A Simple Route to Strong Carbon‐13 NMR Signals Detectable for Several Minutes

Abstract Nuclear magnetic resonance (NMR) and magnetic resonance imaging (MRI) suffer from low sensitivity and limited nuclear spin memory lifetimes. Although hyperpolarization techniques increase sensitivity, there is also a desire to increase relaxation times to expand the range of applications addressable by these methods. Here, we demonstrate a route to create hyperpolarized magnetization in 13C nuclear spin pairs that last much longer than normal lifetimes by storage in a singlet state. By combining molecular design and low‐field storage with para‐hydrogen derived hyperpolarization, we achieve more than three orders of signal amplification relative to equilibrium Zeeman polarization and an order of magnitude extension in state lifetime. These studies use a range of specifically synthesized pyridazine derivatives and dimethyl p‐tolyl phenyl pyridazine is the most successful, achieving a lifetime of about 190 s in low‐field, which leads to a 13C‐signal that is visible for 10 minutes.

S2 specified number of 90˚pulses with sufficient (at least 5 times T1) relaxation delay between each scans. All NMR experimental parameters, except the number of scans, remain the same for hyperpolarized and thermal measurements.
Singlet lifetime (TS) measurements were evaluated by varying the storage times (߬) in a series of M2S-S2M experiments. [3] Integrated signal intensities were fitted to the mono-exponential equation: ‫ܯ‬ ௭ (߬) = ‫ܯ‬ ݁ ି ఛ ் ೄ ൗ to produce the TS value. Here Mz(߬) represents the integrated signal amplitudes at time ߬, while M0 is a fitting constant.

Solvent:
All the experiments were done in deuterated methanol unless otherwise stated.
Sample preparation: Samples were prepared with the following specifications: Sample prepared for 'shake & drop' method: 2 mg (3.12 μmol) of [IrCl(COD)(IMes)] precatalyst with a substrate loading that leads to a one-fold excess of added agent after formation of the active catalyst in 0.6 mL of solvent. (ii) Sample prepared for 'flow' method: 10 mg (15.6 μmol) of [IrCl(COD)(IMes)] precatalyst with substrate loading that leads to a one-fold excess of agent after formation of the active catalyst in 3 mL of solvent.

S2. Catalyst
In this study, we used [IrCl(COD)(IMes)] as the catalyst precursor in all cases unless otherwise stated. It was synthesized in the laboratory following the standard literature procedure. [4] Here IMes stands for 1,3-bis(2,4,6-trimethylphenyl)imidazole-2-ylidene and COD = cis,cis-1,5-cyclooctadiene. A typical reaction of the catalyst with the substrate ligand (Sub) and p-H2 is shown below. The substrate binds to the metal through a nitrogen center as described below. In all cases reported in this study, the complex produces a single hydride resonance at δ -21.2 ± 0.4 ppm in the corresponding 1 H NMR spectrum as per the reaction shown in Figure S1. The process for pyridazine, and a series of its related forms have been extensively studied previously. [5][6] S3. Synthetic Methods

S3.1 General
Distilled water was employed where detailed. Brine refers to a saturated aqueous solution of NaCl. THF was freshly distilled from sodium and benzophenone ketyl or dried using a Grubbs solvent purification system. All reactions were carried out under O2-free N2 unless otherwise stated.
Flash column chromatography was carried out using Fluka Chemie GmbH silica (220-440 mesh). Thin layer chromatography was carried out using Merck F254 aluminium-backed silica plates. 1 H (400 MHz) and 13 C (100.6 MHz) NMR spectra were recorded on a Bruker-400 instrument with an internal deuterium lock. Chemical shifts are quoted as parts per million and referenced to CHCl3 (H 7.27), CDCl3 (C 77.0). 13 C NMR spectra were recorded with broadband proton decoupling. 13 C NMR spectra were assigned using DEPT experiments when necessary. Coupling constants (J) are quoted in Hertz. Electrospray high and low resolution mass spectra were recorded on a Bruker Daltronics microOTOF spectrometer.

