Resolution Enhancement in SEA XLOC for Heteronuclear NMR Long‐Range Correlation

Abstract

It is shown how the resolution in SEA XLOC NMR spectra for distinguishing between heteronuclear two‐ and three‐bond correlations for all ¹³C multiplicities can be improved by a modified experiment delivering absorptive profiles in the indirect dimension. The method is demonstrated with applications to ibuprofen and strychnine.

In a recent report,1 a two‐dimensional method, SEA XLOC, was introduced for distinguishing between two‐ and three‐bond heteronuclear long‐range correlations in NMR spectroscopy. The distinction is based on different multiplet widths of zero‐quantum (ZQ or echo) and double‐quantum (2Q or antiecho) signals in the indirect dimension of a 2D spectrum of e. g. a ¹H−¹³C correlation one. For three‐bond correlations associated with a negative passive ² J _CH coupling constant, the ZQ multiplet width is larger than the corresponding 2Q multiplet width, because ³ J _HH−² J _CH enters into the former whereas ³ J _HH+² J _CH enters into the latter. For two‐bond correlations associated with positive passive ³ J _CH coupling constant it is opposite, with ³ J _HH−³ J _CH and ³ J _HH+³ J _CH entering into the ZQ and 2Q multiplet widths, respectively.

A potential drawback of SEA XLOC is that, in order to access these multiplet widths, ZQ and 2Q or echo and antiecho spectra must be kept and inspected separately. Thus the common feature of most multidimensional experiments of combining echo and antiecho data to obtain pure absorption peakshapes is not an option in SEA XLOC in which the spectra as a consequence are displayed in absolute‐value mode to date. Absolute‐value peakshapes include dispersive contributions and are therefore wider than pure absorption peakshapes.

The need to keep echo and antiecho or negative and positive ¹³C frequency contributions separate in order to reveal the two‐/three‐bond distinction makes it appear impossible to obtain absorptive peak profiles in SEA XLOC. However, there is a way around this apparent dilemma because there is another pool of echo and antiecho data not used in the first absolute‐value implementation of SEA XLOC.

In operator terms, the SEA XLOC antiecho and echo parts are in the evolution period represented by $I^{-}{S}^{-}$ and $I^{-}{S}^{+}$ , respectively, with $I$ referring to ¹H and $S$ to ¹³C. In the experiment, the chemical shift contribution from the $I^{-}$ operators is suppressed in the t₁ period in order to have pure ¹³C frequencies arising from the $S^{-}$ and $S^{+}$ operators and with multiple‐quantum multiplet structures in F₁ of the 2D spectrum. The other pool of these $S$ operators are contained in the mirror partners of $I^{-}{S}^{-}$ and $I^{-}{S}^{+}$ , namely $I^{+}{S}^{+}$ and $I^{+}{S}^{-}$ , and they can be picked up in another experiment. $I^{-}{S}^{-}$ and $I^{+}{S}^{+}$ have the same multiplet structure and likewise for $I^{-}{S}^{+}$ and $I^{+}{S}^{-}$ , because the former are both double‐quantum coherences and the latter are both zero‐quantum coherences. Thus echo/antiecho pairs necessary for absorptive profiles can be formed by combining $I^{-}{S}^{-}$ data with $I^{+}{S}^{+}$ data and $I^{-}{S}^{+}$ with $I^{+}{S}^{-}$ . Figure 1 shows how the data associated with the different two‐spin coherences are combined to obtain absorptive profiles. Also shown are the associated coherence transfer pathways including the final point to take into account, namely that quadrature detection only detects $I^{-}$ operators in t₂, so the $I^{+}$ operators must be converted to $I^{-}$ by a ${\pi }^I$ pulse at the interface between the evolution and detection periods of the 2D experiment.

Figure 1.

