non-zero doublequantum frequency. Their intensities will show co-sinusoidal variation with A , and co-addition of transients that have been obtained with different value of A will tend to cancel these signals. It can be that pairs of carbon-13 nuclei whose centres lie two further than u' from the pulse will not be observed if the longest delay (A,) is set such that A , = l/u'. Some DOUBTFUL spectra of andibenin B are shown in Figure 24. Even with the poor suppression of the central line, the following effects are seen: no double-quantum signals are obtained in spectrum (a), where the pulse does not coincide with the centre of any coupled pair of resonances; the doublequantum signals appear for the pairs C(l')-C(2') and C(4kC(5) in (b) and (c) if the pulses are appropriately placed and a value of A , of 2.24 ms is used; in (d) the signals for the C(9W(lO) pair as well as for C ( 4 w ( 5 ) are obtained for the pulse that is placed at the centre of C(4)-C(5) but a shorter delay of A , = 0.75 ms is used. Spectrum (d) shows that choosing a A , that is rather less than the exclusion requirement above can result in two or more pairs of carbon-13 nuclei being detected. In the example shown there is no ambiguity. However, it sometimes happens that the high-frequency and low-frequency components of two pairs are too close for the pairs to be distinguished by this technique. In such cases it is a p ~ r o p r i a t e ~ ~ to use a single fixed value of A . This requires a preliminary assignment of the resonances to be made, the pulse being placed at the centre of one pair and set so that A = (8v')-' (where v is the separation of the centres of the pairs of ' resonances). If the assignment is correct the signals from the

Sequence 6C INEPT INADEQUATE : (i) pre-saturation of carbon13 nuclei; (ii) INEPT; 7 = 1/4JcH; (iii) INADEQUATE; 72 = ' 1/4J,,; $ and $ are as in Figure 22.

'H

BB

Downloaded by Instituto Gulbenkian Ciencia (IGC) on 14 July 2011 Published on 01 January 1988 on http://pubs.rsc.org | doi:10.1039/NP9880500101

Sequence 6D DEPT INADEQUATE : (i) pre-saturation of carbon- 13 nuclei; (ii) DEPT; 7' = 1/2JcH; (iii) INADEQUATE; 72 = (I /3, - ; # and are as in Figure 22. 4,)

Sequence 6E DOUBTFUL ; broad-band proton decoupling is applied throughout; q5 and are as in Figure 22.
~

polarization transfer, the carbon- 13 nuclei are pre-saturated by a long sequence (ca 60) of 90" pulses at intervals of 0.1 to 0.2 seconds. The INEPT sequence transfers proton polarization to the carbon- 13 nuclei, at which point the INADEQUATE sequence follows. Since the carbon signals that are generated at this point consist of two anti-phase vectors, corresponding to the a and the p state of the bonded protons, a period A' must elapse (to allow the signal vectors to precess until they come into alignment) before broad-band proton decoupling is applied. As in the INEPT experiment, the value taken by A' determines the signal intensity; for example, a compromise setting of A' = 0.3/Jc, will yield all signals. More transients may be collected in a given time than in the basic INADEQUATE experiment since the delay between successive scans is governed by the faster relaxation of the protons. In practice, an improvement in sensitivity of between two- and three-fold is obtained, which represents a saving of time by a factor of between 4 and 9. Essentially similar results are by Sequence 6D, which uses the DEPT sequence to generate the polarization transfer; here the signal intensities are governed by the value that is given to the 8 pulse. As with all INADEQUATE-based experiments, extensive phase-cycling schemes are required to suppress unwanted central signals. The extent of phase cycling that is necessary will depend on the particular spectrometer being used. Some models may require phase cycling only for the carbon pulses whereas others will require this for the proton pulses also. It has recently been r e c o g n i ~ e dthat both DEPT and ~~ INEPT polarization- transfer versions of INADEQUATE cause the proton polarization to be transferred also to a non-proton-bearing carbon- I3 nucleus that is coupled to the proton-bearing carbon- 13 nucleus. The basic four-step phasecycling scheme that is used in INADEQUATE places polarization alternately on the protonated carbon- 13 nucleus and the non-protonated carbon- 13 nucleus in a H-13C-13C fragment, and this is also true for the more extensive phase-cycling procedures. A further consequence is that a DEPT-

View Online

122

NATURAL PRODUCT REPORTS, 1988

@
0

Downloaded by Instituto Gulbenkian Ciencia (IGC) on 14 July 2011 Published on 01 January 1988 on http://pubs.rsc.org | doi:10.1039/NP9880500101