S4. Theory and simulations
In this work, we use standard density matrix based numerical approaches to study the SABRE process. The complete method has been extensively presented in earlier works. [7][8] The whole catalytic process can be thought of as a three-step time dependent evolution process as described in the figure below.
Step 1: evolution of the full spin system including p-H2 at a defined mixing field; Step 2: evolution of the substrate spins after its dissociation from the catalyst; Step 3: evolution of substrate spins during time dependent magnetic field transfer while transporting the sample in to the spectrometer.

Figure S3. Magnetic field variance during the SABRE process. E1, E2, and E3 represent three evolution periods during time intervals (t0-td), (td-tf), and (tf-tm) respectively. Bmix and Bmeas represent the mixing (transfer) and measurement fields respectively.
After going through the complete procedure with common approximations, the final density matrix that results for a 2spin substrate in the measurement field can be written as follows: where ܴ ௭ and ܶ ௭ represent the longitudinal magnetization terms of two spins (R and T) in the substrate. The ܴ ௭ ܶ ௭ term represents a double quantum term. Apart from these three terms, there will be several cross and higher order terms but their contributions are negligible in most of the cases and hence have been neglected in this study. A, B, and C are constant terms and their values actually reflect the enhancement factor over the thermal signal.
Appropriate routines in Mathematica were used to perform the calculations. The input parameters used in these calculations are taken from the related experimental data that is summarized in Table S1. Figure S3-S5 show simulations of SABRE derived magnetization at the indicated mixing field strength for two types of substrates. Despite this simplified treatment, the experimental data ( Fig. S4-S6) shows good agreement with the results of simulation.   Table S1. The complexity of its spin dynamics as a function of mixing field is reflected in Figures S4-S6. The terms IzRz, IzTz, SzTz, SzRz, Rz and Tz will lead to observable signal when a single 90 o pulse is applied at 13 C for detection as shown in Figure S4. However, if we apply simultaneous 90 o pulses to both nuclei on the 1 H and 13 C channels for detection, then we only observe the terms RZ and TZ terms whilst filtering out the other terms (a de facto double-quantum filter). Figure S5 illustrates the success of this approach of filtration. It should be noted that in this study we primarily focused on polarizing single quantum coherences in order to apply the subsequent LLS sequence, hence all other cross terms have been filtered out during the process.     1 H and 13 C channels), as a function of mixing field and (b) corresponding simulated plot (sum of IZ and SZ terms) derived by the theoretical approach described above that will lead to observable signal. A maximum signal enhancement was observed for a 65 Gauss mixing field.
Agents 4-8, will polarize via the direct mode of transfer as there are no suitable relay mechanisms on offer. This scenario significantly reduces the complexity of the spin dynamics. Almost the entire contribution to the observed signal now results from the single quantum RZ, TZ terms with very little of the double quantum term RZTZ. The experimental field transfer plot closely matches with the theoretical simulations as shown in Figure S8. Figure S8. (a) 13 C NMR SABRE derived response after a 90˚pulse portrayed as the experimentally derived enhancement factor based on relaxed magnetization for 5 as a function of magnetic mixing field (Gauss) and (b) the corresponding simulated plot derived by the theoretical approach described above for the indicated terms produced in the mixing field. The terms Rz and Tz will lead to observable signal.
The resonance condition for the direct mode of polarization transfer can also be verified through calculation as outlined by Theis et al. in their recent paper. [9] ‫ܤ‬ ோ௦ ≈ where JHH and JCH are the hydride J-coupling constant and hydride-to-13 C coupling constant in Hz respectively. ߛH and ߛC represent gyromagnetic ratios of 1 H and 13 C nuclei respectively. Using the equation above, the theoretical estimation for optimum polarization transfer to the 13 C nuclei is these samples are found to be ±0.3 ߤT (3 mG). This calculation, however, does not include several other important factors (e.g. residence time, exchange rates) required to make it precise. Experimentally, we found maxima in the range of 1 mG -20 mG in all the cases.

