Figure 11.8


We saw in Chapter 9 that the INEPT experiment can be extended to a1H-13C 2D correlation experiment ("HETCOR") by simply adding an evolution (t1) delay. The coherence flow diagram (Fig. 9.3) shows that 1H coherence is created and evolves during t1 and then is transferred to 13 C coherence, which is observed in t2. To create the equivalent inverse experiment, we simply reverse the roles of13C and 1H:13C coherence is created and evolves during t1, encoding its chemical shift, and then "reverse" INEPT transfer (mixing) moves it to 1H where it is observed in the FID. This would give us a factor of 16 increase in signal strength in recording the FID, but we would lose a factor of 4 because now we are starting with the population difference of 13C rather than that of 1H. For this reason we use a more complex preparation step: The proton is excited with the first pulse, and then this coherence is immediately transferred to 13C. These two parts of the preparation step can be labeled as step 1a and 1b (Fig. 11.9).

Both experiments yield a 2D spectrum that has 13C chemical shifts on one axis and 1H chemical shifts on the other axis, with crosspeaks representing the one-bond relationship between 13C and 1H. The main difference is that the HSQC spectrum has the 13C chemical shifts on the indirect (F1) axis whereas the HETCOR spectrum has the 13 C chemical shifts in the directly detected (F2) dimension. Thus, an HSQC spectrum looks like a HETCOR spectrum turned on its side (90o). We saw in the last section the consequences of detecting the1H signal directly: We have a much stronger signal (16 x) and better resolution (complex

1H multiplets observed in F2), but we have to solve the technical problems of removing the 12C-bound 1H artifact (100 times larger than the desired signal), and we would like to be able to decouple 13C. Thus, there are two important differences between HETCOR and the inverse one-bond experiments HSQC and HMQC:

1. In HMQC/HSQC, the crosspeaks appear in pairs separated in the F2 (horizontal) dimension by the large one-bond CH coupling 150 Hz) and centered on the 1H chemical shift. This coupling can be eliminated by turning on a 13 C decoupler during the acquisition of the FID, which operates just like the 1H decoupler in a 13C-detected experiment. In many cases, however, this coupling gives useful information because the exact value of 1JCH gives us structural information.

2. Between these pairs there will be a vertical streak (parallel to the F1 axis) that represents the 12C-bound proton signal. Because the 12C-bound proton signal is not modulated in t1, the 13C evolution period, it has no F1 frequency, and so it just appears at all F1 frequencies; that is, as a vertical streak. This problem can be solved by coherence pathway selection using phase cycling or gradients.

11.2.2 One-Bond (HSQC/HMQC) Versus Multiple-Bond (HMBC) 2D Spectra

Consider the molecular fragment HaCa-O-CbHb: In some portion of the molecules (about 1.1%), we will have Ca = 13C and Cb = 12C, giving rise to an F1 = Ca, F2 = Ha crosspeak in the HSQC (or HMQC) spectrum due to 1JCH, and an F1 = Ca, F2 = Hb crosspeak in the HMBC spectrum due to 3 JCH (Fig. 11.10, upper dotted line). The HSQC crosspeak will be a wide doublet separated by the large 1JCH coupling (~150 Hz), and the HMBC peak will be a single "blob." In another portion of the molecules (again ~ 1.1%), Ca will be 12C and Cb will be 13C, giving rise to the F1 = Cb, F2 = Hb crosspeak in the HSQC (HMQC) spectrum and a crosspeak at F1 = Cb, F2 = Ha in the HMBC spectrum (Fig. 11.10, lower dotted line). It is important to keep in mind that these are two different experiments, and the data are superimposed on the same 2D spectrum for comparison only. With these two 2D spectra, we can establish that Ha is three bonds or less distant from Cb, and likewise Hb is

Figure 11.11

three bonds or less away from Ca. We can conclude that Ca and Cb are either directly bonded (Ca-Cb) or are separated by one intervening atom (Ca-C-Cb, Ca-O-Cb, Ca-N-Cb, etc.).

