Principles of the NMRtorch design
In our proposed approach to illuminating NMR samples in situ, a lighthead (Figure 1) containing one or more LEDs is attached directly to the top of an NMRtorch tube, allowing light to enter the walls of the glass tube through the rim. If used, the tube cap should be sufficiently optically clear (Figure 1A). Optional optical elements, such as lenses or condensers, may be used to further focus light, although we found these to be of little benefit given the close proximity of the LED emitter to the top rim of the tube. The NMR tube itself acts as a light guide, with light confined within the optically-transparent wall of the tube, and propagated towards the sample area (standard borosilicate or quartz 5 mm tubes with ~1.4 mm wall thickness work well in our experience). In NMRtorch tubes, the glass around the sample volume at the bottom of the tube is modified by the introduction of light scattering centres (achieved here by etching of the exterior surface). When the LEDs are illuminated, light is scattered here, thus preferentially illuminating the sample area from the outside (Figure 1A).
The lighthead is connected by a multicore electrical cable to an electrical control box containing constant-current power supplies, with their triggers in turn connected to the NMR console controlled by a computer (Figure 1B). Furthermore, in the NMRtorch setup described here, four separate channels connect to the NMR console, allowing the use of multi-channel LED arrays, to maximize light output at a single wavelength, or to provide illumination at different wavelengths. In the current implementation, the NMR tube and attached lighthead are manually lowered into the spectrometer bore by the multicore electrical cable. Gas flow that is typically present in the magnet bore provides sufficient cooling for LEDs with lower power consumption (≤3W) or under operation at lower duty cycles. However, for more powerful LED arrays at constant illumination, a supplementary compressed gas line to the lighthead provides additional cooling, with temperature monitored by an optional temperature sensor in the lighthead. The proposed arrangement enables convenient delivery of high-intensity light (Figure 1C) to the sample area achieving high uniformity (Figure 1D), while essentially allowing the user to handle the NMR sample tube as normal, with easy tube filling and capping, avoiding any inserts or additional sources of magnetic field inhomogeneity in the sample area. The etchings on the outside surface of NMRtorch tubes, in our experience, do not affect NMR tuning, shimming, or water suppression, and no significant effects on the signal lineshape were observed. Furthermore, as all electrical components in the lighthead are positioned sufficiently far away from the NMR detection region, we have not noticed any signs of electromagnetic interference caused by LED switching.
Using NMRtorch For Photo-CIDNP Experiments
Photo-chemically induced dynamic nuclear polarization (photo-CIDNP), where observed NMR signal intensities of aromatic compounds in the presence of a photosensitiser are modulated upon illumination, has been previously demonstrated using illumination with LASERs,18,20,21,36 and more recently, with LEDs.13,31,32 Here, we explored the suitability and effectiveness of NMRtorch for the photo-CIDNP experiments, by recording 19F NMR spectra of 6-fluoroindole (6FI) with photosensitization by flavin mononucleotide (FMN) under blue light illumination. With the NMRtorch setup, 6FI exhibited emissive photo-CIDNP, with negative 19F peaks upon illumination and a maximum 64-fold enhancement achieved with the lighthead containing a single LED with nominal 470 nm peak emission, 3W power consumption (Figure 2A). To our knowledge, this enhancement is the largest ever reported for 19F, which is remarkable given that no attempts were made here to remove oxygen from the samples, or introduce any other measures to prevent dye quenching. Convenient working with multiple samples allowed us to explore the concentration-dependence of photo-CIDNP effects. At 0.2 mM FMN, the highest photo-CIDNP enhancements (α, Equation 1) were observed at 1 mM 6FI, with lower enhancements observed at both higher and lower concentrations (Figure 2B). As the highest absolute signal intensity for illuminated samples was observed at 6 mM 6FI (Figure 2B), this concentration was chosen for further NMR imaging experiments where absolute signal intensity is critical. Photo-CIDNP enhancements were also observed to increase with illumination time, yet began to plateau at values above 6 s (Figure 2C). We found that the easy process of sample tube filling, capping and attaching to the NMRtorch lighthead enabled multiple photo-CIDNP samples to be run efficiently and consistently.
