Fast Proton NMR Detection of Aqueous Ammonia with Relaxation Agent and Nitrogen Decoupling

Abstract

The Haber–Bosch process, which synthesizes ammonia (NH₃) from nitrogen (N₂) and hydrogen (H₂), consumes approximately 2% of the global energy supply. A sustainable alternative is the direct electrochemical conversion of N₂ to NH₃. The selectivity and activity of the electrocatalysts for this process are assessed by quantifying the NH₃ present in the electrolyte. Compared with other analytical methods, ¹H NMR offers a straightforward approach for detecting NH₃ (by analyzing NH₄⁺). ¹H NMR method can also definitely confirm that the detected ammonia originates from the electroreduction of N₂ by comparing results obtained from isotopically labeled ¹⁵N₂ and regular ¹⁴N₂ gases. This capability is unique to the ¹H NMR method, as no alternative approaches offer this level of specificity. However, this method suffers from low sensitivity when measuring NH₄⁺ of low concentration of such as at μM or lower. To address this issue, we developed a novel approach that improves sensitivity by ∼3-fold through the introduction of ¹⁴N decoupling during the ¹H NMR data acquisition. Recently [Kolen M.et al. ACS Omega 2021, 6, 5698–5704], demonstrated a ∼3.5-fold increase in sensitivity by using a 1 mM concentration of the paramagnetic relaxation agent Gd³⁺. By combining our ¹⁴N decoupling technique with the relaxation agent Gd³⁺, we achieved a synergistic enhancement in sensitivity, resulting in an overall ∼10.9-fold sensitivity increase for the ¹H NMR detection of ¹⁴NH₄⁺. This translates to a reduction in NMR detection time by a factor of ∼119 (10.9²). This significant advancement enables the fast detection of ammonia at μM concentration or lower. ¹H NMR of ¹⁵NH₄⁺ with ¹⁵N decoupling was also demonstrated.

Introduction

Ammonia (NH₃) is an efficient hydrogen carrier and has numerous industrial applications, including the production of fertilizers and explosives.¹⁻³ The Haber–Bosch process, which has been used for over 100 years, synthesizes NH₃ from nitrogen (N₂) and hydrogen (H₂) using an iron catalyst at high temperature and pressure. This process, however, consumes ∼2% of global energy,^4,5 drives the search for alternative, more sustainable methods. One promising approach is electrocatalytic nitrogen reduction (e-NRR), which produces NH₃ from nitrogen and water using an applied voltage. This method is considered a green and sustainable alternative for NH₃ synthesis. However, the current yield from electrocatalytic nitrogen reduction remains at the micromolar (μM) level,⁶⁻⁹ making quantitative measurement of NH₃, new catalyst development, and reaction mechanism analysis of e-NRR challenging. In addition, the e-NRR system is highly prone to false-positive results caused by the electroreduction of impurities (e.g., NO₂^–/NO₃^–) present in the electrolytic cell, which can potentially overshadow the true signal entirely.^10,11 Therefore, it is crucial to provide definitive proof that any detected ammonia originates from the intended electroreduction process.¹²

Various analytical techniques have been developed to quantify NH₃, such as ion chromatography,¹³ fluorescence,¹⁴ laser absorption spectroscopy,¹⁵ colorimetric methods,¹⁶ and ¹H NMR.¹⁷⁻²⁷¹H NMR is unique in its ability to provide both accurate quantification of NH₃ and N isotope tracing.^28,29 Specifically, it can definitively confirm that the detected ammonia originates from the electroreduction of N₂ by utilizing isotopically labeled ¹⁵N₂ gas—a capability unmatched by any alternative approach.¹⁷ However, this method suffers from low sensitivity, especially at low NH₄⁺ concentrations, due to interference from the large water ¹H NMR signal and a few other factors. Several approaches have been developed to address these issues. Nielander et al. used a pulsed field gradient spin echo (PGSE) pulse sequence with a selective 180° shaped pulse to target the NH₄⁺ signal while avoiding excitation of the water signal.¹⁸ Other methods, such as excitation sculpting (ES)²⁵ or water suppression by gradient tailored excitation (WATERGATE)²⁶ pulse sequences, have also been employed for water signal suppression. Assafiri et al. reported enhanced sensitivity by using a 3 mm NMR tube in a 5 mm NMR probe due to radiation damping reduction with a 3 mm NMR tube.²⁷ More recently, Kolen et al. demonstrated a ∼3.5-fold sensitivity increase by using the paramagnetic relaxation agent Gd³⁺.²⁵