Now consider the homonuclear couplings of Ha and Hb. Let's expand the fragment to -CH2-CaHa-O-CbHb-CH- so that Ha appears in the XH spectrum as a triplet and Hb appears as a doublet (Fig. 11.11). These multiplicities also show up in the 13C satellites on either side of the main (1H-12C) peaks. In the one-bond inverse correlation spectrum (HSQC or HMQC, in black), these satellite peaks appear at the F1 position of the corresponding 13 C resonance: Ha satellite peaks at Fi = Ca and Hb satellite peaks at Fi = Cb. In fact, the F2 slice at F1 = Ca is a double triplet, with the doublet coupling being the large 1JCH coupling, and the F2 slice at F1 = Cb is a double doublet, with one of the doublet couplings being the large coupling to Cb. The HMBC crosspeaks (in gray and white for positive and negative intensities) also contain the additional coupling to 13C, but it is much smaller because it is a long-range (2JCH or 3JCH) coupling, which is of the same order of magnitude as the homonuclear (JHH) couplings. If the HMBC data are processed in phase-sensitive mode, the coupling to 13C will appear antiphase, just as the active couplings in COSY spectra appear. This is because the INEPT transfer is always antiphase to antiphase (Fig. 11.9, mixing = step 3). In the one-bond experiments (HSQC/HMQC), this antiphase coherence is refocused, so the large 1JCH coupling appears in-phase, but in HMBC the refocusing delay (1/(2J)) would be too long, so it is left antiphase.

Let's move one step farther and consider a real molecule, ethyl acetate (Fig. 11.12). The cartoon shows a superposition of the HSQC/HMQC spectrum (white) and the magnitude-mode HMBC spectrum (black). 13C decoupling is not used in the HSQC/HMQC spectrum, so the one-bond crosspeaks appear as wide doublets centered on the 1H chemical shift in F2. From center-left to upper-right we see the paired one-bond crosspeaks in roughly diagonal fashion, displaying the homonuclear splitting pattern in F2: quartet for the CH2 group, singlet for the acetate CH3 group, and triplet for the ethyl CH3 group. The carbonyl carbon does not show up in the HMQC spectrum because it has no directly attached proton. In the HMBC spectrum (black), there are two crosspeaks at the 13 C position of the ester carbonyl carbon in F1: A double quartet at the 1H shift of the CH2 group in F2 and a doublet at the 1H shift of the acetate CH3 group in F2. The additional coupling in each

Figure 11.12

case is the long-range JCH, which is comparable to the homonuclear couplings (3JHH) in magnitude. The crosspeak between the CH2 protons of the ethyl group and the carbonyl carbon of the acetate portion is especially important: It connects two parts of the molecule that cannot be connected by homonuclear (COSY or TOCSY) 2D experiments. In other words, HMBC is one way to connect one spin system to another. The only other way to do this is with NOE experiments (NOESY and ROESY). Two-bond HMBC correlations are also predicted between the CH2 carbon and the CH3 proton of the ethyl group (double triplet, center right) and between the ethyl CH3 carbon and the CH2 proton (double quartet, upper left). We will see that in the real world, not all predicted HMBC correlations are observed because some of the couplings are too small to give crosspeaks above noise level. Another aspect of real HMBC spectra is that one-bond artifacts (the wide doublets observed in the HSQC/HMQC spectrum without13 C decoupling) will often show up as strong peaks in the HMBC spectrum. In the worst case, the HMBC spectrum of ethyl acetate would contain all of the crosspeaks shown in Figure 11.12 (black and white).


The HMQC spectrum of 3-heptanone is shown in Figure 11.13, with the 1H spectrum shown at the top. The assignments come from the COSY analysis (Chapter 9). Note the prominent vertical streaks of noise at the F2 frequencies of the most intense 1H peaks: 0.78 ppm (H-7), 0.93 ppm (H-1), and 2.30 ppm (H-2 and H-4). Because this is not a gradient experiment, the 12C-bound 1H artifact is removed by subtraction using a phase cycle, and these are subtraction artifacts. The signal we are removing by subtraction is 100 times larger than the signal we are selecting in the phase cycle, so the artifacts are similar in magnitude to

