Introduction

Solid-state nuclear magnetic resonance (SS-NMR) spectroscopy, with atomic resolution, permits probing the structure and dynamics of biological and material systems^[1-7]. However, the lacks of sensitivity and resolution are often the two major limitations for the investigation of complex solid systems. Both limitations are attenuated by using high static magnetic field B₀ and fast magic angle spinning (MAS) frequencies to remove the anisotropic parts of interactions. Through-space homo-nuclear and hetero-nuclear correlations are widely used under MAS for resolution purpose to probe the structures. However, the dipolar couplings have to be restored under MAS. Among various homo-nuclear recoupling (HOMCOR) methods, spin diffusion is widely used since it is little sensitive to dipolar truncation and can probe long distance correlations. For ¹³C-¹³C experiments, proton-driven spin-diffusion (PDSD)^[1], dipolar-assisted rotational resonance (DARR)^{[8, 9]}, phase-alternated recoupling irradiation schemes with xy phases (PARIS_xy)^[10-12], second-order Hamiltonian among analogous nuclei generated by hetero-nuclear assistance irradiation (SHANGHAI)^[13], and second-order Hamiltonian among analogous nuclei plus (SHA+)^[14] have been demonstrated. All these sequences are second-order recoupling methods. In PDSD [Fig. 1(b)] experiments, no recoupling pulse is sent during the mixing time (τ_mix). While in SHA+ experiments [Fig. 1(c)], recoupling pulses with different lengths and/or phases are sent. It has been shown that by using recoupling pulses in the mixing time, the signal transfer between ¹³C nuclei is more efficient than that without any pulse.

However, for ¹⁵N-¹⁵N experiments, PDSD is usually preferred since the dipolar couplings between ¹⁵N nuclei are always very small, hence leading to very long (≈ 1 s) mixing times, which make most of the second-order methods with recoupling pulses inefficient. In this communication, using g-C₃N₄ which is a material used in photocatalysis as a sample, we demonstrate that the SHA+ recoupling pulse sequence can be used for ¹⁵N HOMCOR and it is much more efficient than PDSD. The SHA+ scheme has been derived on the basis of the super-cycling of SHANGHAI sequence using Floquet theory^{[15, 16]}. It concatenates ¹H irradiation with three different pulse lengths:T_R/2, T_R and 2T_R.T_R=1/v_R, is the rotor period, and v_R is the spinning frequency. The super-cycles will last for 64 T_R. The use of different pulse lengths and phases allows increasing the number of modulation-sidebands and hence the number of efficient recoupling conditions. Our results show that SHA+ pulse sequence can be useful for the structural analysis of different kinds of nitrogen-doped carbon materials, such as nitrogen-doped graphene.

1 Experimental

Experiments have been performed on a Bruker AVANCE-Ⅲ 600 MHz spectrometer (B₀=14.1 T). A commercial Bruker HX double-resonance MAS probe with rotor of 4 mm outer diameter was used, and spinning frequency v_R was set to 10 kHz for all experiments. In this case, the same use of probe, sample volume and mixing time allows an easy quantitative comparison of the transfer efficiencies. The temperature is controlled by cooling gas.

For increasing the NMR detection sensitivity of the ¹⁵N nucleus, a ¹⁵N-enriched g-C₃N₄ (enrichment of ca. 20%) was synthesized starting from ¹⁵N-labeled melamine, which was prepared according to the procedure introduced by Jurgens et al^[17].

The simulations were performed using SIMPSON V4.1.1 software^[18]. The powder averaging was accomplished using 1 680 crystallite orientations. The 168 (α_MR, β_MR)-pairs, which relate the molecular and rotor frames, were selected according to the REPULSION algorithm^[19], and the 10 g_MR-angles were regularly spaced between 0°~360°.

2 Results and discussion

Fig. 2 shows the experimental 2D SHA+ [Fig. 2(a)] and PDSD [Fig. 2(b)] ¹⁵N-¹⁵N correlation spectra of g-C₃N₄sample displaying at the same counter levels. The ¹⁵N signals were referenced with respect to nitromethane. Typical 90°pulse lengths of ¹H and ¹⁵N were 4 ms and 9 ms, respectively. The contact time for cross-polarization was 1 ms, and SPINAL-64 proton dipolar decoupling sequence was used during evolution and detection periods^[20]. Recycle delays were set to 2 s for all experiments. For each t₁ step, 96 scans were accumulated. τ_mix was set to 0.8 s or 1.6 s. The ¹H radio frequecy (rf) power of the SHA+ sequence was set to 10 kHz. Additional experimental details are given in the figure captions.