In all these studies, ¹H NMR signals of NH₄⁺ were measured directly, resulting in a triplet for ¹⁴NH₄⁺ and a doublet for ¹⁵NH₄⁺, respectively.^7,9,17−27 Here we introduce a new approach that increases ¹H NMR sensitivity by ∼3-fold through ¹⁴N decoupling during ¹H NMR acquisition. By combining this technique with the use of Gd³⁺, we achieved a synergistic effect, enhancing sensitivity by ∼10.9-fold for ¹H NMR of ¹⁴NH₄⁺. This corresponds to a reduction in NMR detection time by a factor of ∼119 (10.9²). ¹H NMR of ammonia with ¹⁵N decoupling was also demonstrated. This advancement allows for the detection of much lower concentration of NH_3.

Experimental Section

¹⁴NH₄Cl (99.995%), ¹⁵NH₄Cl (≥98 atom %, ¹⁵N ≥ 99% CP), maleic acid (≥99%), and H₂SO₄ (≥97.5%) were purchased from Sigma-Aldrich. Gadolinium(III) nitrate hexahydrate (99.9%) was obtained from Fisher Scientific. DMSO-d₆ (99.9% D) was obtained from Cambridge Isotope Laboratories. NMR was conducted with a 400 MHz Bruker Avance III NMR spectrometer with a 5 mm broadband observe (BBO) or a 10 mm BBO probe at 25 °C. ¹H NMR transmitter frequency was set on water chemical shift, ¹⁴N and ¹⁵N decoupling frequencies were set on-resonance on the ¹⁴N NMR signal of ¹⁴NH₄⁺ and ¹⁵N NMR signal of ¹⁵NH₄⁺, respectively. Detailed experimental parameters are listed in the Results and Discussion section and Supporting Information (SI).

Results and Discussion

We began our experiments using a 400 MHz 5 mm BBO probe, where the inner coil is the X-coil and the outer coil is the ¹H-coil. To start, we measured the ¹⁴N 90° pulse using a sample containing 500 μL of 10 wt % NH₄Cl and 50 μL of DMSO-d₆. On our spectrometer, the ¹⁴N 90° pulse is 19.25 μs at 120 W radio frequency (RF) power. The power level for a 200 μs composite decoupling (CPD) pulse was then calculated using Bruker’s “edprosol” program. For the continuous wave (CW) decoupling portion, we applied an additional −6 dB of power (i.e., increasing power by 2-fold) when employing the bi-Waltz65–256pl method for ¹⁴N decoupling, based on our previous studies.^30,31

As the WATERGATE method is associated with baseline roll and signal phasing issues³² and the excitation sculpting (ES) NMR pulse sequence is well-known for its effective water suppression for the NMR analysis of NH₄⁺ signals,²⁵ we evaluated four ES pulse sequences in the studies:

• (1)Bruker “zgesgp”: A standard ES pulse sequence for water suppression.
• (2)“zgesgpig”: An enhanced version of “zgesgp” that includes inverse gated decoupling on the f2 channel.
• (3)“zgesfpgp”: This pulse sequence incorporates a 90° water flip-back pulse at the start of the “zgesgp” sequence on the f1 channel. The first two pulses (90°flip-back and hard 90°) will store most of the water magnetization along +z (since 90°flip back pulse is applied on −x, while hard 90°pulse is on +x, so effectively 0° rotation). The remaining water magnetization in the transverse plane (caused by radiation damping) will then be defocused by the following excitation sculpting element. These can be advantageous when using a short relaxation delay (D₁). It also reduces radiation damping, which can broaden the NH₄⁺ signals and decrease the signal-to-noise ratio (S/N). Note that the radiation damping becomes a significant issue with a cryoprobe on a high field NMR spectrometer.²⁷
• (4)“zgesfpgpig”: This combines the features of “zgesfpgp” with inverse gated decoupling on the f2 channel.

The new pulse sequences “zgesgpig” and “zgesfpgpig” are shown in Figure 1 (detailed Bruker pulse sequence codes and parameters are provided in the Supporting Information).