Figure 11.13

the crosspeaks. The contour threshold is set high to minimize the artifacts, so we see only the most intense peaks: triplets for H-1, H-7, and H-4, but only the central doublets of the H-6 sextet and the H-2 quartet and the central triplet of the H-5 quintet. The one-bond couplings (1 JCH) can be readily measured from the crosspeak separations in F2; all are in the range 125-128 Hz, typical for sp3-hybridized carbon with no bonds to electronegative atoms. Notice the quasi-diagonal pattern of crosspeaks, extending from lower left to upper right, even for the relatively narrow range of chemical shifts represented. This is because the factors that shift protons downfield (in this case proximity to the slightly positive C-3) have a similar effect on the attached carbons.

The HMQC spectrum of sucrose in D2O is shown in Figure 11.14. The glucose-1 cross-peak is found in the lower left because the two bonds to oxygen (anomeric position) lead to downfield shifts in both 1H and 13 C relative to the more typical (for a sugar) singly oxygenated carbon. The "pack" of CH-O positions appears at the center right side, with the CH2OH positions in a tight group at the upper right. The effect of substitution on 13 C shifts was mentioned in Chapter 1, Section 1.3: Every time an H is replaced by C, we get a down-field shift of about 10 ppm. Thus, for singly oxygenated sp3-hybridized carbons, CH3-O is roughly 50-60, CH2-O is 60-70, CH-O is 70-80, and Cq-0 is 80-90 ppm. The "upward" shift in the HMQC spectrum from CH-O to CH2-O is thus a result of the sensitivity of13 C shifts to steric crowding, which is not observed in 1H chemical shifts. Any time there is a significant deviation from the roughly diagonal appearance of the HSQC/HMQC spectrum, it is due to different sensitivities of 1H and 13 C to the chemical environment, and this can be very valuable information in structure determination.

The anomeric position (g1) shows a much larger 1JCH (170 Hz) than the rest of the protonated carbons, which fall in a tight range 141-149 Hz. Oxygen substitution tends to

Figure 11.14

increase the one-bond coupling: We see typical values of 125-128 Hz with no bonds to oxygen (e.g., 3-heptanone), 140-150 Hz for one bond to oxygen and 160-170 Hz for two bonds to oxygen. As we saw with lactose and glucose, the anomeric 1JCH is 160-170 Hz with a orientation falling near 170 and orientation near 160. The g1 position in sucrose is in the a orientation (170 Hz). Comparing to the HETCOR spectrum of sucrose (Fig. 9.7), rotating the HETCOR counterclockwise by 90° and then flipping it left-for-right gives the basic pattern seen in the HMQC (Fig. 11.14), except that in the HMQC the 1JCH couplings expand each crosspeak into a wide doublet in F2 because we are not decoupling 13C in this experiment.

Figure 11.15 shows the upfield (CH3) region of the HMQC spectrum of cholesterol. In general, because inverse experiments are 1H detected, methyl groups give much more intense signals than CH or CH2 groups because there are three equivalent protons. It is almost always possible to increase the contour threshold to a point where only the crosspeaks due to CH3 groups are observed. This is very useful, but the corollary is that CH3 crosspeaks from impurities can sometimes be as strong as the CH and CH2 crosspeaks from the compound being studied. Ignoring the wide 1JCH couplings, one can see that two of the CH3 crosspeaks are singlets; that is, they have no homonuclear couplings. This means that the CH3 group is attached to a quaternary carbon: CH3-Cq (the possibilities of CH3-O or CH3-N can be ruled out because of the upfield 1H and 13C chemical shifts). These singlet methyl groups are very useful in structure determination because of their unambiguous interpretation in the HMBC spectrum. The crosspeak with a 13C shift of 18.7 ppm shows a clear doublet structure in F2: It is a methyl group attached to a CH carbon. Two more "doublet" methyl groups can be seen at F1 = 22.58 and 22.80 ppm, partially overlapped. The "diagonal" shape of the crosspeaks (see insets) makes it clear that the upfield 13C (22.58 ppm) is connected to the upfield 1H signal, even though these two doublets are not resolved in the 1H spectrum (top).

0 0

Post a comment