As light intensity is an important experimental parameter, it needs to be controlled and measured in the sample area. LED brightness can be routinely dimmed by controlling duty cycle through pulse width modulation (PWM). Here, we incorporated this PWM control into the NMR pulse sequence, such that light intensity (between 0 and 100% duty cycle) before RF pulses was controlled directly by the NMR console (see Supplemental Appendix S1). Both the 19F NMR photo-CIDNP effect for 6FI, and the light intensity measured ex situ on the surface of the sample area, were observed to be linear with LED duty cycle (Figure 2D). This linearity of 6FI photo-CIDNP effect with increasing light intensity means this parameter can be used to assess the uniformity of light distribution across the length of the NMR sample in ‘NMR imaging’ experiments. Although photo-CIDNP has previously been used to assess the effectiveness of sample illumination approaches29 and light distribution uniformity,28 here, to account for non-linearity of magnetic field gradients et the edges of NMR-active volume, we propose to use position-dependant photo-CIDNP enhancement factors, rather than a raw photo-CIDNP spectra 28 obtained in imaging experiments. As the photo-CIDNP signal intensities and enhancement factors are directly proportional to the light intensity, the profiles formed by position-dependent enhancement factors (\({\alpha }_{Z}\), defined by Equation 3) report directly on light intensity distribution inside the sample, along Z-axis.
In the NMRtorch setup, the positioning of light scattering centres (created here by etchings on the outside surface of the tube) controls light distribution and intensity (Figure 3). Excessive etchings at the top of the sample may cause the majority of light to be scattered at the top, with less light reaching the bottom of the sample area. We therefore propose that etchings should be distributed non-uniformly across the sample length, to achieve more uniform light distribution. To demonstrate the effect of such positioning of etchings, light distribution in an illustrative set of NMRtorch tubes was assessed by both pixel brightness analysis of photo images, and photo-CIDNP position-dependent enhancement factors \({\alpha }_{Z}\).
Here, smoothed pixel brightness characterisations of light distribution (Figure 3B, with ± 2.5 mm moving averages) were in good agreement with the observed photo-CIDNP position-dependent enhancements across the NMR-active volume (Figure 3C). In the control unetched Tube 1, scattering was primarily observed at the sample meniscus and tube bottom, resulting in poor sample illumination and weak photo-CIDNP effects. Conversely, in sample tubes with various etched patterns (Tubes 2 – 4), light scattering results in greater sample illumination and thus greater position-dependent signal enhancement (\({\alpha }_{Z}\)). This distribution of etchings can be used to guide light distribution in the sample: for example, in Tube 2, the etched bottom portion experiences greater sample illumination than the unetched top portion of the sample region, whereas adding etchings in this area, in Tubes 3 and 4, increased illumination in the upper portion. Additionally, the degree of etching can be used to control overall illumination intensity, with the more frequent etchings on Tube 4 resulting in greater light intensity (averaging 260.9 ± 44.9 µmol m−2 s−1 across the sample area) and photo-CIDNP enhancements than Tube 3 (117.2 ± 33.3 µmol m−2 s−1). Notably, in the illustrative Tubes 2-4 shown here, light was more intense at the bottom of the sample, further away from the LED, where the etchings were denser. These experiments show that by controlling the positioning of the etchings it is possible to control the light distribution around the sample area of the NMRtorch tube, with the position-dependent photo-CIDNP signal enhancement parameter \({\alpha }_{Z}\) being a convenient indicator of both the uniformity and overall light intensity in the sample area.
Studying Chemical Photostability Under UV Illumination
We next tested whether the NMRtorch setup could deliver high-intensity UV light to the NMR sample. Studies of photoreactions, or photodegradation of pharmaceutical and biopharmaceutical products, would both benefit from simultaneous online observation of changes in NMR spectra under very intense light. The irradiation dosage typically required for stability testing of pharmaceutical products is defined in the International Council for Harmonisation of Technical Requirements for Pharmaceuticals for Human Use (ICH) Guidelines Q1B,34,35 and suggests the use of 2% (w/v) quinine hydrochloride as an actinometric system, with prescribed dosage corresponding to greater than 0.5 increase in light absorbance at 400 nm following the UV-A exposure. In practice, such dosage is typically achieved after several hours in specialist UV illumination chambers, which do not allow easy online monitoring of the photodegradation process.
