Electron Paramagnetic Resonance

Ecole Polytechnique Fédérale de Lausanne

Section de Physique

Travaux Pratiques III

Responsable : Dr. Arnaud Magrez Superviseur : Felix Blumenschein

Lausanne, 6 juin 2017 TP III Electron Paramagnetic Resonance

1 Introduction

This TP III experiment is about electron spin resonance (ESR) spectroscopy, in english more often called electron paramagnetic resonance (EPR) and in french résonance param- agnétique électronique (RPE). In EPR spectroscopy we measure the energy differences between the discrete energy states of quasi-free electrons in form of free radicals and paramagnetic centres by measuring their absorption of electro-magnetic radiation. In our present set-up a continuous wave EPR set- up is used. By modulating the magnetic field in combination with the continuous electro- magnetic (EM) microwave (MW) field one can gain insight into the molecular structure, chemical configuration and dynamics of the studied substrate. EPR spectroscopy is used for the detection of free radicals and paramagnetic centres in biology, chemistry and physics. In medicine the application of ESR is rather difficult, as radicals are reactive.

2 Basic ESR theory

1 Electrons have a spin quantum number of s = 2 and corresponding magnetic moment 1 of ms = ± 2 . Without an external ﬁeld, the spin-orientation is randomly distributed. However, in the presence of an external magnetic ﬁeld of strength B0, the spin orientates 1 1 itself parallel (− 2 ) or antiparallel ( 2 ) to B0 with corresponding energy

E = msgeµBB0, with the Landé g-factor (ge = 2.0023 for free electrons) and the Bohr magneton µB = eh¯ = 9.2740 × 10−24 J/T 2me . This splitting of the Energy within an external magnetic ﬁeld is called Zeeman-Eﬀect and leads to a separation of

∆E = geµBB0 (1) between the lower and upper energy state for free electrons, which is directly linear proportional to B0, as depicted in ﬁgure 1. According to Planck’s law, a free electron can be transferred between the two levels by emission or absorption of an photon of frequency v and energy ∆E = hv. This leads to the fundamental equation of EPR spectroscopy, the resonance condition

hv = geµBB0. (2)

ms = +1/2 |↑i

∆E = E+1/2 − E−1/2 Energy

ms = −1/2 |↓i

B0 = 0 B0 6= 0 Figure 1: Splitting of the electron spin state in an external magnetic ﬁeld, Zeeman eﬀect.

1 TP III Electron Paramagnetic Resonance

Most of the standard EPR experiments happen in the MW range of 9 GHz to 10 GHz (≈ 33 mm to 30 mm). Regarding the resonance condition, this corresponds to ﬁelds of around 0.35 T. In most EPR experiments, the MW frequency is kept constant and the magnetic ﬁeld is increased between an interval (sweeping) around the expected signal (continuous wave, "CW" EPR). Reaching the resonance condition, the gap-energy ∆E(B0) matches the MW energy hv and the electrons can transit between the two spin states. The Maxwell- Boltzmann distribution predicts a higher population of the lower energy state nlower than the upper energy state nupper, which leads to more absorption than emission:

nupper Eupper − Elower ∆E hv = exp − = exp − = exp − , (3) nlower kBT kBT kBT

−23 with the Boltzmann constant kB = 1.3806 × 10 J/T. This net resonant absorption due to spin-flip is what we observe in EPR-spectroscopy. An example EPR scan for DPPH is shown in figure 2 on the right in blue, as absorption signal over the swept external magnetic field B0. The basic set-up consists of an electro-magnet for B0 and a MW generator. The MW are guided through wave-guides (metallic, hollow tubes), which allow only single modes. To optimise the EPR absorption, a resonator (cavity) is used. The MW are guided into the resonator where they are reflected from the walls, forming a standing wave. The MW reflected out of the cavity is detected with e.g. a shottky-diode. The dimensions of the cavity therefore have to be adjusted, such that at a certain MW frequency no signal is reflected out of the chamber anymore. Figure 3 shows the reflected signal of the cavity for a frequency sweep. The resonance frequency, where no MW is reflected out of the chamber anymore is visible as the dip in that scan. The adjustment process to find this resonance frequency of the cavity is called matching and tuning. For the EPR spectrum of a sample in the matched and tuned cavity, the MW frequency is kept constant, while B0 is swept. The sample absorbs MW energy at the field corresponding to the resonance condition (eq. 2), B0,res.(hv). The absorption changes the cavity impedance and therefore the coupling condition of the cavity. As the cavity at B0,res.(hv) is now no longer critically coupled, MW is reflected out of the cavity. However, this absorption induced reflection-signal is very small. Therefore, we measure the derivative signal with a Lock-In detector by additional B-field variation coils which induce a sinusoidal field parallel to B0. This corresponds to the red signal on the left in figure 2.

