Magic Angle Spinning Solid-State 13C Photochemically Induced Dynamic Nuclear Polarization by a Synthetic Donor–Chromophore–Acceptor System at 9.4 T

Solid-state photochemically induced dynamic nuclear polarization (photo-CIDNP) is a nuclear magnetic resonance spectroscopy technique in which nuclear spin hyperpolarization is generated upon optical irradiation of an appropriate donor–acceptor system. Until now, solid-state photo-CIDNP at high magnetic fields has been observed only in photosynthetic reaction centers and flavoproteins. In the present work, we show that the effect is not limited to such biomolecular samples, and solid-state 13C photo-CIDNP can be observed at 9.4 T under magic angle spinning using a frozen solution of a synthetic molecular system dissolved in an organic solvent. Signal enhancements for the source molecule larger than a factor of 2300 are obtained. In addition, we show that bulk 13C hyperpolarization of the solvent can be generated via spontaneous 13C–13C spin diffusion at natural abundance.

N uclear magnetic resonance (NMR) spectroscopy is an indispensable tool for chemical and structural analysis owing to its outstanding ability to capture changes in the local environment at the atomic scale.−9 These hyperpolarization methods in NMR spectroscopy boost the sensitivity of the technique by orders of magnitude, allowing complex analyses to be performed, which would otherwise not be possible. 3,10In the solid state, the most widespread NMR hyperpolarization protocol is dynamic nuclear polarization (DNP), 11,12 where, upon microwave irradiation, thermal electron spin polarization is transferred from a paramagnetic species to neighboring nuclei, typically 1 H spins, and then is relayed to the bulk of the sample by spontaneous spin diffusion. 13,14he signal enhancement (ε) provided by DNP is limited, among other factors, by the magnitude of the thermal electron spin polarization and cannot theoretically exceed a factor of 658 for 1 H nuclei or 2610 for 13 C nuclei (ε DNP max = |γ e /γ N |, where γ e and γ N are the electron and nuclear gyromagnetic ratios). 2 On the other hand, optically induced NMR hyperpolarization methods exploit transient excited states as the initial source of polarization.−25 In the solid state, photo-CIDNP is observed in donor−acceptor systems that undergo charge separation upon irradiation with light, generating a transient spincorrelated radical pair (SCRP). 26The SCRP is typically generated in the singlet state, |S⟩.At high fields, the SCRP coherently evolves between the singlet and the |T 0 ⟩ triplet state, potentially accumulating nuclear hyperpolarization during the process. 22,27,28−44 Recently, we reported the first example of direct 1 H photo-CIDNP in solids, at 0.3 T, using a synthetic donor−chromophore−acceptor (D−C−A) molecular system dissolved in a glassy frozen matrix. 45In this study, 1 H− 1 H spin diffusion was exploited to relay the polarization from the D− C−A system to the glassy matrix, leading to bulk uniform polarization of the sample.This was in contrast with most solid-state photo-CIDNP studies on photosensitive proteins, where the poor spin diffusion efficiency of 13 C and 15 N nuclei at natural abundance confines the polarization in the vicinity of the donor−acceptor system.
Here, we extend the concept of solid-state photo-CIDNP hyperpolarization by a D−C−A molecule to the high magnetic fields required for high-resolution NMR (here, 9.4 T).The advantages of small molecules over biomolecular systems, such as flavoproteins and photosynthetic reaction centers, are the easier tuneability of the photoactive machinery and broader solvent compatibility.However, to the best of our knowledge, no solid-state photo-CIDNP activity has been reported on D− C−A molecular systems at fields >1 T.
In the present work, we show that solid-state 13 C photo-CIDNP can occur in a D−C−A molecule in a frozen solution at 9.4 T, where D is benzobisdioxole aniline (BDX), C is 4aminonaphthalene-1,8-dicarboximide (ANI), and A is naphthalene-1,8:4,5-bis(dicarboximide) (NDI).We refer to this motif as CarboPol.The structure of CarboPol is shown in Figure 1, together with its photocycle.Upon absorption of a photon by the chromophore, an excited state D−C*−A is populated, which then undergoes an intramolecular electron transfer to form a charge-separated state.This charge-separated state is a singlet-born SCRP, indicated as 1 (D + • −C−A − • ) in Figure 1.The radical pair then evolves between that and the | T 0 ⟩ state, denoted as 3 (D + • −C−A − • ), and eventually decays to either the initial state or the neutral triplet state via charge recombination (CR).In the case of CarboPol, the neutral triplet that is created after CR in the triplet channel ( 3 CR) is localized on the acceptor and has the form D−C− 3 A. 46 CarboPol was synthesized as detailed in the Supporting Information.−56 It shares features with the previously introduced PhotoPol molecule that showed a 1 H photo-CIDNP response at 0.3 T. 45 CarboPol was designed to have a stronger electron− electron coupling as a result of the shorter donor−acceptor distance.
−60 13 C photo-CIDNP experiments were carried out at 9.4 T (100.6 MHz 13 C Larmor frequency) and cryogenic temperatures (100−220 K) using an Avance III HD Bruker NMR spectrometer and a commercial low-temperature magic angle spinning (MAS) DNP probe, in which a portion of the microwave waveguide was modified to allow for optical irradiation of the spinning sample (Figures S1−S3 of the Supporting Information).Samples were packed in a 3.2 mm sapphire rotor to allow for light penetration and sealed with a silicone plug.The MAS rate was 8 kHz in all of the experiments.Sample irradiation was achieved with a continuous wave 450 nm blue laser (the laser output power is 1.3 W, but the power delivered to the sample is likely to be significantly lower).In the laser-on experiments, the laser was on for the whole duration of the experiments.Experiments at 0.3 T were performed as described previously, 45 using a laser intensity of 4.2 W cm −2 .
First, we note that CarboPol shows solid-state 1 H photo-CIDNP activity at 0.3 T (Figure S13 of the Supporting Information), giving a bulk 1 H signal enhancement factor of ε 1H bulk = −27 at a 20 s recycle delay (ε = I on /I off ).This bulk enhancement factor is 1.7 times larger than that previously observed with PhotoPol under similar conditions.CarboPol has a larger electron−electron coupling compared to that of PhotoPol (d ≈ −60 MHz versus −5.5 MHz), as measured in butyronitrile at 85 K (details in the Supporting Information).This larger coupling could enable three-spin mixing at higher fields. 27,28igure 2 shows the direct excitation 13 C NMR spectra of a 1.5 mM frozen solution of CarboPol in OTP at 9.4 T and 100 K with (blue) and without (red) 450 nm laser irradiation, using a recycle delay of 1.8 s.In the absence of irradiation, only the OTP solvent signal is detected.With light irradiation, the 13 C signals of the CarboPol molecule can clearly be observed.Interestingly, all of the 13 C photo-CIDNP signals are negatively enhanced, irrespective of whether they belong to the donor or acceptor part of the photoactive machinery.A tentative assignment is also shown based on 13 C chemical shifts and solution-state NMR data (Figures S4−S7 of the Supporting Information).Note that the large signal from the silicone plug is due to its short 13 C T 1 compared to OTP, given the short experimental recycle delay.The reduction of the OTP signal at 126 ppm in the laser-on spectrum could be due to an overlap with negatively enhanced 13 C signals of the NDI acceptor (Figure S5 of the Supporting Information).