Illustration of coherence transfer pathways and associated operators for obtaining absorptive profiles along the F₁ dimension of the SEA XLOC 2D spectrum, as described in the text. The first two lines refer to the pulse sequence without the dashed π^H pulse in Figure 2 whilst the remaining two lines refer to the version with that pulse included. Echo data from the first experiment are combined with antiecho data from the second and vice versa. Active operators are in boldface.

The SEA XLOC pulse sequence based on the $I^{-}{S}^{-}$ and $I^{-}{S}^{+}$ operators is outlined in Figure 2 and does not include the final dashed π^H pulse, in contrast to the one based on the $I^{+}{S}^{-}$ and $I^{+}{S}^{+}$ operators including that pulse.

Figure 2.

Pulse sequences for F₁ absorptive SEA XLOC employing a 2^nd order low‐pass J filter. The version without the final dashed π^H pulse selects two‐spin coherences containing $H^{-}$ and the version with the final dashed π^H pulse included selects two‐spin coherences containing $H^{+}$ during t₁. Filled and open bars refer to π/2 and π pulses, respectively. τ₁=0.5[¹ J _min+0.146 (¹ J _max−¹ J _min)]⁻¹, τ₂=0.5[¹ J _max−0.146 (¹ J _max−¹ J _min)]⁻¹. ϕ₁={x, −x, −x, x}, ϕ₂={x, x, −x, −x}, ϕ₃=x and alternating receiver phase {x, −x}. The phase cycle can be repeated with ϕ₃=−x. Δ is the delay for evolution under heteronuclear long‐range couplings, δ a gradient delay, and $t_1^{max}$ <Δ+τ₁+τ₂−2δ−t(180(C)). The delays a and b are ($\Delta +t\left(180\left(C\right)\right)-{\tau }_1-{\tau }_2+t_1)/2$ and $(\Delta -t\left(180\left(C\right)\right)+{\tau }_1+{\tau }_2-t_1)/2$ – $\delta$ , respectively. Selection of the coherences {H⁻C⁻}, {H⁺C⁺}, {H⁻C⁺}, {H⁺C⁻} in t₁ is done by the gradient triples {G₁,G₂,G₃}={−3; 5/2; 15/8}, {3; −5/2; 15/8}, {5/2; −3; 15/8}, {−5/2; 3; 15/8}, respectively. The amplitude of the first three gradients is set an order of magnitude lower than the amplitude of the other ones selecting coherence transfer echo or antiecho. It is an option to place a (π/2)_−x(π/2)_x pulse pair at the point of the dashed π^H pulse in the version without that pulse in order to apply equal rf power in both experiments.

The experiment in Figure 2, however, cannot deliver full 2D pure absorption peakshapes and the reason is the same as the equivalent one for the HMBC experiment which is somewhat related to SEA XLOC. The J _HH coupling constants cause evolution during the entire time between the first π/2 pulse and the start of the t₂ period resulting in mixed phases across multiplets in F₂. The procedure developed for HMBC of combining echo and antiecho data and discarding the dispersive part after the Fourier transformation along t₁ prior to Fourier transformation along t₂ and absolute‐value calculation2 is equally applicable to the new SEA XLOC experiment in Figure 2. The result is absorptive profiles in the indirect dimension of the 2D spectrum.

In order to compare the experiment in Figure 2 with the earlier SEA XLOC pulse sequence1 it must be kept in mind that the former for a given phase cycle requires twice as many scans per t₁ increment as the latter. Thus for the same instrument time the choice is between running the experiment in Figure 2 with a given set of parameters or the earlier version with the same parameters and twice the number of t₁ increments.

If the width at half height of an absorptive Lorentzian line is $\lambda$ the corresponding width for the line in absolute‐value mode is $\sqrt{3}·\lambda$ . Therefore, it is relevant to distinguish the two regimes of coarse and high resolution. High resolution shall refer to the case in which a multiplet in a spectrum is displayed with approximately its natural width. On the other hand, coarse resolution shall refer to the case where a multiplet in a spectrum appears broader than its natural width due to only a small number of recorded t₁ increments.