Figure 2: EPR spectroscopy signal of DPPH, taken with the MS400. Shown is the absorption signal over the swept external magnetic ﬁeld B0, on the left as direct Lock-In derivative absorption signal in red and on the right the integrated absorption signal in blue.

2 TP III Electron Paramagnetic Resonance

The set-up sensitivity depends strongly on cavity, MW frequency v, MW power P and sample volume V . The quality of the cavity system is defined through the unloaded quality factor Q0 of the microwave cavity and the filling coefficient kf . Q0 indicates how efficiently the microwave energy is stored (reflected) by the cavity and is defined as

2πEstored Q0 = . (4) Edissipated

Estored and Edissipated correspond to the energy stored and lost while sweeping once over the resonance frequency. The loss for a perfectly matched and tuned cavity is due to the electrical currents generated in the cavity walls by the MW, which again produce heat. Eﬀectively Q0 is found by measuring the position of the resonance frequency vres. and the width ∆v of the absorption signal at half it’s height in ﬁg. 3, together with formula 4 rewritten as

vres. Q0 = . (5) ∆v

A high Q0 value is important for a good spin sensitivity. The lowest number of detectable spins Nmin is calculated as

k1V Nmin = , (6) 2 1 Q0kf v P 2 with constant k1. A good set-up sensitivity is achieved with a large number of spins and small Nmin. Obviously in equation 6, also higher MW power and lfrequency is of advantage for a good Nmin value. The signal’s size, i.e. the integrated signal intensity, is proportional to the concentration of unpaired electrons. The MW intensity in combination with the electron relaxation time τ has another inﬂuence on the signal. The absorbed energy is mostly emitted again in form of phonons, which happens with the rate 1/τ. Too large an MW intensity can lead to an slower relaxation rate compared to the MW excitation rate, i.e. a [ saturation] of the upper level, which

-0.5

0.0 Amplitude of the reflected signal [V] 9.30 9.35 9.40 9.45 9.50 Frequency [GHz]

Figure 3: Reﬂected signal for a MW frequency scan to ﬁnd the resonance position of the cavity.

3 TP III Electron Paramagnetic Resonance again leads to a decrease of the MW absorption. Increasing saturation appears in the EPR spectra as decreasing absorption lines. The theory above is explained for the case of free electrons. In reality however, we look at radicals or paramagnetic centers, the electrons are under the influence of one or more atoms. The influence of the molecule on the electron is visible in the EPR signal, as a change of the g-value, the line-shape, and hyper-fine coupling: g-factor and line shape

In the presence of an atom or molecule, the electron possesses, besides its spin s, some orbital momentum L, changing its total angular momentum. This spin-orbit coupling leads to a change of the g-factor away from ge of an unpaired electron. Effectively, the local magnetic field is changed from B0 to Beff = B0(1 − σ). σ is hereby the influence of the magnetic fields from atoms and molecules. By applying this on equation 2, we get

hv = geµBB0(1 − σ)

⇒ hv = g · µBB0. (7)

σ is the difference of g from the free electrons value ge and is easily determined from the EPR spectrum. Moreover, σ contains information about the substrates’ orbital-structure, as the coupling magnitude depends on the masses. Organic free radicals, with only hydrogen, oxygen, carbon and nitrogen, only have small deviations from ge, while e.g. metals having a more significant influence. For e.g. DPPH, the influence of the molecule is on average zero, leading to an EPR signal for DPPH corresponding to g = 2.0036, which is quasi-free. In general g is a 2nd order tensor with three main or principal components (of the diag- onalised tensor) gx, gy and gz. In the case of an anisotropic electronic structure of the molecule or atomic structure, g is such a tensor. This leads to a signal structure for such anisotropic orbital molecules, which is dependent on their orientation to the direction of B0. By measuring different orientation of the substrate in the external field, additional informations about the substrates structure can be collected. For single crystals, all the molecules are in the same orientation. A molecule with e.g. two different principal components (e.g. an axial-symmetric crystal with the main-axis gx = gk and gy = gz = g⊥) will have different signals, depending on the crystals orientation to B0 at Bx or By,z. For any orientation in between Bx and By,z a superposition is to be expected. However, in liquids the molecules can rotate and the isotropic g value, i.e the fast motion leads to an average of the gi values:

1 giso. = (gx + gy + gz) (8) 3 For powders or frozen liquids however, the spins are randomly oriented and immobile. The resulting signal is an average of all the existing orientations of the molecules and their orbitals, a so-called powder-spectrum. As mentioned above, the signal intensity is proportional to the square root of the MW power for low MW power, and eventual appearance of saturation for too high powers, which also broadens the line-width. In general, the line shape of EPR spectra are rather asymmetric, it can therefore make sense to not calculate the line width but the halfwidth to both sides from the line-centre. Measured is hereby till the point of half of the maximal absorption value in the lines centre.

4 TP III Electron Paramagnetic Resonance

Hyperﬁne Coupling

The electronic spin also couples with the magnetic moment of non-zero nuclear spin, hyper- ﬁne coupling. This splits the resonance signal into multiple (depending on the amount of inﬂuencing nuclei) lines. The number of EPR lines is 2I +1 in the presence of a nuclei with spin I, as the corresponding magnetic moment of the spin has mI = −I, −I + 1, ..., I − 1, I possible values. The nuclear spin of important atoms for their main-isotopes are

Atom 1H 12C 14N 16O 31P 32S 35Cl 55Mn 56Fe 59Co 58Ni 63Cu Spin 1/2 0 1 0 1/2 0 3/2 5/2 0 7/2 0 3/2

Table 1: Nuclear spin of important atoms for their main-isotopes.

If more than one nucleus plays a part, the line-structure becomes more complex. Different nuclei at different distances produce certain splittings which altogether superimpose to a complex splitting structure. As the splitting is proportional to the bonding of the unpaired electron to a nucleus, the splitting decreases in general with increasing distance of the a nucleus from the unpaired electron. For M nuclei of the same spin I and same distance to the unpaired electron, the number of corresponding hyperfine lines is 2MI + 1. Taking the methyl-radical C H3 as example, we get 2 · (3) · (1/2) + 1 = 4 hyperfine lines, as I = 0 for carbon and I = 1/2 for the 3 hydrogen atoms. The four lines are shown in fig. 4. For the methoxymethyl radical H2C (OCH3) it is more complicated: C and O both have I = 0 and therefore no influence. The two hydrogen atoms closer to the unbound elecetron split the EPR line into (2MH1IC + 1) = 3 lines. Each of these three lines are split by the 3 equivalent hydrogens, which are further away, again into (2MH2IC + 1) = 4. The total number of lines is therefore 3 · 4 = 12, as visible in fig. 5. However complicated the structure will be, the total intensity will stay constant and the hyperfine splitting lines are always centred around g. For only one nucleus, the lines have the same height, for different nuclei the height increases in the direction to the centre g. More complicated than the analysis above is however the reverse, assigning the lines to the specific nuclei, but if done well this gives important informations on the molecular

Figure 4: Simulated EPR spectroscopy signal of the methyl-radical C H3 [Detection of free radicals in low-temperature gas-grain reactions of astrophysical interest by R. A. Zhitnikov and Yu. A. Dmitriev, A&A, 386 3 (2002) 1129-1138, DOI: 10.1051/0004-6361:20020268].

Figure 5: Simulated EPR spectroscopy signal of the methoxymethyl-radical H2C (OCH3)[https: //commons.wikimedia.org/wiki/File:EPR_methoxymethyl.png].

5 TP III Electron Paramagnetic Resonance structure. In reality this is even more complicated as the hyperﬁne structure is weakened if visible at all due to powder-broadening. Therefore the hyperﬁne structure might only be visible in solution as for e.g. DPPH.