I
10
4

1'

v]' I, where J is the homonuclear coupling constant and vo ,and v, are the frequencies of the pulse and the centre of the A doublet respectively. The 90" pulse of the spin-echo sequence restores the population for one A line, inverts the population for the other, and also excites the signal from the X pair and those from all other carbon-13 nuclei. However, all signals except those from the AX system are cancelled if the phase of the 90" pulse is alternated on successive scans and if this is combined with alternation of both the detector phase (between y and - y ) and the value of the spin-echo delay (between 1/4J and 3 / 4 4 every other scan, as shown in Sequence 6F. The sequence is accompanied throughout by proton broad-band decoupling. Doublets resulting from polarization transfer appear as in-phase signals and the initially excited resonance appears as a singlet. Since the excitation pulse is not placed directly at the resonance position of the carbon from which polarization is to be transferred, and resonances lying at vof n/t' Hz are semi-selectively excited, extra care must be taken when setting up the experiment. Typical conditions are a DANTE train of 42 pulses, of flip angle 3", at intervals of 0.0042 seconds, which would site the pulse 2400 Hz from the resonance that is to be semi-selectively excited. A series of experiments could be carried out by altering the frequency of the excitation pulse. Since, on most spectrometers, it is not possible to change the power of the pulses between the DANTE pulses and the spin-echo pulses, some fixed attenuation may therefore be necessary so that the DANTE pulses are not too short and yet the 90" pulses and the 180" pulses are not excessively long. Small unwanted signals show up in some spectra, particularly if too short a pre-excitation delay is used or if significant relaxation occurs during the spin-echo sequence.

90

80

70

60

50

40

P.P.m

7 The Location and Correlation of Proton Resonances


Almost since routine proton n.m.r. spectrometers became generally available, spin decoupling has been the most frequently employed method for correlating proton resonances. Measurements of nuclear Overhauser enhancements were much less used, due to the difficulties in measuring peak intensities sufficiently accurately. For satisfactory results from both types of experiment, the irradiated and affected resonances must be sufficiently well resolved from nearby resonances. For molecules whose spectra are not amenable to spin decoupling (as well as for many that are) the two-dimensional COSY (Correlation SpectroscopY)7 7 has more recently often been used successfully, although interpretation of spectra is not always as straightforward as might be hoped. The twodimensional NOE experiment, NOESY,78 is generally only applicable to biological macromolecules, for which the correlation times are long, and is usually unsatisfactory for smaller molecules, which show positive homonuclear Overhauser enhancements. The range of applications of homonuclear double-resonance techniques has, however, been considerably extended by combining these with difference spectroscopy. These methods will be discussed below. This is followed by a short discussion of how spectra may be simplified by using multiple-quantum-excitation methods.

Figure 24 Carbon-13 n.m.r. spectra of andibenin B (500 mg in 2 ml CDCl,) at 90 MHz. Spectra (a)---(d) are DOUBTFUL carbon-13 spectra, and were obtained by using Sequence 6E (with d being incremented by 15 ,us every 128 scans) carried out to assign C-2' and C-8 : in (a) the pulse was midway between C- 1' and C-8, as indicated by the arrow; in (b) the pulse was midway between C-I' and C-2'; in (c) and (d) the pulse was midway between C-4 and C-5. Spectra (a)--(c) were obtained with 150 increments and (d) with 50 increments. Note that the responses are clear in spite of the poor suppression of the centre line. Spectrum (e) is the proton-broadband-decoupled spectrum.
e

Sequence 6F DANTE + spin-echo : (i) DANTE; (ii) spin-echo : 4 = y, -y, y , and --y consecutively; 7 = 1/4J, 1/4J, 3/4J, and 3/4J respectively; @ = y, y, - y , and - y respectively.

pair that is centred on the pulse will be obtained but the signals from the other pair will be absent. A completely different method76 for selective correlation of carbon nuclei does not employ double-quantum excitation but makes use of transfer of polarization from one of a pair of coupled carbon-13 nuclei to the other. Sequence 6F uses a DANTE sequence and the 90" pulse of a spin-echo sequence to inve.rt the populations of the energy levels corresponding to one line of the A pair in an AX system. The DANTE sequence arranges the spin vectors corresponding to the two lines from A in opposite directions along the x' axis of the rotating frame. This is known as semi-selective excitation. It requires that the number (m)of pulses for which the flip angle is 8 and their separation is t' be adjusted so that the following relationships hold: WI = [~O"/COS-'(COS~ 1 ; 0 = 180Jt'"; t' = I [v,0/2)]-

7.1 Homonuclear Double Resonance Difference Spectroscopy The principle of difference spectroscopy was described briefly above (Section 5.2.2) in connection with the measurement of heteronuclear Overhauser enhancements. Several different types of homonuclear double-resonance difference spectroscopy have been described, and these differ in the position and power of the second applied radiofrequency and whether this is applied before or during the acquisition of data or both. Since all of these experiments and their applications are the subject of an excellent comprehensive review56they will be discussed only briefly below, and for further information the reader is directed to that review.

View Online

NATURAL PRODUCT REPORTS, 1988-1.

H. SADLER

123

Downloaded by Instituto Gulbenkian Ciencia (IGC) on 14 July 2011 Published on 01 January 1988 on http://pubs.rsc.org | doi:10.1039/NP9880500101

L
I
1.0

1
3.0

1
2 -5

I
2.0 p. p. m.

I
1.5

Figure 25 Proton spectra of anditomin at 300 MHz. Spectrum (a) is the proton spectrum and (b) is the decoupling difference spectrum, obtained by irradiating the 7'-H multiplet at 6 = 3.0; the decoupler power was 9 dB below 0.2 W.