In the limit of high resolution, there is no benefit from doubling the number of t₁ increments in the earlier implementation of SEA XLOC1 compared to the version in Figure 2 that therefore is expected to offer the best spectral resolution in that regime. In the limit of coarse resolution with double number of t₁ increments in the earlier implementation of SEA XLOC1 the line widths of$\lambda$ and $\sqrt{3}·\lambda$ above in principle become $\lambda$ and $\sqrt{0.75}·\lambda$ indicating similar resolution in the two alternative implementations.

Figures 3 and 4 show excerpts of SEA XLOC spectra of ibuprofen (left in Scheme 1) obtained with the earlier SEA XLOC pulse sequence1 (referred to as F₁‐magnitude) and with the new experiment in Figure 2 (referred to as F₁‐absorptive).

Figure 3.

Overlay of 1D sections of the H2−C3 two‐bond correlation of ibuprofen. Left: F₁‐absorptive ZQ (red solid) and DQ (black solid) recorded with 30 Hz/point resolution superimposed with F₁‐magnitude ZQ (red dotted) and DQ (black dotted) recorded with 15 Hz/point resolution. All data were processed without apodization functions. Middle: F₁‐absorptive ZQ and DQ signals obtained after multiplying data with a cos² function. Right: F1‐magnitude ZQ and DQ signals obtained after multiplying data with a 30° shifted sine function. All sections were obtained by summing up columns across the peaks and scaled to the same intensity for comparison of multiplet widths. For better visualization expanded signals can be found in the Supplementary Information.

Figure 4.

Overlay of 1D sections of the H10−C8 three‐bond correlation of ibuprofen. Left: F₁‐absorptive ZQ (red solid) and DQ (black solid) recorded with 30 Hz/point resolution superimposed on F₁‐magnitude ZQ (red dotted) and DQ (black dotted) recorded with 15 Hz/point resolution. All data were processed without window functions. Middle: F1‐absorptive ZQ and DQ signals obtained after multiplying data with a cos² function. Right: F₁‐magnitude ZQ and DQ signals obtained after multiplying data with a 30° shifted sine function.

Scheme 1.

Structures of ibuprofen (left) and strychnine (right).

The results are in accordance with the general remarks above. As described in the original paper,1 the ZQ profile (red) being narrower than the 2Q profile (black) is indicative of a two‐bond correlation and the opposite is indicative of a three‐bond correlation.

The impact of digital resolution from high to coarse resolution is illustrated in Figure 5 for the H15a−C21 correlation in strychnine (right in Scheme 1) in which only little difference is observed in the limit of coarse resolution but there is a clear advantage to the experiment of Figure 2 for high resolution.

Figure 5.

Overlay of 1D sections of the H15a−C21 three‐bond correlation of strychnine obtained with digital resolution of 14, 28, 56 and 112 Hz/point (from left to right) in F₁‐absorptive and 7, 14, 28 and 56 Hz/point in F₁‐magnitude experiments. The colors and styles of lines are the same as in Figures 3 and 4 and no apodization functions were employed.

In all the above examples, the Difference in Multiplet Width (DMW) of ZQ and 2Q correlations is larger in the F₁‐magnitude than in the F₁‐absorptive spectrum. This effect is expected to be exacerbated for strong coupling because the latter experiment includes the final π^H pulse causing mixing of spin states. For the H23a−C21 correlation in strychnine there is strong coupling between the H23a and H23b protons and it is seen in Figure 6 that assignment as a three‐bond correlation in the F₁‐absorptive spectrum can be made safely only for high resolution.

Figure 6.

Overlay of 1D sections of the H23a−C21 three‐bond correlation of strychnine obtained with digital resolution of 14, 28, 56 and 112 Hz/point (from left to right) in the F₁‐absorptive and 7, 14, 28 and 56 Hz/point in the F₁‐magnitude experiments. No apodization functions were employed.

