Visualizing unresolved scalar couplings by real-time J-upscaled NMR

Abstract

Scalar coupling patterns contain a wealth of structural information. The determination, especially of small scalar coupling constants, is often prevented by merging the splittings with the signal line width. Here we show that real-time J-upscaling enables the visualization of unresolved coupling constants in the acquisition dimension of one-dimensional (1D) or multidimensional NMR spectra. This technique, which works by introducing additional scalar coupling evolution delays within the recording of the FID (free induction decay), not only stretches the recorded coupling patterns but also actually enhances the resolution of multiplets, by reducing signal broadening by magnetic field inhomogeneities during the interrupted data acquisition. Enlarging scalar couplings also enables their determination in situations where the spectral resolution is limited, such as in the acquisition dimension of heteronuclear broadband decoupled HSQC (heteronuclear single quantum correlation) spectra.

Figure 1

Pulse sequences used for real-time J-upscaling of (a) regular 1D spectra and (b) a 1D selective TOCSY. Thin and thick black rectangles are nonselective 90° and 180° pulses, respectively. A selective 180° pulse is indicated by a white half ellipse. The following phase cycles were used: (a) ϕ₁ = x, −x, −x, x, y, −y, −y, y; ϕ₂ = x, −x; ϕ₃ = −x, x; ϕ_rec = x, −x, −x, x, y, −y, −y, y; (b) ϕ₁ = x, −x; ϕ₂ = x; ϕ₃ = x, x, −x, −x; ϕ₄ = −x, −x, x, x; ϕ_rec = x, −x. The evolution of chemical shift (CS) and scalar coupling (J) is illustrated in part a. During the actual data acquisition both chemical shift and scalar coupling evolve. In the middle of the interruption delay chemical-shift evolution is refocused.

Figure 2

Regular 1D ¹H spectrum of propanol in DMSO-d₆ together with a close-up image of the central CH₂ group data (1.42 ppm) and the same signal J-upscaled by a factor of 4 and 7. The J-scaling factor λ is indicated for each peak, together with the line width at half height w and the relative resolution enhancement r, which describes λ/w relative to 1/w for the regular proton spectrum. For all spectra, 32k data points were recorded for a spectral width of 8 kHz. For the J-upscaled spectra, 100–200 loops of 10–20 ms were added.

Figure 3

Regular 1D ¹H spectrum of nicotinic acid in DMSO-d₆ together with close-up views of all peaks with upscaling factors of up to 12-fold. All 3-, 4-, and 5-bond coupling constants could be obtained from the upscaled spectra and are indicated. For the regular ¹H spectrum 128k data points were recorded, and 32k for all J-upscaled spectra. The number of loops n was between 60 and 100 for upscaling between λ = 3 and 12, respectively. This corresponds to chunking times t₁/n between 11 and 19 ms.

Figure 4

Structure of azithromycin with its numbering scheme and the region around H-2′ in a regular 1D ¹H NMR spectrum in CDCl₃ (red), together with various H-2′ selectively excited, J-upscaled spectra. For the regular spectrum 32k data points, 16 scans, and a spectral width of 8 kHz were used, while 144 scans were accumulated for the upscaled spectra, with otherwise identical acquisition parameters. All spectra were processed with a 0.1 Hz exponential window function. It is possible to measure J without overlapping nearby signals.

Figure 5

Regular 1D spectrum of azithromycin (50 mg/mL) in CDCl₃ is drawn in red. J-upscaled selective TOCSY spectra (excitation at 4.11 ppm) of azithromycin are shown in blue (λ = 2) and purple (λ = 4). A TOCSY spin lock (MLEV17) with a duration of 150 ms was used. Additionally, a portion of the region between 2.90 and 3.20 ppm is enlarged to see how a double doublet emerges from an overlapped triplet.

Figure 6

Close-up image of the region between 55 and 110 ppm (¹³C) and between 1.5 and 6 ppm (¹H) of a regular HSQC of azithromycin in CDCl₃ is shown in blue. The 3-fold J-upscaled HSQC is shown in red and the 6-fold J-upscaled version in green. For all spectra data matrices of 1024 × 64 data points were recorded with 72 scans each. The spectral widths were 4 kHz (¹H) × 21 kHz (¹³C). For λ = 3 the data chunks had a length of around 9 ms, and for λ = 6 the data chunks were ∼5 ms. All HSQCs were processed with sine square 90° window function after 3-fold zero filling along the direct dimension and zero filling to 256 points and the same window function in the indirect dimension. All peaks visible in this figure are multiplet components.

Figure 7

(a) Regular 1D spectrum of cholic acid (10 mg in methanol-d₄), recorded with 16 scans, 32 768 data points, and a spectral width of 8000 Hz. (b) A regular HSQC spectrum without J-scaling (λ = 1), 128 increments with 16 scans each, and 512 data points in the direct dimensions. The spectral widths of direct and indirect dimensions are 1500 × 6290 Hz. (f) A J-upscaled HSQC with λ = 6 and the same spectral parameters as in part b but double the number of scans. (c–e) 1D extracts of one particular signal, as indicated in red from a regular spectrum and in blue from a J-upscaled HSQC. The signal indicated by an asterisk is from a λ = 9 HSQC to resolve the splittings completely.