7.1. I Decoupling Diflerence Spectroscopy In these experiments a 'control' spectrum, which is obtained by applying radiation in a blank region of the spectrum, is subtracted from one in which the centre of a specific multiplet is irradiated, all other experimental parameters being identical. No pre-excitation delay is necessary and the irradiation is applied continuously throughout excitation and while data are being acquired. The power of the irradiation is normally less than that required for complete decoupling. This method is particularly suited for revealing multiplets that are ill-defined (or submerged in regions of the spectrum which are not clearly resolved) and are coupled to isolated multiplets which can be cleanly irradiated. The best results are obtained from spin systems which are essentially first-order. Figure 25 (a) shows a crowded region of aliphatic proton resonances in the 360 MHz spectrum of anditomin. The corresponding difference spectrum [Figure 25 (b)], which was obtained by decoupling the 7'-proton (8 = 3.0), clearly shows the 6-protons. The decoupled spectrum and the control spectrum show, respectively, as positive and negative signals. The main problem arises from the shift in position of every resonance in the spectrum that is caused by irradiating the sample during the acquisition of data. These shifts, known as Block-Siegert shift^,'^ may result in imperfect subtraction of the non-affected resonances. This is particularly noticeable for sharp singlets, which also appear in the difference spectrum as dispersion signals, as well as for the desired responses from the affected multiplets. The magnitude of the shifts increases with irradiation power but decreases with increasing separation of the irradiation position from the observed signals. The unwanted signals can be reduced by siting the control irradiation close to the irradiated multiplets, which will reduce the relative shifts of any particular resonance in the two spectra. Reduction of the irradiation power below that normally employed in a conventional decoupling experiment is also advantageous and acceptable, since a small change in the position or the width of a line that is not readily discernible in a normal spectrum shows up clearly in a difference spectrum. In some cases the use of very high digital resolution for the transformed spectra and a facility for

displacing one spectrum with respect to another before subtraction will also yield satisfactory results. Decoupling difference spectroscopy is not satisfactory where the irradiated multiplet is in a crowded region of the spectrum or is partly obscured. In these cases it is necessary to rely on populationtransfer experiments, effected by irradiation of a single line. 7.1.2 Selective Population Transfer Difference Spectroscopy The simplest procedure uses the SPT (Selective Population Transfer) experiment, in which irradiation is applied to a single line of a multiplet at lower power, to saturate one transition, and the power is gated off during the excitation pulse and the period when data are being acquired.s0 This irradiation of a single line causes changes in the populations of energy levels of the coupled nuclei, so that subtraction of a control spectrum shows positive and negative responses for the coupled nuclei that are reminiscent of a continuous-wave INDOR spectrum. The exact appearance of the responding multiplets depends on the relaxation times of the nuclei involved, the duration of the pre-excitation irradiation, and the flip angle of the excitation 82 The advantages of this technique over decoupling difference spectroscopy are (i) improved selectivity, due to the use of lower power levels, (ii) the absence of Block-Siegert shifts, since the power is not applied while data are being acquired, and (iii) only a single transition need be clearly visible, so that partly obscured resonances may be irradiated. Another application of this technique, which has receiveds3 very little use but which should not be ignored, makes use of the internal population transfer which occurs between the component lines of a multiplet. A long pre-irradiation period must be used. In this case the multiplet to which the irradiation line belongs appears as an inverted signal, with some distortion of intensities, in the difference spectrum. If one of the lines near the centre of the multiplet is saturated and a 90" excitation pulse is used then the intensity distortions may not be too great, and coupling constants may be measured on an otherwise inaccessible multiplet. Such simplifications are not possible if the irradiated multiplet is coupled to another which overlaps it. Application of this technique to some hidden and partly

View Online

I 24

NATURAL PRODUCT REPORTS, 1988


H

U
Downloaded by Instituto Gulbenkian Ciencia (IGC) on 14 July 2011 Published on 01 January 1988 on http://pubs.rsc.org | doi:10.1039/NP9880500101

- 2 '1.

2.0

1.8

1.6 p.p.rn.

1.4

1.2

(a)

Figure 26 The low-frequency region of the proton n.m.r. spectrum of andilesin A and the saturation difference spectra of multiplets (aHd), which were obtained by subtracting from a control spectrum ( i e . one in which the line that is indicated with an arrow was irradiated at 42 dB below 0.2 W during a pre-excitation delay of 10 s).

5.5

5.0

4.5 p.p.r . n

4.0

3.5

obscured multiplets in the proton spectrum of andilesin A is shown in Figure 26. Irradiation was applied at the transitions marked. Other techniques that are related to SPT difference spectroscopy are pseud0-1NDOR,~~ which the irradiation is in continued during excitation and data acquisition and which also suffers from the Block-Siegert effect, and SPI (Selective 84 Population Inversion) difference spectroscopy,s2~ in which the pre-excitation irradiation is replaced by a selective 180" pulse. The latter experiment is more sensitive than the SPT experiment but requires more care in execution. The SPT difference experiment has been shown to be very effectivess in separating out individual multiplets with true relative line intensities if an exact 90" excitation pulse is used and if a Gaussian envelope is superimposed on the selective-inversion pulse. However, very few spectrometers are at present equipped with pulse modulators for this purpose and they only supply approximately rectangular pulses.
7.1.3 Nuclear Overhauser Eflect Diflerence Spectroscopy Difference spectroscopy has raised the status of measurements of homonuclear Overhauser enhancements from a method which is unreliable, and therefore little used, to one of the most powerful methods for deducing spatial relationships between protons in the same molecule. The maximum increase in signal intensity which can be observed for a proton that is relaxed by a nearby proton whose energy levels are saturated is 50%. However, observed increases in intensity are frequently only a few percent and sometimes less than one percent. Such small increases can be reliably and reproducibly measured by using difference spectroscopy. The experiments6 is carried out essentially as described earlier (Section 5.2.2) for the heteronuclear version but with irradiation gated off during the excitation pulse and while data are being acquired. The duration of the pre-excitation irradiation should be one or two times the average longitudinal

