Technical Note Cite This: Anal. Chem. XXXX, XXX, XXX−XXX
pubs.acs.org/ac
Suppression of Protonated Organic Solvents in NMR Spectroscopy Using a Perfect Echo Low-Pass Filtration Pulse Sequence Peter W. A. Howe* Syngenta, Jealott’s Hill Research Centre, Bracknell, Berkshire, RG42 6EY, U.K. S Supporting Information *
ABSTRACT: Proton NMR spectra are usually acquired using deuterated solvents, but in many cases it is necessary to obtain spectra on samples in protonated solvents. In these cases, the intense resonances of the protonated solvents need to be suppressed to maximize sensitivity and spectral quality. A wide range of highly effective solvent suppression methods have been developed, but additional measures are needed to suppress the 13C satellites of the solvent. Because the satellites represent 1.1% of the original solvent signal, they remain problematic if unsuppressed. The recently proposed DISPEL pulse sequences suppress 13C satellites extremely effectively, and this Technical Note demonstrates that combining DISPEL and presaturation results in exceptionally effective solvent suppression. An important element in the effectiveness is volume selection, which is inherent within the DISPEL sequence. Spectra acquired in protonated dimethlysulfoxide and tetrahydrofuran show that optimum results are obtained by modifying the phase cycle, cycling the pulse-field gradients, and using broadband 13C inversion pulses to reduce the effects of radiofrequency offset and inhomogeneity.
I
n a recent Letter in Analytical Chemistry, Moutzouri et al.1 describe two highly effective NMR pulse sequences that suppress resonances from protons attached to 13C nuclei. They show that the sequences, which they name Destruction of Interfering Satellites by Perfect Echo Low-pass filtration or “DISPEL”, simplify the detection of trace level impurities by NMR spectroscopy, because 13C satellites of major components no longer complicate the spectra. The sequences are based on a combination of isotope editing and the Perfect Echo sequence and have the advantages that they are simple to implement and relatively undemanding of NMR spectrometer hardware. In the final sentence of their Letter, Moutzouri et al. note that DISPEL sequences have potential application for suppression of protonated organic solvents. This Technical Note aims to highlight practical and theoretical aspects of solvent suppression using DISPEL sequences which are not mentioned by Moutzouri et al. In typical applications of NMR, solvents are present at concentrations of >1000× that of the solutes of interest. Although NMR has reasonable dynamic range, intense protonated solvent resonances obscure parts of NMR spectra and distort the spectral baseline making quantitative analysis difficult. NMR spectra are usually acquired using deuterated solvents, but this can be extremely expensive with unusual solvents, such as some used in chemical synthesis or in cases where a subsample of a larger volume of solution is taken for NMR analysis, such as quality control of chemical libraries. If protonated solvents need to be used, a range of solvent suppression methods are available.2,3 The most effective are those based on WATERGATE4 and Excitation Sculpting,5,6 but © XXXX American Chemical Society
these suppress resonances across broad regions of NMR spectra (typically several tens of hertz). The most selective suppression method, and the longest established, is presaturation.2,7,8 This can achieve good suppression of solvent signal(s) without hindering the observation of nearby resonances. However, this selectivity also means that 13C satellites of solvent resonances are not significantly suppressed so they remain as the largest resonances in proton NMR spectra. The combination of DISPEL and presaturation offers a simple method of suppressing solvent resonances and their associated 13C satellites. One important aspect of solvent suppression is “volume selection”, described in detail in the papers by Neuhaus et al.9 and McKay.8 Put briefly, NMR spectra require magnetic field homogeneity because regions which experience different magnetic fields resonate at different frequencies. These regions are normally detected with low sensitivity, but because solvent resonances are so intense these regions contribute significantly to the NMR spectrum and, because they resonate at different frequencies from the central homogeneous regions of the sample, they broaden the area of the spectrum obscured by the solvent. Spectra are significantly improved if signal from inhomogeneous regions is removed. This can be achieved by canceling or dephasing magnetization from regions of poor radiofrequency (rf) homogeneity because regions of poor field Received: February 6, 2018 Accepted: March 19, 2018 Published: March 19, 2018 A
DOI: 10.1021/acs.analchem.8b00621 Anal. Chem. XXXX, XXX, XXX−XXX
Technical Note
Analytical Chemistry
A final practical aspect of the sequences applies if they are implemented on a cryo-cooled NMR probe (Moutzouri et al. show data acquired using a room-temperature probe). As has been recently noted by other authors with excitation sculpting22 and PURGE23 solvent suppression sequences, baseline artifacts and peak distortions are often observed on cryo-cooled probes. These are sufficiently severe to compromise the effectiveness of the sequences and their cause is unclear but they can be removed by summing two experiments where all PFG pulses are reversed in polarity. This solution can also be used effectively with the DISPEL-2 sequence, albeit with the consequence of doubling the number of transients required for a spectrum.