Figure 1.

New pulse sequences of zgesgpig and zgesfpgpig. Blue rectangles are hard, blue bullets are shaped and black bullets are pulsed field gradient pulses, respectively.

Sample without Gd³⁺ in a 5 mm NMR Tube with a 5 mm BBO Probe

To evaluate the pulse sequences and different decoupling methods, we first tested a 32 mM NH₄⁺ high-concentration sample. According to previous reports, the ¹H spin–lattice relaxation time (T₁) of NH₄⁺ at 25 °C is ∼2.2 s.²⁵ Based on this, we used a relaxation delay (D₁) of 10.36 s and an acquisition time (AQ) of 0.64 s in our experiments (D₁ + AQ = 5 × T₁). Figure 2 shows the ¹H NMR spectra of the 32 mM NH₄⁺ sample acquired with the four ES pulse sequences discussed previously. When using the “zgesgp” and “zgesfpgp” pulse sequences (without ¹⁴N decoupling), typical ¹H triplet signals of NH₄⁺ were observed. The triplet pattern arises from the coupling between ¹H and ¹⁴N nuclei of the NH₄⁺. In contrast, when using “zgesgpig” and “zgesfpgpig” sequences (with ¹⁴N decoupling), the ¹H–¹⁴N heteronuclear coupling pattern of the NH₄⁺ signal was eliminated, resulting in a singlet signal. The bi-Waltz65–256pl decoupling method was employed for ¹⁴N decoupling in Figure 2, as the NH₄⁺ signal is strong at this concentration. Using conventional Waltz65 decoupling^30,31 at such a high NH₄⁺ concentration produced decoupling artifacts (Figure S1).

Figure 2.

¹H NMR spectra of 32 mM NH₄⁺ with different NMR pulse sequences. The decoupling method used in “zgesgpig” and “zgesfpgpig” is bi-Waltz65–256pl. The sample used was 525 μL of 41.1 mM NH₄Cl, 50 μL of 0.5 M H₂SO₄, 50 μL of DMSO-d₆, 25 μL of 12.5 mM maleic acid and 25 μL of H₂O.

Table 1 summarizes the detailed results for the 32 mM NH₄⁺ sample, with the S/N values representing the average of three experiments. Notably, a ∼ 3-fold increase in sensitivity was achieved with both “zgesgpig” and “zgesfpgpig” pulse sequences.

Table 1. ¹H NMR S/N Ratio Results for a 32 mM NH₄⁺ Solution Obtained with Different NMR Pulse Sequences and Decoupling Methods by Using a 5 mm NMR Tube in a 5 mm BBO Probe^a.

pulse sequence	decoupling method	S/N	S/N gain with decoupling
zgesgp	no decoupling	901
zgesgpig	waltz65	2685	3.0
zgesgpig	bi-waltz65–256pl	2647	2.9
zgesfpgpig	waltz65	2511	2.8
zgesfpgpig	bi-waltz65–256pl	2571	2.9

^aThe S/N ratio represents the average of 3 experiments. Sample: 525 μL of 41.1 mM NH₄Cl, 50 μL of 0.5 M H₂SO₄, 50 μL of DMSO-d₆, 25 μL of 12.5 mM maleic acid, and 25 μL of H₂O.

Sample without Gd³⁺ in a 3 mm NMR Tube with a 5 mm BBO Probe

As mentioned in the introduction, Assafiri et al. reported that using a 3 mm NMR tube in a 5 mm probe could enhance S/N, even with the same sample concentration. The improvement was attributed to the reduction in radiation damping.²⁷ We re-evaluated this claim by analyzing the same 32 mM NH₄⁺ sample, previously tested in a 5 mm NMR tube with a 5 mm BBO probe, in a 3 mm NMR tube with the 5 mm BBO probe. The results, averaged from three experiments, are shown in Table 2. The findings reveal three key points:

• 1.A ∼3-fold increase in sensitivity was achieved with both the “zgesgpig” and “zgesfpgpig” pulse sequences, consistent with our previous results.
• 2.The “zgesfpgpig” pulse sequence performed similarly to the “zgesgpig” sequence under these experimental conditions.
• 3.Contrary to the expectation, using a 3 mm NMR tube in the 5 mm BBO probe resulted in ∼72% loss of sensitivity, rather than a gain, compared to the S/N values listed in Table 1.