The picture outlined in Figure 1 strictly holds only for two‐spin systems, because the final dashed π^H pulse produces complementary E.COSY multiplet patterns,3 i. e., with all passive spin states inverted. Thus the F₁‐absorptive SEA XLOC experiment combines normal and complementary E.COSY patterns, and they do not overlap perfectly, which is required for pure absorption profiles. However, as seen in the examples above the technique can deliver clear resolution enhancement, because resolution enhancement occurs even in the case of only partial overlap of normal and complementary E.COSY patterns. It takes relatively large coupling constants of a passive spin to both the two active spins for the resolution enhancement to vanish. Such an example is shown in Figure 7 for the H11a−C10 correlation in strychnine in which the two passive coupling constants to H11b are J(H11a−H11b)=−17.2 Hz and J(H11b−C10)=−7.9 Hz. It is seen in Figure 7d that there is not much difference between these corresponding multiplet widths in the spectra recorded with the two methods.

Figure 7.

The H11a−C10 two‐bond correlation in strychnine. a) and b) obtained using the pulse sequence in Figure 2 without and with the final π^H pulse, respectively, employing a digital resolution of 14 Hz/point and presented in F₁‐magnitude mode. c) Data in a) and b) combined and spectrum presented in F₁‐absorptive mode. In the contour plots 2Q peaks (black) are displaced horizontally with respect to ZQ peaks (red) for better visualization. d) 1D sections extracted in analogy to Figures 3–6, i. e. dashed lines from a 7 Hz/point F₁‐magnitude experiment and full lines from the data in c). The ZQ and 2Q parts are in red and black, respectively, and no apodization functions were employed.

In conclusion, we have presented a way to obtain absorptive profiles and thus resolution enhancement in the indirect dimension of a 2D SEA XLOC spectrum. The experiment has value for high‐resolution applications and when the aim is limited to distinguishing between two‐ and three‐bond correlations. If only a coarse resolution is employed or if coupling constants are to be measured via the original XLOC experiment4 the first version of SEA XLOC is the one of choice.1

Experimental Section

All experiments were performed on a 700 MHz Bruker Avance Neo spectrometer equipped with a TCI z‐gradient prodigy probe. ¹H and ¹³C 90° pulses were 7.56 (1 M ibuprofen in CDCl₃)/7.99 (0.4 M strychnine in CDCl₃) and 12.0 μs, respectively and the sample temperature was 298 K.

SEA XLOC spectra were acquired using the parameters: Δ=83 ms, ¹ J _min=125 Hz and ¹ J _max=165 Hz, spectral widths of 8.4 ppm (¹H) and 180.0 ppm (ibuprofen)/160 ppm (strychnine) (¹³C), relaxation delay 1.6 s. Ibuprofen: F₁‐magnitude spectrum was recorded with 2048 increments in t₁ giving a digital resolution of 15 Hz/point, F₁‐absorptive spectrum was recorded with 1024 increments in t₁ giving a digital resolution of 30 Hz/point, both with 4 scans per increment and 2048 data points in t₂. Strychnine: F₁‐magnitude spectrum was recorded with 4096 increments in t₁ giving a digital resolution of 7 Hz/point, F₁‐absorptive spectrum was recorded with 2048 increments in t₁ giving a digital resolution of 14 Hz/point, both with 2 scans per increment and 2048 data points in t₂. The experiment time was about 4 h in all cases.

Conflict of interest

The authors declare no conflict of interest.

Supporting information

As a service to our authors and readers, this journal provides supporting information supplied by the authors. Such materials are peer reviewed and may be re‐organized for online delivery, but are not copy‐edited or typeset. Technical support issues arising from supporting information (other than missing files) should be addressed to the authors.

Supplementary

T. Gyöngyösi, K. E. Kövér, O. W. Sørensen, ChemistryOpen 2019, 8, 1048.