the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
Paramagnetic Relaxation Enhancement to Improve Sensitivity of 1,1-ADEQUATE NMR
Abstract. The 1,1-ADEQUATE NMR experiment provides easily interpretable spectra to identify two-bond proton-carbon correlations in the structural analysis of small molecules. Despite this, the low natural abundance of 13C drastically reduces the efficacy of this experiment due to its dependence on C-C magnetization transfers. A reduction of T1 and T2 relaxation times via paramagnetic relaxation enhancement (PRE) represents an opportunity to improve sensitivity and reduce experiment times by facilitating a reduction in recycle delays. In this work, we demonstrate Cr3+ at a concentration of 10 mM to be an excellent PRE agent. Paramagnetic relaxation enhancement was then applied in 1,1-ADEQUATE analysis of several compounds, limonene, camphor, amoxapine and quinidine, demonstrating signal-to-noise improvements up to 4.5-fold over diamagnetic conditions.
- Preprint
(927 KB) - Metadata XML
-
Supplement
(594 KB) - BibTeX
- EndNote
Status: open (until 04 Dec 2026)
- CC1: 'Comment on mr-2026-11', Tom Barbara, 28 Oct 2026 reply
-
CC2: 'Comment on mr-2026-11', Julien Furrer, 09 Nov 2026
reply
This study focuses on the 1,1-ADEQUATE experiment, a powerful structural tool whose low sensitivity typically restricts its use in routine NMR. Thus, methods to enhance its signal-to-noise ratio are highly valuable; here, the authors successfully address this limitation by utilizing paramagnetic relaxation enhancement. The proposed PRE-enhanced 1,1-ADEQUATE is valuable, and merits to be shared among the NMR community. In my opinion, there are three key aspects that the authors should investigate further or highlight more prominently in their manuscript:
- The authors recorded their data on a 600 MHz equipped with a CryoProbe, equipment that is not universally available in typical analytical facilities or routine research laboratories. To demonstrate the broad applicability of this PRE-enhanced 1,1-ADEQUATE method, it would be highly valuable to include 1–2 experiments acquired on a more standard configuration (e.g., a 400 or 500 MHz spectrometer with a room-temperature probe). Additionally, given that the current sample concentrations exceed 100 mM, the authors should discuss or experimentally define the practical lower concentration limit under which this experiment can be reasonably envisioned.
- It appears that the PRE-enhanced 1,1-ADEQUATE experiment does not yield significant S/N improvements for amoxapine and quinidine—the samples that would benefit most from a sensitivity boost, and for which the 1,1-ADEQUATE is helpful. Conversely, the other examples are very simple that the practical necessity of a 1,1-ADEQUATE experiment is debatable (I fully understand the need to choose simple and reliable molecules for demonstrating the efficacy of the method, though). To truly demonstrate the utility of the proposed method, the authors should include a more challenging, "real-world" application, if I may use that expression. Testing the sequence on a mixture of diastereomers or a demanding structural scaffold (such as cholesterol or strychnine) would greatly clarify its true analytical value.
- Normally, an objective SNR comparison between different samples must be performed on data processed without linear prediction. The used 4x linear prediction can have the consequence that the noise value is falsely low and calculated SNR too high. Since comparisons between resonances and identically processed spectra are carried out, this should not be too detrimental for the quality of the data; nevertheless, a check for one sample and spectra processed without LP should be performed.
Other points:
Several figure captions (Figs. 2–7) contain result interpretations rather than purely descriptive text. Example Fig 2: “…. Nickel and Vanadium had minimal… occurring in samples prepared with chromium”. These interpretative sentences should be moved to the main text or removed if the points are already addressed there.
Lines 10-11: May be inserting the sentence “maximizing sensitivity per unit of time”, as in the main text, would be suitable for the abstract.