Table 2. ¹H NMR S/N Results Obtained on a 32 mM NH₄⁺ Solution with Different NMR Pulse Sequences and Decoupling Methods by Using a 3 mm NMR Tube with a 5 mm BBO Probe^a.

pulse sequence	decoupling method	S/N	S/N gain with decoupling
zgesgp	no decoupling	249
zgesgpig	waltz65	721	2.9
zgesgpig	bi-waltz65–256pl	728	2.9
zgesfpgpig	waltz65	720	2.9
zgesfpgpig	bi-waltz65–256pl	711	2.9

^aS/N represents the average of three experiments. Sample: 525 μL of 41.1 mM NH₄Cl, 50 μL of 0.5 M H₂SO₄, 50 μL of DMSO-d₆, 25 μL of 12.5 mM maleic acid, and 25 μL of H₂O.

A 3 mm NMR tube has 28% of the sample volume of a 5 mm NMR tube. Note that the thin wall thickness of the glass tube is 0.38 mm, and the sample diameters of 3.0 and 5.0 mm tubes are 2.24 (3.0–2 × 0.38) and 4.24 (5.0–2 × 0.38) mm, respectively. Thus, the ratio of the NMR detection area is 2.24²/4.24² = 0.28. Therefore, the 72% loss of sensitivity using a 3 mm NMR tube compared with a 5 mm NMR tube is consistent with the amount of sample volume reduction.

Our results contradict the findings of Assafiri et al.²⁷ One possible explanation is that the radiation damping is not a significant issue with our 5 mm 400 MHz BBO probe, whereas Assafiri et al. used a 5 mm 600 MHz triple channel inverse (TCI) cryoprobe,²⁷ which has much higher sensitivity and may be more prone to radiation damping effects. Additionally, while our “zgesfpgpig” pulse sequence was designed to mitigate radiation damping and enhance S/N, it did not outperform the “zgesgpig” pulse sequence in our experiments (see Tables 1 and 2). This further supports the conclusion that radiation damping is not a significant concern with our 5 mm BBO probe setup.

Sample with Gd³⁺ in a 5 mm NMR Tube with a 5 mm BBO Probe

Kolen et al. reported that maleic acid is an effective internal standard for NH₄⁺ quantification, and Gd³⁺ is an excellent relaxation agent. By using 1 mM Gd³⁺, they were able to reduce ¹H T₁ of maleic acid from 2.1 to 0.13 s and NH₄⁺ from 2.2 to ∼0.14 s, respectively.²⁵ Consequently, the quantitative NMR acquisition interdelay (D₁ + AQ = 5 × T₁) can be shortened from ∼11.0 to ∼0.75 s with 1 mM Gd³⁺. This approach increased sensitivity by approximately 3.5-fold,²⁵ allowing Kolen et al. to reduce the NH₄⁺ NMR detection time by a factor of 3.5² = 14.

Inspired by this finding, we investigated the combined use of ¹⁴N decoupling and the Gd³⁺ relaxation agent to achieve a synergistic sensitivity enhancement. We adopted Kolen et al’s sample preparation method,²⁵ which included

525 μL of 56 μM NH₄Cl

50 μL of 0.5 M H₂SO₄

50 μL of DMSO-d₆

25 μL of 12.5 mM maleic acid

25 μL of 27 mM Gd³⁺

This mixture resulted in a final concentration of 43.6 μM NH₄⁺ and 1 mM Gd³⁺. We then conducted our experiments in a 5 mm NMR tube using a 5 mm 400 MHz Bruker BBO probe, with relaxation delay (D₁) of 10.34 s and acquisition time (AQ) of 0.64 s without Gd³⁺, and with D₁ = 0.11 s and AQ = 0.64 s with Gd³⁺. The Waltz65 decoupling was employed, as decoupling artifacts were not an issue with the weak ¹H signal at low NH₄⁺ concentration.