Figure 27 The proton spectrum (a) and the proton NOE difference spectra [(b)--(d)] of methyl 1-thio-P-D-mannoside tetra-acetate. Spectrum (b) was obtained by irradiating at the centre of the 5-H multiplet, at 39 dB below 0.2 W, during a pre-excitation delay of 3 s; a 20" excitation pulse was used; 1-H and 3-H show clear enhancements whereas &Ha and 6-H, show SPT effects. To produce spectrum (c) the irradiation was cycled through the lines of the 5-H multiplet at 42 dB below 0.2 W during the pre-excitation delay of 8 s; a 90" excitation pulse was used; SPT effects are absent. The same irradiation as in (c) was used to produce (d) for the 2-H multiplet ;the spectrum shows clear enhancements for 1-H and 3-H and small dispersion-type signals where there are no NOE or SPT effects.

relaxation time, e.g. around 8 seconds for molecules of medium molecular weight (100--600 Da), and the data-acquisition period should be as short as is consistent with satisfactory digital resolution. In molecules of this size the contribution of dissolved oxygen to relaxation is not normally significant, and it is not necessary to degas solutions. As would be expected, it is preferable to irradiate methyl groups and observe methine protons rather than vice versa, although the latter can yield some very small NOES,~' and NOES between closely situated methyl groups are regularly observed.88 In the difference spectrum the irradiated resonance appears as a strong negative signal. It has been customary to irradiate multiplets at their centre. However, this often results in the appearance in the difference spectrum of positive-type responses of overall intensity zero for multiplets that are coupled but not spatially close to the irradiated multiplet, due to SPT effects. Multiplets that are spatially close but not coupled to the irradiated multiplet appear as truly positive signals, but multiplets which are both coupled and spatially close to the irradiated multiplet may show a response due to a combination of NOE and SPT effects. Frequently, if SPT effects are present, it is not clear whether any NOE is present, as shown in Figure 27 (b). Irradiation of the centre of the multiplet from the proton at C-5 of methyl 1-thio-P-D-mannoside tetra-acetate results in strong SPT signals for the protons at C-6 in the difference spectrum. These effects can be avoided by rapid cycling8' of the irradiation frequency through each of the lines of the chosen

View Online

NATURAL PRODUCT REPORTS, 1988-1. H. SADLER

125

Downloaded by Instituto Gulbenkian Ciencia (IGC) on 14 July 2011 Published on 01 January 1988 on http://pubs.rsc.org | doi:10.1039/NP9880500101

multiplet, in turn, during the pre-excitation period and by employing a 90" excitation pulse. When this procedure is employed, the SPT components cancel or are very much reduced and the true NOE responses are much clearer. This procedure also allows lower irradiation powers to be used, which improves selectivity. Similar results may be obtained by acquiring a separate FID for the irradiation of each line of the multiplet for the same number of scans and adding these together before subtracting a control spectrum of the same total number of scans.9oThe result of cycling through the eight resolved lines of the multiplet from the proton at C-5 in the same sugar derivative is shown in Figure 27(c). Positive NOE signals are now clearly visible for both of the protons at C-6 and also, unexpectedly, for the proton at C-4. A negative signal is observed for the proton at C-2, illustrating the three-spin e f f e ~ t * in# ~ l the effect of irradiating the proton at C-5 is ~ which relayed to the proton at C-2 by the proton at C-1 or that at (2-3. In the general case, if a nucleus B is sufficiently close to two nuclei (A and C) which do not relax each other, then irradiation of A may result in a negative signal for C as well as a positive signal for B. Where no NOE or SPT signals are evident, as for the protons at C-6 in Figure 27 (d), small dispersion signals will be evident where the subtraction is imperfect. Negative signals, apart from those described above, are sometimes obtained in other specific circumstances. If the resonance of a proton that is undergoing slow exchange with another is irradiated, as is possible with OH groups, then strong negative responses are found at both sites, due to saturation transfer. Positive signals will be those that would be obtained if both resonances were irradiated. Negative NOE's are also obtained where molecular motion is sufficiently slow, either due to molecular size or because a viscous solvent is employedg2to increase the molecular correlation times.

Sequence 7A Multiple-quantum excitation; the final pulse (----) is applied on alternate scans only; extensive phase cycling is required.

9'

6 '

14

+. .

. .