Lines 32-34: The authors may want to cite the 2J,3J-HMBC (JMR, 2000, 146, 232), and the H2OBC, HOBC (MRC, 2017, 55, 515 ; Chem Comm. 2019, 55, 12208). Note that the i-HMBC, based on isotopic shifts, offers a general and reliable solution for distinguishing 2JCH from 3JCH correlations, even though the experiment is demanding experimentally (long measurement time).
Lines 35-36: An important reference is missing for the 1,1-ADEQUATE : Chem. Commun. (2022) 58 (93): 13011–13014, which demonstrates an S/N improvement over the standard ADEQUATE sequence used by the authors. It would be useful to assess whether this improved ADEQUATE method also benefits and to which extent from the present PRE method.
Table 2: Processing parameters: If the authors applied a pure cosine square function, as seems to be the case, the LB and Gaussian parameters are completely ignored.
Tables 4-6: The total acquisition time for Camphor, amoxapine and quinidine should be indicated.
Figure 1. I personally have difficulty differentiating 2JCH correlations from HSQC artifacts (color-blind person). Dark blue or black peaks would be better !
Relaxation measurements: While the authors provide measured T1 relaxation times, T2 values are missing. It would be nice to have them as well (even if the line broadening can be used as a guideline).
Lines 158-160: did the author investigate higher concentration of Cr3+? Certainly, as mentioned by the authors, the S/N of several correlations would probably decrease further, but some may also increase further (i.e. H4>C5, H4>C4, H8>C7).
Figure 5 and lines 226-247. It is not immediately clear to the reader that these results refer to the limonene sample. While the provided S/N ratios are relative, reporting individual ratios would be useful as well, as resonance intensities likely vary depending on the chosen delay during INEPT. Specifically, the H4>C3 and H9>C7 correlations are expected to be more intense when INEPT is optimized for 150 and 180 Hz. Also, the experimental conditions and procedure (line 90-97) are not completely clear, Table S3 being absent in the SI. Is Figure S3 meant here?. Yet, Figure S3 provides the S/N changes of limonene upon addition of Fe(acac)3.
Figure 7: Some correlations are missing, for the 3-4 carbons near C11/H11 (potentially H8/C8, H9/C9, and H10/C10, depending on how the atoms were labeled)? Is there a reason for that? In the case of overlapped resonances, this should be mentioned.
Citation: https://doi.org/10.5194/mr-2026-11-CC2 -
RC2: 'Reply on CC2', Julien Furrer, 10 Nov 2026
reply
Please note that the original was intended as my reviewer assessment.
It was mistakenly posted as a general comment, and I sincerely apologize for the confusion.
Citation: https://doi.org/10.5194/mr-2026-11-RC2
-
RC1: 'Comment on mr-2026-11', Julien Furrer, 10 Nov 2026
reply
This study focuses on the 1,1-ADEQUATE experiment, a powerful structural tool whose low sensitivity typically restricts its use in routine NMR. Thus, methods to enhance its signal-to-noise ratio are highly valuable; here, the authors successfully address this limitation by utilizing paramagnetic relaxation enhancement. The proposed PRE-enhanced 1,1-ADEQUATE is valuable, and merits to be shared among the NMR community. In my opinion, there are three key aspects that the authors should investigate further or highlight more prominently in their manuscript:
- The authors recorded their data on a 600 MHz equipped with a CryoProbe, equipment that is not universally available in typical analytical facilities or routine research laboratories. To demonstrate the broad applicability of this PRE-enhanced 1,1-ADEQUATE method, it would be highly valuable to include 1–2 experiments acquired on a more standard configuration (e.g., a 400 or 500 MHz spectrometer with a room-temperature probe). Additionally, given that the current sample concentrations exceed 100 mM, the authors should discuss or experimentally define the practical lower concentration limit under which this experiment can be reasonably envisioned.