The results are shown in Figure 3, with the S/N values averaged from two experiments. Upon adding Gd³⁺, the sensitivity increased by a factor of 3.8 (6.4/1.7), consistent with the expected value derived from the reduction in acquisition time . Notably, our S/N values, both with and without Gd³⁺, were lower than those reported by Kolen et al, despite using the same 400 MHz NMR spectrometer.²⁵ This difference may be attributed to their use of a 5 mm Agilent OneNMR probe, which is optimized for ¹H detection and offers excellent salt tolerance.³³

Figure 3.

¹H NMR spectra of 43.6 μM NH₄⁺ (56 μM NH₄⁺ was diluted in NMR tube). The S/N values shown represent the average of two experiments. Bottom: without Gd³⁺ (D₁ = 10.34 s, AQ = 0.64 s, with zgesfpgp pulse sequence). Middle: with 1 mM Gd³⁺ (D₁ = 0.11 s, AQ = 0.64 s, with zgesfpgp pulse sequence). Top: with 1 mM Gd³⁺ (D₁ = 0.11s, AQ = 0.64 s, with zgesfpgpig pulse sequence and Waltz65 decoupling).

Additionally, the ¹⁴N decoupling provided a further 2.9-fold increase in sensitivity (S/N ratio of 18.6/6.4). The combined effect of Gd³⁺ and ¹⁴N decoupling resulted in a 10.9-fold increase in sensitivity (18.6/1.7). In other words, the NMR acquisition time can be reduced by a factor of 10.9² = 119, more than 2 orders of magnitude.

Kolen et al. proposed a simple and convenient method to calculate NH₄⁺ concentration by using internal reference maleic acid, with the following eq 1(25)

1

where I, N, and C are the integral area, number of nuclei, and concentration of NH₄⁺ and standard, respectively. However, many researchers still prefer to use a calibration curve to quantify NH₄⁺ concentration. To facilitate this, we prepared four samples with NH₄⁺ concentrations ranging from 7.5 to 18.9 μM, each containing 1 mM Gd³⁺, following Kolen et al’s sample preparation method.²⁵ We then acquired ¹H NMR spectra using the zgesfpgpig pulse sequence with Waltz65 decoupling. The NH₄⁺ concentrations were calculated using eq 1, and the resulting data were used to plot a calibration curve (Figure 4). The integral of the maleic acid was set to 100. Researchers can use this calibration curve (Figure 4) to quickly estimate the NH₄⁺ concentration in their samples.

Figure 4.

¹H NMR of 525 μL solutions of 7.5 to 18.9 μM NH₄Cl, 50 μL of 0.5 M H₂SO₄, 50 μL of DMSO-d₆, 25 μL of 12.5 mM maleic acid, and 25 μL of 27 mM Gd³⁺. NMR was carried out with zgesfpgpig pulse sequence and Waltz65 decoupling (D₁ = 0.11 s, AQ = 0.64 s).

To confirm the origin of the detected NH₄⁺, researchers sometimes use ¹⁵N₂ gas to generate ¹⁵NH₃ with their catalyst and process.¹⁷ To explore the effect of ¹⁵N decoupling, we briefly tested its performance using a 5 mm NMR tube with a 5 mm BBO probe. A 10 wt % of ¹⁵NH₄Cl sample was used to measure the 90° pulse, which was 19.35 μs at 70 W power on our spectrometer. The corresponding CPD and CW decoupling parameters were calculated with Bruker’s “edprosol” program.

A 16.7 mM ¹⁵NH₄⁺ sample was tested with and without decoupling, the results are shown in Figure 5. As expected, the ¹H NMR sensitivity gain of ¹⁵N decoupling was approximately 2-fold. The best performance was observed with the zgesfpgpig pulse sequence with Waltz65 decoupling method. Additionally, representative ¹H NMR spectra of a blend containing 8.3 mM ¹⁵NH₄⁺ and 8.3 mM ¹⁴NH₄⁺, both with and without ¹⁴N or ¹⁵N decoupling, are shown in Figure S2. As seen in Figure S2, ¹⁵N decoupling resulted in a singlet that overlapped with the central peak of the triplet caused by ¹⁴N heteronuclear coupling at the 400 MHz field strength (9.40 T). With an ultrahigh field 3-channel NMR spectrometer, it might be possible to simultaneously decouple ¹⁴N and ¹⁵N, allowing for the observation of two distinct ¹H NMR peaks for ¹⁴NH₄⁺ and ¹⁵NH₄⁺ in future studies.

