Reaction Monitoring Using SABRE-Hyperpolarized Benchtop (1 T) NMR Spectroscopy

The conversion of [IrCl(COD)(IMes)] (COD = cis,cis-1,5-cyclooctadiene, IMes = 1,3-bis(2,4,6-trimethyl-phenyl)imidazole-2-ylidene) in the presence of an excess of para-hydrogen (p-H2) and a substrate (4-aminopyridine (4-AP) or 4-methylpyridine (4-MP)) into [Ir(H)2(IMes)(substrate)3]Cl is monitored by 1H NMR spectroscopy using a benchtop (1 T) spectrometer in conjunction with the p-H2-based hyperpolarization technique signal amplification by reversible exchange (SABRE). A series of single-shot 1H NMR measurements are used to monitor the chemical changes that take place in solution through the lifetime of the hyperpolarized response. Non-hyperpolarized high-field 1H NMR control measurements were also undertaken to confirm that the observed time-dependent changes relate directly to the underlying chemical evolution. The formation of [Ir(H)2(IMes)(substrate)3]Cl is further linked to the hydrogen isotope exchange (HIE) reaction, which leads to the incorporation of deuterium into the ortho positions of 4-AP, where the source of deuterium is the solvent, methanol-d4. Comparable reaction monitoring results are achieved at both high-field (9.4 T) and low-field (1 T). It is notable that the low sensitivity of the benchtop (1 T) NMR enables the use of protio solvents, which when used here allows the effects of catalyst formation and substrate deuteration to be separated. Collectively, these methods illustrate how low-cost low-field NMR measurements provide unique insight into a complex catalytic process through a combination of hyperpolarization and relaxation data.


Quantification of substrate to catalyst ratios and the level of deuterium incorporation
The substrate to catalyst ratio was calculated from the ratio between the integrals of the peaks associated with appropriate proton resonances of the free substrate and the catalyst. In the cases where a mixture of iridium catalyst species were detected, the substrate integral was referenced to the sum of the proton resonances responsible for the same proton sites for the different bound forms of the substrate in the catalyst.
The deuteration percentage was calculated using Equation S1, where Sortho and Smeta are the integrated signal intensities for the ortho and meta resonances of the free substrate, respectively. In all cases, sufficient time was left between observations to allow for relaxation and accurate integration.

Evidence of the 4-aminopyridine (4-AP) ortho position deuteration
Figure S1 -Change in the relative integrals and line-shapes of the ortho and meta proton resonances of 4-AP as a function of reaction time in the regime where H2 (4 bar absolute) was added once at the beginning of the experiment. Time zero denotes the time at which the H2 gas was first introduced to the sample. The line-shape of the ortho resonance is relatively unchanged with time; however, the line-shape of the meta resonance tends to a singlet over the course of the experiment. This change in line-shape is indicative of the substitution of 1 H for 2 H in the adjacent ortho position. Figure S2 -13 C NMR SABRE signals from a partially deuterated sample of 4-aminopyridine (5 eq, 4-AP) in the presence of activated catalyst [Ir(H)2(IMes)(4-AP)3]Cl (4) in MeOD under 4 bar p-H2. The ortho carbon signal of free (blue circle) and bound (orange circle) 4-AP shows splitting from deuterium (J CD = 26.6 Hz) in both the 13 C and 13 C{ 1 H} SABRE NMR spectra. This splitting is removed in the 13 C{ 2 H} SABRE NMR spectrum. This confirms its origin from deuteration of the ortho position. The resonance for the meta carbon (purple and green triangles) signal appears in antiphase and consequently vanishes on proton decoupling in the corresponding 13 C{ 1 H} SABRE NMR spectrum; there is minimal change when a 13 C{ 2 H} SABRE NMR spectrum is recorded. Figure S3 -Change in the relative intensity and line-shape of the ortho and meta proton resonances of 4-AP as a function of reaction time in the regime where the p-H2 was refreshed between each scan. Analogous to Figure S1, the ortho resonance line shape loses some of the second order splitting, which is characteristic of progressive 2 H substitution at the second ortho position on the ring, but retains the doublet line-shape indicating that the meta resonance itself is not undergoing deuteration. The line-shape of the meta resonance evolves into a singlet over time which suggests that the ortho position has been significantly deuterated. We note that the line-shape of the meta resonance indicates that significant incorporation of deuterium in the ortho position has occurred after only 25.6 min under these reaction conditions (see Fig. S4). Time zero denotes the time at which the H2 gas was introduced to the sample. In black: deuteration level for the ortho position of the substrate 4-AP during the catalyst activation. The deuteration curve for 4-MP shows an initial fall that is not real, but instead reflects a decrease in the free meta peak integral due to partial overlap with a pre-catalyst resonance, a change in line-broadening also contributes to this.