Resonances have been assigned according as the literature^[21]. The signals of δ_N-260~-280 were assigned as protonated nitrogen sites of NH₂ and NH (referred as N_H in Fig. 3). The signals of δ_N-190~-210 were assigned as tertiary nitrogen N_tert from N-C(NH) or N-C(NH₂)^[21].And the signals of δ_N-230~-240 were assigned as the central nitrogen atom (N_c) of the heptazine core^[21].

It is obvious from the Fig. 2(a) and Fig. 2(b) that with τ_mix=0.8 s, PDSD is not able to probe any correlations between nitrogen nuclei, while the SHA+ identifies a cross-peak from the N_H to the N_tert. There is no cross-peak from the N_tert to the N_H, which is due to the short spin-lattice relaxation time (T₁) of N_tert. It is obvious from Fig. 3 that the N_H atoms are close to the N_tert, with a distance of r ≈ 0.22 nm corresponding to a dipolar coupling of |d/2π| ≈ 115 Hz. And the existence of protons in the N_H groups (NH or NH₂) accelerates the magnetization transfer from the N_H to the N_tert. On the contrary, the N_C is far away from the N_H groups (r ≈ 0.38 nm, |d/2π| ≈ 22 Hz), and therefore SHA+ cannot probe this correlation. Furthermore, both SHA+ and PDSD sequences can not probe the correlation between the N_tert and the N_C, in spite of their short distances (r ≈ 0.22 nm, |d/2π| ≈ 115 Hz) due to the absence of the protons between them. These results demonstrate the importance of protons in the second-order recoupling methods.

To further compare the transfer efficiency of SHA+ and PDSD in this experiment, we show in Fig. 2(c) the row slices taken at δ_N-270 for these two approaches with τ_mix=0.8 s and τ_mix=1.6 s. It is noticed that the cross-peak is more intense with SHA+ recoupling scheme. Moreover, the comparison of the slices proves that SHA+ accelerates the magnetization transfer between different ¹⁵N sites by a factor of ca. 2. Indeed, the same cross-peak intensity is observed in SHA+ with τ_mix=0.8 s as in PDSD with τ_mix=1.6 s. It should be pointed out that we use only 10 kHz ¹H rf power for SHA+ recoupling at v_R=10 kHz and the irradiation time set to 1.6 s on ¹H channel which would not damage the ¹H amplifier.

To compare with our experimental results, we have simulated the transfer efficiency versus τ_mix for SHA+ and PDSD sequences. Simulations were performed on a N₁H-N₂ spin-system, where the inter-nuclear distances are r(N₁-N₂)=0.22 nm, r(N₁-H)=0.105 nm, and r(N₂-H)=0.311 nm. The chemical shift anisotropies (CSAs) of ¹⁵N and ¹H nuclei were 0, and the isotropic chemical shifts of N₁ and N₂ were±2.4 kHz. This configuration is similar with the configuration of ¹⁵N species in the g-C₃N₄ samples. It is obvious from Fig. 4 that SHA+ is much more efficient than PDSD.

It should be mentioned that the simulation results are different from the experimental results, especially the simulation of the PDSD, since (1) the simulations were done for three-nuclei system of N₁H-N₂, while the real systems have more protons; (2) the effect of the relaxation has not been considered in the simulations; (3) the simulations do not consider the longitudinal spin exchange phenomenon has not been considered in the simulations^[22]. Without the longitudinal spin exchange phenomenon, the simulation results for PDSD are always 0. It should be also pointed out that although the intensity of the SHA+ simulation with N₁H-N₂system reaches the maximum at τ_mix=0.6 ms, in the real system the efficiency of SHA+ increases with the increase in the recoupling time as demonstrated in Fig. 2, since the real system has more protons, among which some are far away from ¹⁵N while some are close to ¹⁵N.

3 Conclusion

We have proposed a broadband spin diffusion experiment assisted by ¹H irradiation—SHA+, which allows more efficient ¹⁵N-¹⁵N correlations than PDSD at high magnetic fields and moderate MAS frequencies. This method is little sensitive to dipolar truncation, allowing polarization transfer between ¹⁵N atoms with one nitrogen bonded to protons. Our investigations should become valuable for SS-NMR experiments in material science, especially it can be applied to nitrogen-doped carbon-based materials.

Acknowledgement: Authors would like to acknowledge Large Instruments Open Foundation of East China Normal University, National Natural Science Foundation of China (21373086), National Science Fund of China for Excellent Young Scholars (21522303), and Basic Research Project of Shanghai Science and Technology Committee (14JC1491000).