S5. Long-lived singlet states (LLS)
A pair of coupled spin-1/2 particles can be defined by the total spin of the system. This can takes value of 1, in which case there are three possible quantum state degeneracies, commonly known as triplet states (analogous to those found in the ortho forms of dihydrogen molecule). The other possibility is that the total spin is 0, representing a non-magnetic state termed a singlet state, analogous to the para isomer of dihydrogen. In a coupled spin-1/2 system where both spins are chemically equivalent (e.g. H2 molecule), the singlet and triplets forms correspond to exact Eigen states of the spin Hamiltonian and formation of the singlet state occurs naturally. However, it is not possible to access the singlet state (spin 0 = non-magnetic) without performing a symmetry breaking process. In 2004, it was first shown by Levitt and coworkers that it's possible to create similar singlet states in chemically inequivalent spin-1/2 pair via suitable radio frequency (rf) pulse sequencing. Since the singlet can be created experimentally, it can also be retrieved on demand via similar rf pulse sequencing. The importance of singlet state in NMR is huge considering the fact that it's unaffected by S21 one of the major relaxation mechanism. Consequently, a singlet state lifetime can be much longer than the standard nuclear spin lifetime that is dictated by T1/T2 relaxation time constants.
In this study, we used singlet state as the polarization bank. At first we create accessible hyperpolarized spin-order that is then converted into singlet order to preserve its lifetime. When required, the stored hyper-singlet states are converted back into an observable mode for detection. We used standard rf pulse sequencing protocols to create the hyper-singlet states as described below.
The M2S-S2M pulse sequence is among the most elegant to study long-lived singlet states (LLS), shown in Figure S9.
Here M2S stands for 'magnetization-to-singlet' which converts magnetization into singlet states in an adiabatic fashion by applying a number of ߨ pulses spaced over an optimized duration of time. After M2S, a variable waiting times were applied (߬delay) to measure the singlet state decay time constant (TS). During the waiting period a reverse order pulse S2M ('singlet-to-magnetization') was applied to detect the signal. A T00 filter was also used in the pulse sequence to eliminate any possible magnetization that may have accumulated over the long storage times. Figure S9: Pulse sequence to study long-lived singlet states.
In order to efficiently run the LLS pulse sequence, it is required to evaluate the exact parameters ߬, n1 and n2 of Figure  S9. A series of J-synchronized experiments along with Spindynamica simulations were employed to calculate those values accurately in each sample.

J-synchronization experiment
The pulse sequence for J-synchronization experiment is shown Figure S10. It consists with a series of 180˚pulses at equal intervals. By varying the delay (߬) in between those pulses or by varying the number of 180˚pulses (n1), we can accurately calculate the optimized parameters required for M2S-S2M pulse sequence. For chemically equivalent but magnetically inequivalent spin pair (Type-1), the theoretical values can be estimated using the following formulas, Where, ‫ܬ߂‬ ு is the difference between two out of pair heteronuclear J-couplings in the spin system 1.
For Type-2 spin system, the theoretically estimated values are calculated as follows, Where ߂ߜ, ߥ and J represent chemical shift difference, 13 C Larmor frequency and scalar coupling constant between the 13 C spin pair respectively.
In all cases, Spindynamica simulation [10] closely matches the experimental findings as shown below.       Table S2. Singlet lifetime constants (TLLS) were measured by employing the complete sequence of M2S-S2M and lead to the results shown in Table S3. It should be noted that for agent 2, it was not possible to extract any information by running J-synchronization due to the extremely large number of refocusing 180˚pulses required that resulted in a very long duration pulse sequence and consequently no signal. Also fast relaxation of deuterium plays a part. In the case of agent 3, no symmetry breaking mechanism is available due to the spin-pair's magnetically symmetric environment and hence no singlets can be populated by the means of rf pulse sequencing. Agent 8 represents a weakly coupled (J ≈ Δߥ) scenario at 9.4 T and consequently low singlet lifetime.  [11] .

S7. NMR spectral data
The following examples of NMR spectra reflect the substrate spin systems studied here. Samples were prepared as described in section S1. All spectra were measured in a 400 MHz NMR spectrometer at room temperature.