7.2 Simplification of Proton Spectra by Multiple-Quantum Filtration In many proton spectra, the high degree of complexity arises from the overlap of resonances from a number of separate spin systems, comprising two or, more frequently, a large number of coupled nuclei. It would therefore be valuable to be able to present the spectrum of each individual spin system separately, or at least to separate a complex spectrum into subspectra that show all the spin systems of a given number of coupled spins. Both of these objectives can be achieved to some extent by multiple-quantum excitation methods60-62and by using the special phase properties that are associated with the derived signals to filter out signals from unwanted spin systems. The INADEQUATE sequence (see Section 6) for obtaining spectra from coupled pairs of carbon-13 nuclei, free from the strong signals due to isolated uncoupled carbon-13 nuclei, is an example of a double-quantum filter experiment, and the refocussed version has been used to separate doublet signals from singlet signals in the aromatic region of the proton spectrum of simple polynucleotides.93 The DOUBTFUL sequence (also described above) has also been used successfully for i d e n t i f ~ i n g ' ~ , ~ ~ coupled pairs of protons in these and in other molecules. In these cases, all of the examples comprised AX or AB pairs where the doublet couplings in a particular spectrum were of similar sizes. In a more general situation, proton coupling constants vary over a sufficiently wide range that uniform double- or higher-order quantum excitation cannot be achieved for all spin systems by the simple or refocussed INADEQUATE sequences. In these, multiplequantum excitation is optimum where the coupling constant J = (47)-l. Intensities are lower (Iop, 90 Jr") for other values sin of J and the spectra frequently show strong phase distortions. Uniform excitation and the absence of phase distortion can be producedg5 by the modified refocussed INADEQUATE sequence (Sequence 7A). Here the pulse interval t is regularly incremented during the total number of scans and the last 90" pulse is applied on alternate scans. In a refocussed INADE-

4 p.p.m.

Figure 28 Proton n.m.r. spectra of andibenin B. (a) The normal proton spectrum. (b) The double-quantum-excitation spectrum, using Sequence 7A, showing the expected AX doublets and higher-order spin systems at reduced intensity: t was incremented by 20 ms from 10 to 500 ms. (c) The double-quantumfiltered spectrum, obtained by using Sequence 7B, showing AX doublets with all other signals almost eliminated; t was fixed at 21 ms, A was incremented by 20 ms from 20 to 400 ms. (d) The 'singlets only' spectrum, obtained using Sequence 7C; t was incremented by 10 ms from 10 to 200 ms.

QUATE experiment the intensities of doublet signals vary with sin290Jro,and thus variations in intensity average out over a wide range of values of t, leading to essentially uniform excitation. In a typical experiment t would be varied from 10 ms to 500 ms in steps of 20 ms, thereby covering a range of values of J from 1 to 50 Hz. If the simple INADEQUATE sequence is used, variation of r leads to vanishing signal intensities. The presence of the final 90" pulse removes anomalies in the phases of the signals. Singlet signals are suppressed by cycling the phases ($) of the first three pulses through the sequence x, y , - x , and -y whilst the detector phase is alternated between + x and -x. A double-quantumexcitation proton spectrum of andibenin B that has been obtained in this way is shown in Figure 28(b). Comparison with the standard spectrum [Figure 28 (a)] shows the absence of the methyl singlets and, in addition to the expected doublets, signals from higher-order spin systems are also obtained (but at a reduced intensity). These occur because double-quantum

View Online

126

NATURAL PRODUCT REPORTS, 1988

department) for the sample of methyl 1 -thio-P-D-mannoside tetra-acetate, and Dr P. G. Waterman (of Strathclyde University) for the sample of the triterpene derivative.
Sequence 7B Spin filtering; the final pulse (----) is applied on alternate scans only ; extensive phase cycling is required.