Because the CarboPol signal without irradiation is too weak to be observed at 1.5 mM with 30 000 scans (15 h), it is not possible to determine the enhancement factor directly.Using the noise level in the laser-off spectrum, we can obtain a lower bound for the magnitude of the 13   The Journal of Physical Chemistry Letters recycle delay (ε = I on /N off , with I and N being the signal intensity and the noise level).To obtain a higher sensitivity laser-off spectrum, we repeated the same laser-off experiment on two separate 15 mM CarboPol samples in OTP, i.e., 10 times more concentrated than the sample used to acquire the spectra in Figure 2. Further details are given in the Supporting Information.Even at 15 mM, no CarboPol aromatic signals were observed without laser irradiation after 30 000 scans (Figures S8−S10 of the Supporting Information).This led to an improved estimate of the enhancement factor of the signals at 138 ppm at a 1.8 s recycle delays of |ε 13C | > 2300.
Degradation of the 1.5 mM CarboPol sample was also investigated during a 10 h period of continuous 450 nm laser irradiation at 100 K. Figure 3 reports the normalized integrated intensity of the BDX donor aromatic signals at 138 ppm as a function of time.After 10 h, the peak area was reduced only by 3%, suggesting that degradation of the sample as a result of laser irradiation is very slow under MAS at this temperature.Therefore, long experiments on this and similar systems are, in principle, feasible, without significant loss in quality over time of the recorded NMR signal.
The effect of the temperature on the 13 C photo-CIDNP activity in CarboPol has also been investigated with a 1.8 s recycle delay within the 100−220 K range, below the glass transition temperature of OTP (243 K). 57 The magnitude of the 13 C photo-CIDNP-enhanced signals decreases with an increasing temperature, as shown in Figure 4.The effect is likely due to a shorter SCRP lifetime at higher temperatures, but other parameters (e.g., the electron−electron coupling d) might depend upon the temperature as well, affecting the photo-CIDNP activity.The increasing magnitude of the OTP signal at 126 ppm with the temperature is ascribed to either faster 13 C T 1 relaxation for OTP at higher temperatures or a reduction of the negative photo-CIDNP effect for the CarboPol resonances underneath the OTP signal (especially those from the carbons of the NDI acceptor; Figure S5 of the Supporting Information) or to a combination of the two effects.No sample degradation occurred during the acquisition of the spectra in Figure 4, as suggested by the full recovery of the photo-CIDNP effect when the sample was cooled down again to 100 K (Figure S11     The Journal of Physical Chemistry Letters lapping OTP signal.The characteristic build-up time of the photo-CIDNP-enhanced signal was estimated to be 1.8 s.The characteristic build-up time of OTP was not accurately determined here as a result of the long experimental time that would be required. From the build-up curve in Figure 5, it is evident that, for longer recycle delays, the overlapping OTP solvent signal also becomes negatively enhanced, as the overall integrated intensity of the signals at 138 ppm continues to decrease even at times longer than 5T b,fast .This can be seen directly from the other OTP peak at 126 ppm, which becomes negative in the laser-on spectrum with a recycle delay of 20 s (Figure S12 of the Supporting Information).This evidence demonstrates that the 13 C polarization propagates from the CarboPol molecules to the OTP solvent molecules in the bulk of the sample via spontaneous 13 C− 13 C spin diffusion, despite the very low efficiency of the process (all compounds here are at natural 13 C abundance).Indeed, 13 C− 13 C spin diffusion can still occur because of the very long 13 C T 1 of OTP, which is expected to be on the order of 1000 s at 100 K. Assuming that the spin diffusion coefficient (D) is proportional to γ N 2 c 1/3 , 62 where c is the concentration of the nuclear spins (0.86 M for 13 C in pure OTP), we can estimate its value for 13 C spins in bulk OTP to be D = 11 nm 2 s −1 by rescaling the measured spin diffusion coefficient of 19 62 Considering a 13 C T 1 of 1000 s, the 13 C spin diffusion length, defined as λ = DT 1 , is approximatively 100 nm.This analysis does not account for the influence of MAS on D, 63 which could significantly reduce the value of λ for dilute and low-γ spins because of the reduced homonuclear dipolar interaction.Nevertheless, to explore the extent of bulk 13 C polarization that can be generated at 100 K via 13 C− 13 C spin diffusion from the CarboPol molecules in this formulation, we measured two spectra with a long recycle delay of 2 h (Figure 6).Without light (red), conventional T 1 relaxation restores the OTP polarization to its thermal equilibrium, while in the presence of light (blue), the large negative 13 C polarization is relayed via spin diffusion from CarboPol, which now effectively acts as a polarizing agent, to the bulk of the sample.The 13 C bulk enhancement at a 2 h recycle delay under 8 kHz MAS is ε 13C bulk = −4.7 (ε = I on /I off ).Note that the signal from the silicone plug is not enhanced and, with this long recycle delay, is now dwarfed by the OTP signal.The large difference between the high local enhancements on CarboPol and the modest OTP bulk enhancement is ascribed to the very limited 13 C− 13 C spin diffusion efficiency. 64e now consider the photo-CIDNP mechanism at play here.The simplest Hamiltonian used to describe high-field photo-CIDNP of a single nuclear spin can be expressed as 28 where ω 1e and ω 2e are the electron spin Larmor frequencies of the SCRP, ω N is the nuclear spin Larmor frequency, d is the electron−electron interaction strength, and a and b are the secular and pseudo-secular components of the hyperfine interaction (coupling with only one electron of the SCRP is assumed).Solid-state photo-CIDNP generates nuclear hyperpolarization according to three different mechanisms: differential relaxation (DR), 30,65 differential decay (DD), 41 and three-spin mixing (TSM). 66,67These mechanisms are not mutually exclusive and could act simultaneously; 27,28