Reaction monitoring with SABRE signal
A series of SABRE hyperpolarization experiments were performed using a 400 MHz 1 H NMR spectrometer for detection. Fresh parahydrogen was added to the head-space of the NMR tube and mixed into solution (via shaking) between each measurement to achieve SABRE hyperpolarization. This hydrogen refreshment step will change the observed rate of reaction for both the activation and the deuteration processes. To establish a suitable control experiment, a thermal 1 H NMR spectrum was recorded at each step of the experiment for comparison with the SABRE response. The evolution of the standard and SABRE-enhanced 1 H NMR signal intensities as a function of time, where p-H2 is refreshed between each data point, is presented in Figure S5a and S5b,c, respectively.
Here we use the meta proton resonance of 4-AP in free solution (δ 6.58 ppm) as the probe because no deuteration of this position is observed on the timescale of our experiments.
An exponential fit of the standard 1 H NMR response ( Figure S5a) indicates that the activation process was complete after 235 min. As expected, the SABRE-enhanced NMR response increases on the same timescale due to the increased concentration of the active SABRE catalyst (4). Interestingly, the SABRE signal continues to grow after the activation is complete. These results suggest that the SABRE-enhanced NMR signal is affected by both the activation and deuteration processes. However, differentiation of these effects is challenging (see paper). Similar trends were observed at both 9.4 T ( Figure S5b) and 1 T ( Figure S5c). Figure S5 -(a) Ratio of free substrate to catalyst as a function of reaction time, determined from standard 1 H NMR spectra acquired at 9.4 T in parallel with the SABRE measurements. The time to activation (235 min) was determined from a fit to a mono-exponential decay. Normalised raw SABRE-enhanced 1 H NMR signal intensity of the meta proton of 4-AP (5 eq relatively to 1) during the activation of SABRE complex in methanol-d4 as measured using detection at (b) 9.4 T and (c) 1 T. Figure S6 -Schematic representation of a single-shot hyperpolarized T1 measurement with a small flip angle sequence.

Variable flip angle single-shot sequence for relaxation time measurement
In the variable flip angle pulse sequence illustrated in Figure S6 a series of N FIDs is recorded where each FID is associated with an RF pulse with a unique flip angle, .
, and , are defined as the amplitude of the transverse and longitudinal magnetization vectors immediately following the nth RF pulse, respectively. The time between RF pulses is ∆ , where the subscript n denotes the delay between and +1 and ∆ 0 = 0 by definition. The initial magnetization along the z axis prior to the first RF pulse, 0 , is the initial amplitude of the hyperpolarization. If we consider only the decay of this hyperpolarization, i.e. neglecting the build-up of any thermal magnetization, we can write the following recurrence relationships to describe the amplitude of the transverse and longitudinal components of the magnetization following the nth RF pulse, where T1 is the hyperpolarization lifetime and ,0 = 0 : To directly monitor the decay of the magnetization with an optimal signal-to-noise ratio, we choose the variable flip angles according to the following two constraints: (1) In the absence of T1 relaxation, the amount of transverse magnetization resulting from each RF pulse is the same for all N RF pulses.
(2) As much of the magnetization along z is sampled during the experiment as possible. Therefore the angles are chosen such that , → 0 while not violating condition (1).
In the absence of T1 relaxation, Eqs. S2 and S3 simplify to Eqs. S4 and S5 and we define the tangent of the flip angle, , in terms of the ratio of the transverse and longitudinal magnetization (Eq. S6).
Substituting Eq. S6 into Eq. S4, we obtain the following: To fulfil condition (1), , = , −1 for all n in the absence of T1 relaxation. Therefore Eq. S7 simplifies to Eq. S8, which is defined for all N flip angles as long as | −1 | ≤ 4 . sin = tan −1 (S8) Condition (2) is met by determining the maximum value of 1 for which | −1 | ≤ 4 . Table S1 presents the set of angles that fulfil conditions (1) and (2) where N = 15. In the absence of T1 relaxation, 25.6% of the initial hyperpolarization, M0, is excited by each of the 15 variable flip angles.
In the presence of relaxation, the transverse magnetization created by each RF pulse is given by Eq. S9.
Substituting in Eq. S6, we get the following recurrence relationship for the transverse magnetization: where is defined in Eq. S11.