9 References
1 F. Bloch, W. W. Hansen, and M. Packard, Phys. Rev., 1956, 70, 474. 2 E. M. Purcell, H. C. Torrey, and R. V. Pound, Phys. Rev., 1946, 69,37. 3 R. R. Ernst and W. A. Anderson, Rev. Sci. Instrum., 1966, 37, 93. 4 H. C . Torrey, Phys. Rev., 1949, 76, 1059. 5 A. D. Bain, J. Magn. Reson., 1984, 56,418. 6 D. I. Hoult and R. E. Richards, Proc. R. SOC.London, Ser. A , 1975, 344,311. 7 E. L. Hahn, Phys. Rev., 1950, 80, 580. 8 M. R. Bendall, D. M. Doddrell, and D. T. Pegg, J. Am. Chem. SOC.,1981, 103, 4603. 9 J. H. Noggle and R. E. Schirmer, 'The Nuclear Overhauser Effect', Academic Press, New York, 1971. 10 K. G. R. Pachler and P. L. Wessels, J. Magn. Reson., 1973, 12, 337. I 1 R. Freeman and G. A. Morris, J. Am. Chem. SOC.,1979, 101, 760. 12 R. R. Ernst, J. Chem. Phys., 1966, 45, 3854. 13 F. Pei and R. Freeman, J. Magn. Reson., 1982, 48, 318. 14 D. W. Brown, T. T. Nakashima, and D. L. Rabenstein, J. Magn. Reson., 1981, 45, 302. 15 D. J. Cookson and B. E. Smith, Org. Magn. Reson., 1981, 16, 111. 16 S. L. Patt and J. N. Shoolery, J. Magn. Reson., 1982, 46, 535. 17 J. C. Madsen, H. Bildsoe, H. J. Jakobsen, and 0. W. Sorensen, J. Magn. Reson., 1986, 67, 243; D. Canet, C. Millot, and J. Brondean, Magn. Reson. Chem., 1986, 24, 951. 18 H. Bildsoe, S. Donstrup, H. J. Jakobsen, and 0. W. Sorensen, J. Magn. Reson., 1983, 53, 154. 19 0. W. Sorensen, S. Donstrup, H. Bildsoe, and H. J. Jakobsen, J. Magn. Reson., 1983, 55, 347. 20 M. R. Bendall, D. T. Pegg, D. M. Doddrell, S. R. Johns, and R. I. Willing, J. Chem. SOC., Chem. Commun., 1982, 1138; V. Rutar, J. Magn. Reson., 1983, 53, 235. 21 M. R. Bendall and D. T. Pegg, J. Magn. Reson., 1983, 53, 272. 22 V. Rutar, J. Magn. Reson., 1982, 48, 155. 23 M. R. Bendall and D. T. Pegg, J. Magn. Reson., 1983, 52, 136; B. K. John and R. E. D. McClung, J. Magn. Reson., 1983, 54, 533. 24 P. Schmitt, J. R. Wesener, and H. Gunther, J. Magn. Reson., 1983,52, 51 1 ; D. M. Doddrell, J. Staunton, and E. D. Laue, ibid., p. 523; J. R. Wesener, P. Schmitt, and H. Gunther, J. Am. Chem. SOC.,1984, 106, 10. 25 G. A. Morris, J. Am. Chem. SOC.,1980, 102, 428; D. P. Burum, and R. R. Ernst, J. Magn. Reson., 1980, 39, 163; P. H. Bolton, ibid., 1980, 41, 287. 26 M. R. Bendall, D. T. Pegg, and D. M. Doddrell, J. Chem. SOC., Chem. Commun., 1982, 872; J . Magn. Reson., 1983, 52, 81. 27 D. T. Pegg, D. M. Doddrell, and M. R. Bendall, J. Chem. Phys., 1982, 77,2745; D. M. Doddrell, D. T. Pegg, and M. R. Bendall, J. Magn. Reson., 1982, 48, 323. 28 D. T. Pegg, D. M. Doddrell, and M. R. Bendall, J. Magn. Reson., 1983, 51, 264. 29 G. A. Morris. Top. Carbon-13 N M R Spectrosc., 1984, 4, 179. 30 H. J. Jakobsen, 0. W. Sorensen, and H. Bildsoe, J. Magn. Reson., 1983, 51, 157. 31 J. M. Bulsing, W. M. Brooks, J. Field, and D. M. Doddrell, J. Magn. Reson., 1984, 56, 167. 32 P. L. Rinaldi and N. J. Baldwin, J. Am. Chem. Soc., 1982, 104, 5791 ; ibid., 1983, 105,7523. 33 T. T. Nakashima, R. E. D. McClung, and B. K. John, J. Magn. Reson., 1984, 58, 27. 34 D. M. Doddrell, J. Staunton, and E. D. Laue, J. Chem. SOC., Chem. Cornrnun., 1983, 602; M. R. Bendall, D. T. Pegg, J. R. Wesener, and H. Gunther, J. Magn. Reson., 1984, 59,223. 35 A. Bax, Top. Carbon-I3 NMR Spectrosc., 1984, 4, 220. 36 G. Bodenhausen, R. Freeman and G. A. Morris, J. Magn. Reson., 1976, 23, 171; G. A. Morris and R. Freeman, ibid., 1978, 29, 433. 37 I. H. Sadler, J. Chem. SOC., Chem. Commun., 1987, 321. 38 0.W. Sorensen and R. R. Ernst, J. Magn. Reson., 1983, 51, 477.

Sequence 7C Single spin filter; the final pulse (---)

is applied on

alternate scans only.

excitation can occur in all systems of two or more coupled nuclei. In some cases it is possible to reduce further (or even to eliminate) the signals from the higher-order spin systems by introducing a further variable delay, A , between the two central 90" pulses and placing a 180" pulse at the centre of this delay, as shown in Sequence 7B. In such cases it is probably simpler to select a suitable average value for t whilst varying A . Figure 28 (c) shows a double-quantum-excitation spectrum of andibenin B that was obtained by setting t at 21 ms and incrementing A in steps of 20 ms from 20 to 400 ms. The sequence presented here and the simple INADEQUATE sequence also cause higher-order multiple-quantum excitation, and spectra from three or more coupled nuclei may be obtainedg6 by appropriate phase cycling. In particular, the spectra from systems of n spin-coupled nuclei dominate, those from fewer coupled nuclei are rejected, and those from more coupled nuclei are reduced if the phases (+) of the generation pulse are cycled through values of = 180" k / n [where k = 0, 1 , 2 ... (2n - l ) ] and the acquired signals are alternately added and subtracted. Also, for systems in which there are an odd number of spins, the phases of the 90" pulses on either side of the delay d must be shifted by 90". Only the most recent commercial spectrometers, fitted with digital phase-shifting networks, allow phase shifts of other than 90" and 180" to be specified, and experiments of this kind cannot at present be carried out on the majority of spectrometers, where phase shifts are limited to 90" and 180". A highly simplified version of Sequence 7B can be usedg7to obtain spectra of singlets only. It consists of a simple spin-echo sequence, with a variable inter-pulse delay period t . A 90" pulse may also be applied on alternate scansg5immediately before data are acquired, as in Sequence 7C. Variation of the delay causes sinusoidal change in the intensities of all coupled signals which, apart from the line located at the chemical-shift position for multiplets with odd numbers of lines, will average to zero if sufficient increments are employed. A spectrum of andibenin B that was obtained by varying t from 10 to 200 ms in increments of 10 ms is shown in Figure 28(d).