homogeneity typically also have poor rf homogeneity. Volume selection in this way is usually achieved by π-pulses which are either sandwiched by pulsed field gradient (PFG) pulses or phase cycled, so that any signal that is not refocused/inverted is dephased or canceled.9−11 A cursory inspection of the DISPEL sequences reveals that they inherently achieve volume selection through the presence of two refocusing pulses sandwiched by PFG pulses which are additionally phase cycled. This inherent property contributes significantly to the quality of spectra that the DISPEL sequences deliver. It does have the consequence of an increased offset dependence.11 As well as this important benefit, four practical aspects of the DISPEL sequences merit further consideration. First, Moutzouri et al. propose two different pulse sequences. Their recommended sequence contains four separate filtering steps so will be referred to as “DISPEL-4”. This yields exceptionally effective satellite suppression across a wide 1JCH range but requires a total refocusing delay of 16.2 ms. Their alternative sequence contains two filtering steps (hence referred to as DISPEL-2), so yields less effective suppression across a wide 1 JCH range but requires a shorter total refocusing delay of 8.4 ms. In the case considered here of suppressing protonated solvents, there are usually only one or two narrow ranges of 1 JCH to be suppressed, so the DISPEL-2 sequence is preferable owing to its shorter refocusing delay which minimizes relaxation losses. The second practical aspect that merits consideration is the phase cycle proposed by Moutzouri et al. which differs from those proposed in the original publications of the Perfect Echo.12,13 As they note in their Supporting Information, better results are obtained using the original phase cycle where the first proton excitation pulse is cycled independently of the second one; this suppresses artifacts caused by pulse imperfections and relaxation during the echo delay (see pages 33−34 of van Zijl et al.12). Using the original phase cycle reduces the need for the final z-filter because, in the case of weak coupling, the echo delay is very much shorter than 1/JHH so almost all magnetization is refocused in-phase. For example, the echo delay in the two-stage DISPEL sequence is less than 8.2 ms and during this delay, for the protons of an AX3 system with unusually high J-couplings of 20 Hz, more than 99.95% of magnetization is refocused in-phase.12,13 In the case of strong coupling, transfer of magnetization is independent of the echo delay but the effect of the refocusing delay is minimal if it is short compared to the inverse of the frequency difference of the strongly coupled spins.14 The shorter echo delay of DISPEL-2 is an advantage to minimize relaxation and coupling evolution, but it has the counterbalancing disadvantage that it achieves editing using 13C inversion pulses during echo delays. Moutzouri et al. do not address the impact of the wide 13C spectral width on this sequence, even though it is well-known that the 13C spectral width is ineffectively covered by simple inversion pulses.15 Resonance offset and pulse inhomogeneity have been shown to affect the efficiency of 13C editing16,17 but effectiveness can be easily restored by using composite,15 adiabatic,18−21 or shaped pulses17 which are less affected by offset and inhomogeneity. As an example, a simple 24 μs inversion pulse with an rf field of 20.8 kHz only inverts 90% of magnetization at a 5 kHz offset while a basic 90x240y90x composite pulse with the same rf field inverts 99%. Therefore, the simple 13C inversion pulses of DISPEL-2 should be replaced by adiabatic, composite, or shaped pulses to achieve its expected effectiveness.
■
EXPERIMENTAL SECTION Two examples were used to demonstrate the effectiveness of the modified DISPEL-2 sequence. The first was a sample of 0.1 mg of Linalool 24 in 90% protonated/10% deuterated dimethysulfoxide (DMSO), which mimics typical sample conditions used when verifying pharmaceutical or chemical libraries. The second was a sample of 1.4 mg of chlorsulfuron25 in 90% protonated tetrahydrofuran (THF)/10% deuterated acetonitrile, which mimics conditions used for reaction monitoring by NMR spectroscopy. Full details of sample preparation, pulse sequence delays and phase cycling, and spectrometer acquisition parameters are given in the Supporting Information (section S1).
■
RESULTS AND DISCUSSION Figure 1 shows the protonated DMSO peak of the Linalool sample without solvent suppression and with the DISPEL
Figure 1. Part of NMR spectra of the sample of 0.1 mg of Linalool in 90% protonated DMSO acquired using (a) Pulse-acquire and (b) DISPEL-2 with presaturation.
sequence preceded by presaturation. This demonstrates effective suppression of the main and satellite DMSO peaks. Figure 2 shows enlargements of spectra acquired using the DISPEL sequence with and without application of 13C pulses. This demonstrates that the DISPEL sequence attenuates the 13 C satellite peaks 80× compared to the simple Pefect Echo and thus allows observation of the resonances close to the DMSO peaks. The improvement owing to PFG polarity inversion is shown in the Supporting Information (Figure S2). Solvent suppression is of slightly better quality than that obtained using NOESY-PRESAT (see Supporting Information Figures S3 and S4), with the additional advantage that a higher receiver gain can be used resulting in slightly higher signal-to-noise. The B
DOI: 10.1021/acs.analchem.8b00621 Anal. Chem. XXXX, XXX, XXX−XXX
Technical Note
Analytical Chemistry
Figure 2. Subspectra of the Linalool sample acquired using DISPEL-2 with presaturation and (a) without and (b) with application of 13C pulses.