= ∑ ∆ −1 =1
(S11) Therefore, the signal observed using the variable flip angle sequence will have an initial amplitude of 0 sin 1 and will decay according to the hyperpolarization lifetime. It is important to note that this analysis does not model the effects of the variable flip angle pulse sequence on the recovery of the thermal magnetization during the experiment. These effects are removed by performing a reference measurement without hyperpolarization (see Figure S7), as described in the main text.  Figure S7 -The evolution of the hyperpolarized and thermal magnetisation in an example variable flip angle single-shot measurement for free 4-methylpyridine (4-MP) ortho proton (8.39 ppm) in the presence of [Ir(H)2(IMes)(4-MP)3]Cl in methanol-d4 at 298 K. The relaxation of the hyperpolarized signal with time is presented in red circles. Thermal polarization build-up measured in the reference scan is shown in grey squares. The resultant corrected hyperpolarization decay curve (the difference between the hyperpolarized and thermal reference measurements) is presented in blue triangles. Inset is a zoomed region of the graph illustrating the thermal polarization increase with time in the reference scan. Figure S8 presents a comparison of the variable flip angle pulse sequence (red) and a comparable measurement using a constant flip angle sequence with = 5° (black). While the measured relaxation time is the same for both sequences, the use of the variable flip angle sequence improves the SNR by more than a factor of 3 and hence the accuracy of the final T1. The standard error for 5 repeated variable flip angle single-shot T1 measurements was less than 3% for the samples of 25 mM 4-AP in MeOH in the presence of the 1 mM, 2 mM, 3 mM and 4 mM activated SABRE catalyst and less than 6% for the samples of 25 mM 4-AP in MeOH in the presence of the 0.5 mM and 5 mM activated SABRE catalyst measured on the 1 T benchtop NMR spectrometer.

Activation monitoring with 4-methylpyridine using hyperpolarization lifetimes
The formation of the SABRE catalyst [Ir(H)2(IMes)(4-MP)3]Cl in the mixture of 1 and [Ir(COD)(IMes)(4-MP)]Cl in the presence of a 4-fold excess (in starting solution) of 4-methylpyridine (4-MP) was monitored using the hyperpolarization lifetime measurement procedure with benchtop (1 T) NMR detection. Figure S9 shows the change in the ortho (green square) and meta (gold triangles) 1 H hyperpolarization lifetimes of 4-MP as a function of reaction time after the initial addition of p-H2 in methanol-d4 ( Figure S9a) and protio methanol ( Figure S9b). Figure S9 indicates a much faster rate of activation for 4-MP compared to 4-AP (Figure 3 in main text). This is due to the different electronic properties of 4-MP and 4-AP. Furthermore, the hyperpolarization lifetime behaviour for 4-MP is the same in the two solvents. This indicates that there is no significant deuteration of 4-MP on the timescale of this experiment. Control experiments of 1 H NMR line-shapes were acquired at 400 MHz to confirm the absence of significant deuteration on this timescale in the regime when p-H2 is refreshed between the measurements (see Figure S10).  The initial solution was prepared with 9 eq of 4-MP relative to 1. Eleven p-H2 refreshments were completed over 89 min. The 1 H NMR spectra presented were acquired (bottom) before p-H2 addition, (middle) 39 min and (top) 89 min after the first p-H2 addition. No significant change was observed in the line-shape of the ortho and meta peaks of 4-MP, indicating that no significant deuteration occurs on this timescale. Time zero denotes the time at which the p-H2 gas was introduced into the sample.