Downloaded by Instituto Gulbenkian Ciencia (IGC) on 14 July 2011 Published on 01 January 1988 on http://pubs.rsc.org | doi:10.1039/NP9880500101

8 Reviews and Books


In addition to the reviews cited in the text above, three additional review^^^-'^^ on multi-pulse methods and two recent lo2 books on modern techniques in n.m.r. spectroscopylO1~ are strongly recommended. Acknowledgements The spectra presented in this review were obtained in the chemistry department of Edinburgh University, using Bruker WP2OOSY and WH360 spectrometers funded by the S.E.R.C. Some of these spectra (Figures 14, 15,21, and 24-28) have been essential for structural and resonance assignments and are published for the first time. F o r these spectra, Dr T. J. Simpson (of this department) is thanked for samples of andibenin B, andilesin A, and anditomin, Dr J. C. P. Schwarz (of this

View Online

NATURAL PRODUCT REPORTS, 1988-1.

H. SADLER

127 72 P. J. Hore, E. R. P. Zuiderweg, K. Nicolay, K. Dijkstra, and R. Kaptein, J. Am. Chem. SOC.,1982, 104, 4286. 73 S. R. Johns and R. I. Willing, Magn. Reson. Chem., 1985, 23, 16. 74 A. D. Bain, D. W. Hughes, J. M. Goddington, and R. A. Bell, J. Magn. Reson., 1984, 58, 490. 75 P. Schmitt and H. Gunther, J . Magn. Reson., 1983, 52, 497. 76 J. Brondeau, C. Millot, and D. Canet, J. Magn. Reson., 1984, 57, 319. 77 A. Bax and R. Freeman, J . Magn. Reson., 1981, 44, 542; ref. 71, p. 69. 78 C. Bosch, A. Kumar, R. Baumann, R. R. Ernst, and K. Wutrich, J. Magn. Reson., 1981, 42, 159. 79 W. A. Anderson and R. Freeman, J. Chem. Phys., 1962, 37, 85. 80 R. E. D. McClung and N. R. Krishna, J. Magn. Reson., 1978,29, 573. 81 S . Schaublin, A. Hohener, and R. R. Ernst, J. Magn. Reson., 1974, 13, 196. 82 K. Bock, P. Burton, and L. D. Hall, Can. J. Chem., 1976, 54, 3526. 83 S. V. Ley, A. J. Whittle, and G. E. Hawkes, J. Chem. Res. (S). 1983, 210. 84 P. L. Wessels and K. G. R. Pachler, J. Magn. Reson., 1980, 38, 365. 85 C. Bauer and R. Freeman, J . Magn. Reson., 1985, 61, 376. 86 J. K. M. Sanders and J. D. Mersh, Prog. Nucl. Magn. Reson. Spectrosc., 1982, 15, 353; see p. 361 et seg. 87 B. K. Hunter, L. D. Hall, and J. K. M. Sanders, J. Chem. SOC., Perkin Trans. I , 1983, 657. 88 F. Radler de Aquino Net0 and J. K. M. Sanders, J . Chem. SOC., Perkin Trans. I, 1983, 181. 89 M. Kinns and J. K. M. Sanders, J. Magn. Reson., 1984, 56, 518; K. E. Kover, ibid., 1984, 59, 485. 90 D. Neuhaus, J. Magn. Reson., 1983, 53, 109. 91 J. D. Mersh and J. K. M. Sanders, Tetrahedron Lett., 1981, 22, 4029. 92 M. P. Williamson and D. H. Williams, J. Chem. SOC., Chem. Commun., 1981, 165; D. H. Williams, Acc. Chem. Res., 1984, 17, 364. 93 G. Bodenhauser and C. M. Dobson, J. Magn. Reson., 1981, 44, 212. 94 P. J. Hore, R. M. Scheck, A. Volkeda, R. Kaptein, and J. H. van Boom, J. Magn. Reson., 1982,50,328; D. W. Hughes, T. Neilson, J. M. Coddington, and R. A. Bell, Can. J . Chem., 1984, 62, 1214. 95 0. W. Sorensen, M. H. Levitt, and R. R. Ernst, J . Magn. Reson., 1983, 55, 104. 96 A. J. Shaka and R. Freeman, J. Magn. Reson., 1983, 51, 169. 97 P. H. Bolton, J . Magn. Reson., 1981, 45, 418. 98 R. Benn and H. Gunther, Angew. Chem., Znt. Ed. Engl., 1983, 22, 350. 99 C. J. Turner, Prog. Nucl. Magn. Reson. Spectrosc., 1984, 16, 311. 100 G. A. Morris, Magn. Reson. Chem., 1986, 24, 371. 101 A. E. Derome, Modern NMR Techniques for Chemistry Research , Pergamon Press, Oxford, 1987. 102 J. K. M. Sanders and B. K. Hunter, Modern NMR Spectroscopy, Oxford University Press, Oxford, 1987.