Here, we explored whether the NMRtorch setup can deliver such a UV dosage within reasonable timeframe while allowing to monitor sample degradation by 1H NMR spectra. A specialised lighthead was employed housing a 10 W UV LED array (365 nm peak wavelength), and the degradation of quinine hydrochloride was simultaneously monitored. Even when a borosilicate glass tube was used, NMR spectral changes were readily observed within minutes (Figure 4 and Supplemental Fig. S1), with change in optical density at 400 nm reaching 0.74 ± 0.02 after only 2 hours irradiation (Figure 4B), in line with ICH Q1B guidelines. Using a quartz tube instead of a borosilicate one enhanced the degradation rate by around three-fold (Supplemental Fig. S2). In the 1H NMR spectra acquired during UV irradiation of quinine, a series of phenomena were observed. Firstly, the intensity of the intact quinine peaks decreased (Figure 4C) over time, with this occurring over two distinct timescales, and the greatest changes happening over the first 20 min (Figure 4E). Monitoring signal reduction from different chemical moieties of the molecule allows to assess site-specific rates of degradation, with the quinoline moiety exhibiting greater reductions in signal intensity than the quinuclidine moiety. Secondly, significant spectral broadening was observed, particularly in the aliphatic region (Figure 4D). Finally, a number of additional spectral peaks appear, with this occurring on a similar timescale to the reductions in the intact quinine peaks (Figure 4D,F). Therefore, NMRtorch setup enables high-intensity irradiation of NMR samples in situ with UV light, achieving the doses prescribed by the regulatory guidelines within 2 hours of experimental time, while allowing live monitoring of photodegradation.
Studying Multi-colour Triggering Of Photoswitches
Photo-NMR may also be used to study reactions or transitions triggered by various colours of light, and is particularly well suited to studying the kinetics of such processes. The trans-cis isomerisation of azobenzene and its derivatives is one such light induced transition, which may serve as a useful photoswitch in biotechnology applications.37 Although some transitions in such system have been characterised by NMR before with in situ illumination,38–41, exploring transitions triggered by toggling between numerous illumination colours would be more informative. To illustrate the application of the NMRtorch approach to characterise the kinetics of photoswitches, the photoisomerisation of 1 mM 4-aminoazobenzene (AAB) in DMSO-d6 was studied under continuous illumination with a variety of light wavelengths (Figure 5). At equilibrium under darkness (Figure 5B), AAB exists entirely as the trans isomer. However, if the solution is illuminated with short-wavelength light, such as blue, then isomerisation occurs, shifting equilibrium towards the cis isomer state, giving rise to distinct upfield NMR signals (Figure 5B, marked with asterisks). Therefore, the ratio of these distinct signals can be used to derive populations of trans and cis isomers present in the sample, and track the kinetics of photoisomerisation in response to illumination by different light colours (Figure 5C). Toggling between the colours is easily achieved using a multichannel NMRtorch lighthead containing four different (RGBW) LEDs.
The light toggling experiments reveal that, irrespective to the initial equilibrium state, trans → cis isomerisation triggered by blue light is fast, with rate constant k=65 ± 6 µM min−1. Illumination with each colour establishes its own characteristic trans – cis equilibrium. After blue light illumination, re-equilibration towards trans is faster with green light (108 ± 3 µM min−1) than with red (k=93.0 ± 0.1 µM min−1), but much higher equilibrium population of trans is reached with the red light than with the green. Cis → trans photoisomerisation under green and white light (111 ± 5 µM min−1) occurs at similar rates, but result in different ultimate equilibria, with more trans isomer present under green light. It should be noted that in this LED white light is generated by exciting luminophore with blue light, and therefore a strong blue component in this white light is expected. Interestingly, unlike the colour-driven transitions which here all show mono-exponential behaviours, the thermal transition under darkness from cis to trans can be only satisfactory fitted as the sum of two exponents (kfast=17.9 ± 0.9 µM min−1, kslow = 3.5 ± 0.1 µM min−1), highlighting the hidden complexity of this process uncovered by in situ photo-NMR experiments performed using NMRtorch. To our knowledge, the multi-stage mechanism of thermal transition under darkness for AAB has not been described before. One can also easily explore how the combination of colours would affect both the isomerisation rate and equilibrium state, and thus comprehensively characterise the behaviour of photoswitchable systems in response to different combinations of light stimuli and their intensities. The experiments here demonstrate the convenience of using NMRtorch and the potential of the multi-wavelength illumination approach in studying the kinetics of photoswitches, as well as other photo-induced phenomena which can be detected and monitored in real time by NMR spectral changes.