Figure 5.

¹H NMR spectra of a sample of 500 μL of 18.7 mM ¹⁵NH₄Cl, 50 μL of DMSO-d₆, and 10 μL of 1 M H₂SO₄ with and without ¹⁵N decoupling. The S/N values shown are the average of two experiments. D₁ = 3 s, AQ = 1 s, dummy scans = 8 and number of scans = 16.

Finally, we explored the analysis with a 10 mm 400 MHz BBO probe. To begin, a sample of 10 wt % of NH₄Cl was used to measure the ¹⁴N 90° pulse, which was 36 μs at 99 W power on our spectrometer. The decoupling parameters were calculated with Bruker’s “edprosol” program. The experiments were conducted with a sample of a mixture of 2 mL of 18.7 mM NH₄Cl, 40 μL of 1 M H₂SO₄, and 200 μL of DMSO-d₆. The NMR parameters are D₁ = 3 s, AQ = 1 s, dummy scans = 16 and number of scans = 32. The bi-Waltz-256pl decoupling method was applied for ¹⁴N decoupling. The results are summarized in Table 3 (NMR spectra are shown in Figure S3). It was observed that “zgesfpgpig” pulse sequence performed better than the “zgesgpig” pulse sequence, potentially due to its capability to reduce the radiation damping effects. We plan to further investigate the radiation-damping effects in the future when a cryoprobe becomes available. To date, most studies have focused on using a 5 mm NMR probe.^7,9,17−27 However, it may be worthwhile to attempt detecting very low concentrations of NH₄⁺ with a 10 mm NMR probe using 1 mM Gd³⁺ and ¹⁴N decoupling, provided that good probe tuning/matching and efficient water suppression can be achieved.

Table 3. ¹H NMR Results of a Sample of 2 mL of 18.7 mM NH₄Cl, 40 μL of 1M H₂SO₄, and 200 μL of DMSO-d₆ with Different Pulse Sequences and bi-Waltz65-256pl Decoupling by Using a 10 mm BBO Probe.

NMR tube size	pluse sequence	decoupling method	S/N	S/N gain
10 mm	zgesgp	No	3347
10 mm	zgesgpig	bi-Waltz65–256pl	8974	2.7
10 mm	zgesfpgpig	bi-Waltz65–256pl	10,831	3.2

Conclusions

This study presents a significant advancement in the detection of aqueous ammonia using ¹H NMR, enhanced by the integration of ¹⁴N decoupling and the paramagnetic relaxation agent Gd³⁺. This innovative technique achieves a remarkable 10.9-fold increase in sensitivity, significantly reducing the NMR detection time by a factor of 119. The enhanced sensitivity facilitates the rapid detection of ammonia at micromolar concentrations, making it a valuable tool for sustainable ammonia synthesis research. It may also be used in ammonia analysis in industrial wastewater. Additionally, this improvement not only allows for the fast detection of ammonia but also reduces the environmental impact by minimizing the usage of NMR superconducting magnets, thereby lowering the consumption of liquid nitrogen and helium, which are associated with greenhouse gas emissions. The method’s ability to detect ammonia originating from the electroreduction of nitrogen using isotopically labeled ¹⁵N₂ gas further underscores its potential for applications in sustainable ammonia synthesis research. For analyzing the low concentration of ¹⁴NH₄⁺, it is recommended to use either the zgesgpig or zeesfpgpig pulse sequence with a Waltz65 scheme for ¹⁴N decoupling.

Supporting Information Available

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acsomega.5c00337.

• zgesgpig pulse sequence; zgesfpgpig pulse sequence; ¹H NMR spectra of a 32 mM NH₄⁺ solution acquired with different pulse sequences and Waltz65 decoupling, decoupling artifacts are shown (Figure S1); ¹H NMR spectra of a blend of 8.3 mM ¹⁵NH₄⁺ and 8.3 mM ¹⁴NH₄⁺ with and without decoupling (Figure S2); ¹H NMR spectra of a sample of 2 mL of 18.7 mM NH₄Cl, 40 μL 1 M H₂SO₄, and 200 μL of DMSO-d₆ acquired with different pulse sequences and bi-Waltz65–256pl decoupling by using a 10 mm BBO probe (Figure S3) (PDF)

The authors declare no competing financial interest.