39 U. B. Sorensen, H. Bildsoe, H. J. Jakobsen, and 0. W. Sorensen, J . Magn. Reson., 1985, 65, 222. 40 B. Birdsall and J. Feeney, J. Chem. SOC.,Perkin Trans. 2, 1972, 1643. 41 A. Bax, Top. Carbon-13 N M R Spectrosc., 1984, 4, 200. 42 D. J. Cookson and B. E. Smith, J. Magn. Reson., 1983, 54, 354; ibid., 1984, 56, 510. 43 D. J. Cookson and B. E. Smith, J . Magn. Reson., 1984, 60,125. 44 D. M. Doddrell, W. Brooks, J. Field, and R. M. Lynden-Bell, J. Magn. Reson., 1984, 59, 384. 45 D. M. Doddrell, W. Brooks, J. Field, and R. M. Lynden-Bell, J. Am. Chem. SOC.,1983, 105, 6973. 46 D. G. Davis, D. H. Live, W. C. Agosta, and D. Cowburn, J. Magn. Reson., 1983, 53, 350. 47 H. J. Jakobsen, H. Bildsoe, S. Donstrup, and 0. W. Sorensen, J. Magn. Reson., 1984, 57, 324. 48 G. A. Pearson, J. Magn. Reson., 1985, 64, 487. 49 M. A. Delsuc, E. Guittet, and J. Y. Lallemand, J. Magn. Reson., 1985, 23, 213. 50 J. R. Johnson, N. Shankland, and I. H. Sadler, Tetrahedron, 1985, 41, 3147. 51 G. Commenges and R. C. Rao, J. Magn. Reson., 1984, 58, 496. 52 A. Bax, J. M a p . Reson., 1984, 57, 314; A. Bax, J. A. Ferretti, N. Nashed, and D. M. Jerina, J. Org. Chem., 1985, 50, 3029. 53 A. Bax, C.-H. Niu, and D. Live, J . Am. Chem. SOC.,1984, 106, 1150. 54 J. Feeney and J. Partington, J. Chem. SOC., Chem. Commun., 1973, 611. 55 M. F. Aldersley, F. M. Dean, and B. E. Mann, J. Chem. SOC., Chem. Commun., 1983, 107; N. Niccolai, C. Rossi, V. Brizzi and W. A. Gibbons, J. Am. Chem. SOC.,1984, 106, 5732. 56 J. K. M. Sanders and J. D. Mersh, Prog. Nucl. Magn. Reson. Spectrosc., 1982, 15, 353. 57 K. E. Kover and G. Batta, J . Am. Chem. SOC.,1985, 107, 5829. 58 P. Bigler and M. Kimber, Angew. Chem., Znt. Ed. Engl., 1985, 24, 705; Magn. Reson. Chem., 1986, 24, 972. 59 R. M. Horak, P. S. Steyn, and R. Vleggaar, Magn. Reson. Chem., 1985, 23, 995; J. L. Marshall, C-C and C-H Couplings, Verlag Chemie, Weinheim, 1983. 60 A. Bax, R. Freeman, and S. P. Kempsell, J . Am. Chem. SOC., 1980, 102, 4849; J. Magn. Reson., 1980, 41, 349. 61 A. Wokaun and R. R. Ernst, Chem. Phys. Lett., 1977, 52,407. 62 G. Bodenhausen, Prog. Nucl. Magn. Reson. Spectrosc., 1981, 14, 137. 63 A. Bax, Top. Carbon-13 N M R Spectrosc., 1984, 4, 231. 64 M. H. Levitt and R. R. Ernst, Mol. Phys., 1983, 50, 1109. 65 A. Bax and R. Freeman, J. Magn. Reson., 1980, 41, 507. 66 R. Benn, J . Magn. Reson., 1983, 55, 460. 67 0. W. Sorensen, U. B. Sorensen, and H. J. Jacobsen, J. Magn. Reson., 1984, 59, 332; ibid., 1985, 61, 382. 68 0. W. Sorensen, R. Freeman, T. Frenkiel, T. H. Mareci, and R. Schuck, J . Magn. Reson., 1982, 46, 180. 69 S. W. Sparks and P. D. Ellis, J. Magn. Reson., 1985, 62, 1. 70 A. Bax, R. Freeman, and T. Frenkiel, J. Am. Chem. SOC.,1981, 103, 2102; A. Bax, R. Freeman, T. Frenkiel, and M. H. Levitt, J. Magn. Reson., 1981, 43, 478; A. Bax, Top. Carbon-I3 N M R Spectrosc., 1984, 4, 232. 7 1 A. Bax, Two-dimensional Nuclear Magnetic Resonance in Liquids, Reidel, Amsterdam, 1982, Chapter 5.

Downloaded by Instituto Gulbenkian Ciencia (IGC) on 14 July 2011 Published on 01 January 1988 on http://pubs.rsc.org | doi:10.1039/NP9880500101

You might also like