+
. 27,28,66−68 On the basis of the calculated g-factors of the BDX and NDI cationic and anionic radicals in CarboPol, 46 |Δω e | at 9.4 T is expected to be between 100 and 300 MHz for most molecular orientations.At the same time, |ω N | for 13 C nuclei at 9.4 T is 100.6 MHz.Therefore, for maximum efficiency, both DR and DD require hyperfine couplings on the 13 C sites in CarboPol on the order of a few hundred MHz, which is unrealistic for this system. 46On the other hand, the relative sizes of ω N and d (ω N = −100.6MHz and d ≈ − 60 MHz; details in the Supporting Information) suggest that TSM can still occur in a regime for which the matching condition is approximatively satisfied (|ω N | ≈ |d|).Indeed, we designed CarboPol to have a short donor−acceptor distance for which the electron−  The Journal of Physical Chemistry Letters electron coupling is dominated by a scalar interaction on the order of 100 MHz.The sign of the signal enhancement via the TSM mechanism only depends upon the sign of d and is expected to be negative for both the donor and acceptor sites, which is indeed the case.This also adds weight to the hypothesis that TSM dominates over DR and DD, as the acceptor signal would dominate for DR, while for DD, both positive and negative enhancements would be expected. 28,30,41,46,65n conclusion, we have shown that solid-state 13 C photo-CIDNP can be observed in synthetic donor−acceptor systems at a high magnetic field, if the electronic properties can be designed to match the photo-CIDNP conditions.The effect was proven to be active at 9.4 T and up to 220 K under MAS at 8 kHz, yielding 13 C enhancements in CarboPol of |ε 13C | > 2300 at 100 K. Owing to the long 13 C T 1 of the glassy OTP matrix, spontaneous 13 C− 13 C spin diffusion can relay the 13 C photo-CIDNP polarization far from the optically active molecules, resulting in bulk 13 C hyperpolarization, despite the low efficiency of spin diffusion for 13 C spins at natural abundance.−68 These results represent an important step toward generalized signal enhancements in high-field NMR spectroscopy by optical hyperpolarization using photo-CIDNP of tailorable donor−acceptor molecular systems.