Figure 4. Subspectra of the sample of 1.4 mg of chlorsulfuron in 90% protonated THF acquired using (a) Pulse-acquire and (b) DISPEL-2 with presaturation.
residual negative satellites signals which are observed are unaffected by changes in the echo delays or by phase cycling, so are likely to originate from other mechanisms such as zeroquantum or a combination of radiation damping and the nuclear Overhauser effect. Figure 3 shows one alkene resonance
Figure 3. Alkene region of the spectrum of the Linalool sample acquired using DISPEL-2 with presaturation.
Figure 5. Subspectra of the chlorsulfuron sample acquired using DISPEL-2 with presaturation (a) without and (b) with application of 13 C pulses. Resonances marked “P” originate from THF peroxide.
of Linalool; despite the omission of the final z-filter, no phase distortions of the resonance are observed. The trans 3JHH coupling is 17 Hz, at the high end of the range of proton− proton couplings, suggesting phase distortion is unlikely in most common organic molecules. However, the intensity imbalance between the multiplet components of the resonance differs from that observed in a pulse-acquire spectrum; the components are more symmetric and less “roofed” in the DISPEL sequence. This intensity distortion probably originates from strong coupling26 because the inverse of the frequency difference of the strongly coupled spins is not short compared to the echo delay (Δν−1 ≈ 2.85 ms, τ ≈ 1.82 ms), which Aguilar et al.14 identify as being necessary for strong coupling effects to be small. The distortion can be reduced by reinstating the zfilter27 or applying a π/2 pulse with phase y immediately before acquisition (see Supporting Information Figure S5). Figures 4 and 5 shows analogous spectra for the chlorsulfuron sample. Figure 4 shows the region of the THF solvent resonances without solvent suppression and with presaturation and DISPEL-2. Solvent suppression is good,
even though the spectra were acquired using phase-modulated presaturation of the two solvent resonances; better results may have been obtained if separate spectrometer frequency channels had been available to irradiate both solvent resonances simultaneously. Figure 5 shows enlargements of the region between both solvent peaks acquired using DISPEL with and without application of 13C pulses (full spectra are shown in Supporting Information Figure S6). Most of note are the resonances around 1.8 ppm which originate from the explosive impurity THF peroxide;28 these are exceptionally difficult to discern without suppression of the satellite signals. The combination of DISPEL-2 and presaturation clearly delivers effective solvent suppression, but other methods are available. The first option is to combine NOESY-PRESAT with bilevel broadband 13C decoupling.29 As reported in the original publication of DISPEL, this approach causes sample heating which degrades the quality of NMR spectra. This disadvantage was also noted in the recent publication of Kew et al.,30 who proposed the alternative approach of combining NOESYC
DOI: 10.1021/acs.analchem.8b00621 Anal. Chem. XXXX, XXX, XXX−XXX
Technical Note
Analytical Chemistry PRESAT with selective 13C decoupling of all solvent 13C resonances during the presaturation period. Kew et al. demonstrated automatic calibration of the selective 13C decoupling to minimize the power required and thus avoid spectral distortions caused by sample heating. Their results show this provides a viable alternative to the DISPEL approach. The advantage of their approach is that magnetization of interest is not transverse until observation, so relaxation and coupling evolution are minimized, while the DISPEL approach presented here has the advantage that accurate calibrations of 13 C frequencies and decoupling power are unnecessary.