- It appears that the PRE-enhanced 1,1-ADEQUATE experiment does not yield significant S/N improvements for amoxapine and quinidine—the samples that would benefit most from a sensitivity boost, and for which the 1,1-ADEQUATE is helpful. Conversely, the other examples are very simple that the practical necessity of a 1,1-ADEQUATE experiment is debatable (I fully understand the need to choose simple and reliable molecules for demonstrating the efficacy of the method, though). To truly demonstrate the utility of the proposed method, the authors should include a more challenging, "real-world" application, if I may use that expression. Testing the sequence on a mixture of diastereomers or a demanding structural scaffold (such as cholesterol or strychnine) would greatly clarify its true analytical value.
- Normally, an objective SNR comparison between different samples must be performed on data processed without linear prediction. The used 4x linear prediction can have the consequence that the noise value is falsely low and calculated SNR too high. Since comparisons between resonances and identically processed spectra are carried out, this should not be too detrimental for the quality of the data; nevertheless, a check for one sample and spectra processed without LP should be performed.
Other points:
Several figure captions (Figs. 2–7) contain result interpretations rather than purely descriptive text. Example Fig 2: “…. Nickel and Vanadium had minimal… occurring in samples prepared with chromium”. These interpretative sentences should be moved to the main text or removed if the points are already addressed there.
Lines 10-11: May be inserting the sentence “maximizing sensitivity per unit of time”, as in the main text, would be suitable for the abstract.
Lines 32-34: The authors may want to cite the 2J,3J-HMBC (JMR, 2000, 146, 232), and the H2OBC, HOBC (MRC, 2017, 55, 515 ; Chem Comm. 2019, 55, 12208). Note that the i-HMBC, based on isotopic shifts, offers a general and reliable solution for distinguishing 2JCH from 3JCH correlations, even though the experiment is demanding experimentally (long measurement time).
Lines 35-36: An important reference is missing for the 1,1-ADEQUATE : Chem. Commun. (2022) 58 (93): 13011–13014, which demonstrates an S/N improvement over the standard ADEQUATE sequence used by the authors. It would be useful to assess whether this improved ADEQUATE method also benefits and to which extent from the present PRE method.
Table 2: Processing parameters: If the authors applied a pure cosine square function, as seems to be the case, the LB and Gaussian parameters are completely ignored.
Tables 4-6: The total acquisition time for Camphor, amoxapine and quinidine should be indicated.
Figure 1. I personally have difficulty differentiating 2JCH correlations from HSQC artifacts (color-blind person). Dark blue or black peaks would be better !
Relaxation measurements: While the authors provide measured T1 relaxation times, T2 values are missing. It would be nice to have them as well (even if the line broadening can be used as a guideline).
Lines 158-160: did the author investigate higher concentration of Cr3+? Certainly, as mentioned by the authors, the S/N of several correlations would probably decrease further, but some may also increase further (i.e. H4>C5, H4>C4, H8>C7).
Figure 5 and lines 226-247. It is not immediately clear to the reader that these results refer to the limonene sample. While the provided S/N ratios are relative, reporting individual ratios would be useful as well, as resonance intensities likely vary depending on the chosen delay during INEPT. Specifically, the H4>C3 and H9>C7 correlations are expected to be more intense when INEPT is optimized for 150 and 180 Hz. Also, the experimental conditions and procedure (line 90-97) are not completely clear, Table S3 being absent in the SI. Is Figure S3 meant here?. Yet, Figure S3 provides the S/N changes of limonene upon addition of Fe(acac)3.
Figure 7: Some correlations are missing, for the 3-4 carbons near C11/H11 (potentially H8/C8, H9/C9, and H10/C10, depending on how the atoms were labeled)? Is there a reason for that? In the case of overlapped resonances, this should be mentioned.