■ ASSOCIATED CONTENT
C enhancement factor for the signals at 138 ppm (corresponding to four of the aromatic sites in the BDX donor) of |ε 13C | > |−340| = 340 at a 1.8 s

Figure 1 .
Figure 1.(a) Structure of the CarboPol molecule introduced in this work.The donor (BDX), chromophore (ANI), and acceptor (NDI) in the photoactive part of the structure are shown in red, yellow, and blue, respectively.The linker and end group are shown in black.(b) Photocycle of the D−C−A system in CarboPol (hν, photoexcitation; ISC, intersystem crossing; and CR, charge recombination).
of the Supporting Information).The polarization build-up with laser irradiation of the CarboPol signals at 138 ppm was characterized on the same sample (1.5 mM CarboPol in OTP) at 100 K and 8 kHz MAS.Results are shown in Figure 5. Data were fit to a biexponential function I(t) = I fast (1 − e −t/T b,fast ) + I slow (1 − e −t/T b,slow ), where the fast component corresponds to the BDX donor aromatic signals and the slow component corresponds to the over-

Figure 2 .
Figure2.13 C NMR spectra of a 1.5 mM frozen solution of CarboPol in OTP (9.4 T, 8 kHz MAS, and 100 K) without (red) and with (blue) continuous wave 450 nm laser irradiation.The recycle delay (τ) was set equal to the polarization build-up time under laser irradiation (T b = 1.8 s; data shown below), and an excitation pulse of 68°was used to maximize the signal-to-noise ratio per unit time in the laser-on spectrum (Ernst angle for τ = T b ).61Both spectra were acquired with 30 000 scans.The asterisks indicate spinning sidebands, while the daggers denote the silicone plug signals.The colored circles on the spectrum and on the molecular structure shown in the inset indicate the assignment of the photo-CIDNP-enhanced signals (the complete assignment in solution is shown in the Supporting Information).

Figure 3 .
Figure 3. Intensity of the photo-CIDNP effect in CarboPol as a function of the irradiation time with a continuous wave 450 nm laser.Each data point corresponds to the area of the BDX donor aromatic peak at 138 ppm in a direct excitation NMR experiment, each acquired over a 2 h period using an interscan delay of 1.8 s and an excitation pulse of 68°(3972 scans per data point).Data are normalized by the signal area in the spectrum after the first 2 h.Error bars are smaller than the symbol size.

Figure 4 .
Figure4.13 C NMR spectra of a 1.5 mM frozen solution of CarboPol in OTP (9.4 T, 8 kHz MAS) in the presence of 450 nm laser irradiation.Spectra were collected at the indicated temperatures from colder to warmer using a 1.8 s recycle delay, 2000 scans, and an excitation pulse of 68°.
however, the values of the spin interactions in CarboPol and the sign of the photo-CIDNP enhancement allow us to determine which mechanism dominates.DR is most efficient when |Δω e | = |ω 1e − ω 2e | = |a/2|, while DD is most efficient when |ω N | = |a/2|. 30,41TSM can occur under two different regimes: a first regime involving a double matching condition of the form |Δω e | = |ω N | = |a/2| and a second regime occurring when |ω N | = d a /4 2 2

Figure 5 .
Figure 5. Polarization build-up at 100 K in the presence of continuous wave 450 nm laser irradiation of the CarboPol aromatic signals at 138 ppm (from the BDX donor).The solid blue line shows the best fit to a biexponential function (see the main text for details).Error bars for each data point have a comparable size to the symbols and have been omitted.

Figure 6 .
Figure 6. 13 C NMR spectra of a 1.5 mM frozen solution of CarboPol in OTP (9.4 T, 8 kHz MAS, and 100 K) without (red) and with (blue) laser irradiation (2 h recycle delay and 90°excitation pulse).Both spectra were acquired with four scans.The asterisks indicate spinning sidebands, while the dagger denotes the silicone plug signal.