(7) Hoult, D. I. J. Magn. Reson. 1976, 21, 337−347. (8) McKay, R. T. Concepts Magn. Reson., Part A 2011, 38A, 197−220. (9) Neuhaus, D.; Ismail, I. M.; Chung, C.-W. J. Magn. Reson., Ser. A 1996, 118, 256−263. (10) Simpson, A. J.; Brown, S. A. J. Magn. Reson. 2005, 175, 340− 346. (11) Mo, H.; Raftery, D. J. Magn. Reson. 2008, 190, 1−6. (12) van Zijl, P. C. M.; Moonen, C. T. W.; von Kienlin, M. J. Magn. Reson. 1990, 89, 28−40. (13) Takegoshi, K.; Ogura, K.; Hikichi, K. J. Magn. Reson. 1989, 84, 611−615. (14) Aguilar, J. A.; Nilsson, M.; Bodenhausen, G.; Morris, G. A. Chem. Commun. 2012, 48, 811−813. (15) Freeman, R.; Kempsell, S. P.; Levitt, M. H. J. Magn. Reson. 1980, 38, 453−479. (16) Jahnke, W. J. Magn. Reson., Ser. B 1996, 113, 262−266. (17) Hyre, D. E.; Spicer, L. D. J. Magn. Reson., Ser. B 1995, 108, 12− 21. (18) Tannús, A.; Garwood, M. NMR Biomed. 1997, 10, 423−434. (19) Kupče, E̅ .; Freeman, R. J. Magn. Reson., Ser. A 1995, 115, 273− 276. (20) Skinner, T. E.; Kobzar, K.; Luy, B.; Bendall, M. R.; Bermel, W.; Khaneja, N.; Glaser, S. J. J. Magn. Reson. 2006, 179, 241−249. (21) Hwang, T.-L.; van Zijl, P. C. M.; Garwood, M. J. Magn. Reson. 1998, 133, 200−203. (22) Aguilar, J. A.; Kenwright, S. J. Analyst 2016, 141, 236−242. (23) Le Guennec, A.; Tayyari, F.; Edison, A. S. Anal. Chem. 2017, 89, 8582−8588. (24) http://www.chemspider.com/Chemical-Structure.13849981. html (accessed January 2, 2018). (25) http://www.chemspider.com/Chemical-Structure.43209.html (accessed January 2, 2018). (26) Ernst, R. R.; Bodenhausen, G.; Wokaun, A. Principles of Nuclear Magnetic Resonance in One and Two Dimensions, 1st ed.; Oxford University Press: Oxford, U.K., 1987. (27) Thrippleton, M. J.; Keeler, J. Angew. Chem., Int. Ed. 2003, 42, 3938−3941. (28) http://www.chemspider.com/Chemical-Structure.13393661. html (accessed January 2, 2018). (29) Kupče, E̅ .; Freeman, R.; Wider, G.; Wüthrich, K. J. Magn. Reson., Ser. A 1996, 122, 81−84. (30) Kew, W.; Bell, N. G. A.; Goodall, I.; Uhrín, D. Magn. Reson. Chem. 2017, 55, 785−796.
■
CONCLUSIONS These results demonstrate that combining presaturation with a modified DISPEL-2 sequence provides robust and effective solvent suppression. The removal of 13C satellites reduces the risk that resonances of interest are obscured, and this benefit is obtained with an improvement in spectral quality compared to simple presaturation owing to volume selection. It provides a new benchmark for presaturation-based suppression when NMR is applied to samples in protonated organic solvents.
■
ASSOCIATED CONTENT
S Supporting Information *
The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.analchem.8b00621. Experimental conditions, spectra with and without PFG inversion, comparison of NOESY-PRESAT and DISPEL with presaturation, enlargement of the alkene resonance of Linalool, and DISPEL spectra of chlorsulfuron in protonated THF (PDF)
■
AUTHOR INFORMATION
Corresponding Author
*Phone: +44 1344 424701. Fax: +44 1344 455629. E-mail:
[email protected]. ORCID
Peter W. A. Howe: 0000-0002-0066-7193 Notes
The author declares no competing financial interest.
■
ACKNOWLEDGMENTS I would like to thank the Manchester NMR Methodology Group for sharing their pulse sequence prior to publication and for several useful discussions about their methods. I am very grateful to Dr. Phil Sidebottom (Syngenta Ltd.) for comments on this manuscript.
■
REFERENCES
(1) Moutzouri, P.; Kiraly, P.; Phillips, A. R.; Coombes, S. R.; Nilsson, M.; Morris, G. A. Anal. Chem. 2017, 89, 11898−11901. (2) Albert, K. In On-Line LC-NMR and Related Techniques; Albert, K. Ed.; J. Wiley & Sons: Chichester, U.K., 2002; pp 1−22. (3) Claridge, T. High Resolution NMR Techniques in Organic Chemistry, 3rd ed.; Elsevier: Oxford, U.K., 2016. (4) Piotto, M.; Saudek, V.; Sklenár,̌ V. J. Biomol. NMR 1992, 2, 661− 665. (5) Hwang, T.-L.; Shaka, A. J. J. Magn. Reson., Ser. A 1995, 112, 275− 279. (6) Adams, R. W.; Holroyd, C. M.; Aguilar, J. A.; Nilsson, M.; Morris, G. A. Chem. Commun. 2013, 49, 358−360. D
DOI: 10.1021/acs.analchem.8b00621 Anal. Chem. XXXX, XXX, XXX−XXX