Citation: https://doi.org/10.5194/mr-2026-11-RC1 -
RC3: 'Comment on mr-2026-11', Anonymous Referee #2, 11 Nov 2026
reply
Paramagnetic compounds can enhance the signal-to-noise ratio of NMR experiments by accelerating the restauration of longitudinal magnetisation, thus allowing the use of faster repetition rates. The authors applied this very old concept to the 1,1-ADEQUATE experiment, which is challenged by low sensitivity at natural isotopic abundance. The PREs prove to present a highly practical technique which, as this article shows, can deliver remarkable time savings. Interestingly, the paramagnetic relaxation enhancements notably reduce the dependence of the sensitivity on the exact choice of the JCC-evolution delays by reducing the penalty for shortening the JCC evolution delays.
Abstract: The third sentence seems to imply that reducing T2 relaxation times is an advantage. The sentence would be clearer, if T2 were dropped from it.
Table 1: sensitivity is determined by the acquisition times in both dimensions and independent of the spectral widths used. It is confusing and unnecessary to specify the offset frequency, spectral widths and acquisition times with more than 3 significant digits. For better readability, it would be nice to condense Tables 1 and 4-6 into a single table.
Line 48: define the parameter d1 as the relaxation or recovery delay.
Line 49: define abbreviations (like S/N) on first occurrence.
Line 58: it would be helpful to also report the molar concentration of limonene and the other analytes investigated.
Figure 3 shows spectra with very different concentrations of paramagnetic agents. By what criteria exactly were these the ‘optimum PRE conditions’?
Line 222, ‘their unrestrained bond rotation’: the wording is a bit confusing as, in the case of H9, the C6-C7 bond can rotate but not the double bond.
Figure 5: were these data obtained with the limonene sample containing 10 mM chromium? Do the bars report averages over multiple cross-peaks? Did the paramagnetic conditions increase the ranges of signal-to-noise ratios observed between different cross-peaks? A box plot would be a more informative representation of the results.
Line 239: isn’t the default value JHC = 145 Hz?
Line 270: how would nitrogen atoms exchange? Could the tertiary nitrogens replace acetylacetonate from the chromium complex, resulting in slower rotation about the N-C bond between the 6- and 7-membered rings and, hence, in narrower lines of the protons 16-20 as shown in Figure S13?
Figure 8: it would be nice to indicate paramagnetic and diamagnetic results with the same code (cross or diamond) as in Figure 7. It is unfortunate that the recommended complex (chromium acetylacetonate) doesn’t seem to be a silver bullet for all molecules, as the H17->C18 correlation is attenuated by the chromium complex. Further attenuation of the intensity of an already weak cross-peak by the PRE is an unfortunate outcome.
Line 306: is the slight decrease in S/N reported for only a single cross-peak of amoxapine (Figure 7) significant?
The doi link to the data does not seem to work. In any case, the doi link should lead not only to the processed data, but also to the raw data (serial file and parameters) to facilitate reproducing the results.
Figure S10: is the label correct? The diamagnetic sample is reported to deliver greater S/N. When the experiment was performed in triplicate, does this include the preparations of the samples?
Figure S12: what is the explanation for increased T1 relaxation times of methyl groups observed after the addition of iron or chromium? Why is the effect greater with iron?
Figure S13: it would be helpful to add a structure drawing of amoxapine in the figure.
It would be nice to be consistent about writing subscripts in italics or not. Compound names like limonene should be spelled without capital.
Citation: https://doi.org/10.5194/mr-2026-11-RC3
Viewed
| HTML | XML | Total | Supplement | BibTeX | EndNote | |
|---|---|---|---|---|---|---|
| 216 | 68 | 29 | 313 | 65 | 22 | 28 |
- HTML: 216
- PDF: 68
- XML: 29
- Total: 313
- Supplement: 65
- BibTeX: 22
- EndNote: 28
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
Seeing this reminded me of an old paper by Debra Mattellio and Ray Freeman: Blood, Sweat, and Tears. Toward a Rehabilitation of the INADEQUATE Experiment, Journal of Magnetic Resonance vol 135 (1998). They used dissolved oxygen to promote increased relaxation rates, and worked on this when Ray did a sabbatical at Varian.