2.1 Chemicals DPPH

As a standard chemical for calibration, 2,2-diphenyl-1-picrylhydrazyl (DPPH) is often used. DPPH is an organic radical, mostly used as polycrystalline substance (it is not considered as toxic, however as with any radicals avoid skin-contact). It is a stable free radical, with the quasi-free electron situated at the nitrogen atoms between the three benzene rings, as depicted in figure 6. As mentioned, the influence of the molecule is on average zero, due to orbital quenching its g = 2.0036 is very close to the g = 2.0023 of a free electron. However, the hyperfine splitting is observable in solution.

O2N

N N NO2

O2N

Figure 6: Chemical structure of DPPH.

TEMPOL

4-hydroxy-2,2,6,6-tetramethylpiperidin-1-oxyl (TEMPOL) is an organic, stable nitroxyl radical with the structure depicted in ﬁg. 7. Due to its pronounced line-pattern, it is as DPPH also often used for calibration issues. TEMPOL in solution has three characteristic peaks.

H3C CH3 N H3C CH3 O

Figure 7: Chemical structure of TEMPOL.

Copper chloride

Copper chloride dihydrate CuCl2 · 2 H2O is a paramagnetic salt of blue-green colour. The hcp crystal structure eﬀects an anisotropic g-tensor (gx 6= gy 6= gz).

Copper sulfate

Copper sulfate pentahydrate CuSO4 · 5 H2O is a paramagnetic salt of blue colour. The axial-symmetric crystal structure eﬀects also a axial symmetric g tensor. Given x is the axis of axial-symmetry, we have gx = gk and gy = gz = g⊥).

6 TP III Electron Paramagnetic Resonance

3 Preparational Questions

1. What is the electronic structure of atoms, the magnetic moment of cores and the magnetic moment of electrons?

2. What is Zeeman-splitting, spin-orbit coupling and hyper-ﬁne-structure?

3. How do spin-systems interact with EM MW ﬁelds.

4. What is the "X-band" region? Why is mainly X-band used for EPR?

5. What is the basic EPR set-up?

6. Why do we perform Lock-In detection and what is it?

7. What is the meaning of the energy population ratio nlower to nupper for EPR? Cal- culate the ratio for an X-band microwave Frequency of v ≈ 9.73 GHz for a) room temperature b) the boiling point of liquid nitrogen c) liquid helium?

8. Why does a high MW frequency increase the signal? What else can be done to increase the signal?

9. What is saturation? How can saturation be avoided?

10. Why does DPPH have a "free electron"?

11. What information can be interpreted from the position and shape of the ESR-lines?

12. Why does the TEMPOL EPR spectra have three lines? How many lines would dissolved DPPH have?

13. Which values are necessary to calculate the g-values?

14. Which signal (derivative, true signal, integral) is important for a signal to number- of-radicals calibration?

15. What is the unit dB and why do we use it?

7 TP III Electron Paramagnetic Resonance

4 Experiment and Tasks

4.1 Home-made set-up Figure 8 shows the principal set-up, the blue electro magnets are situated with their poles around the brass cavity. The wave-guide goes up to the red box, the MW source. Figure 9 shows the top-view on the cavity between the electro magnets and the poles. Figure 10 shows the opened cavity. At the right side the wave-guide, in the center in grey the actual cavity and under the red plastic the B-ﬁeld variation coils.

Figure 8: Home made set-up.

Figure 9: Home made set-up, top view.

8 TP III Electron Paramagnetic Resonance

Figure 10: Home made set-up, opened cavity with B-ﬁeld variation coils.

Tasks

The tasks in this section are for basic understanding of the set-up. The tasks will be done together with your tutor.

1. Determine the resonance frequency of the ESR cavity (tuning and matching). What is the Q factor for this cavity?

2. Determine the resonance frequency with DPPH in a 5 mm NMR tube. (Avoid skin contact) What caused the resonance change?

3. What is the signal to noise ratio (SNR)?

4.2 MS 400 set-up Figure 11 shows the MiniScope MS 400 EPR spectrometer with external MW frequency indicator. Start the computer and turn both switches on the MS400 on. The log-in name is TPA, PW tpa31416. Start the software MiniScope. It starts as shown in fig. 12. Press Apply params, to apply the red marked parameters. Afterwards auto-adjustment needs to be done by pressing Auto-adjust. Only when no field is red anymore, starting a scan is possible (comp. fig. 13). These two steps have to be done after any change. The dewar in fig. 14 is made of quartz-glass, consisting of a vacuum isolated double-wall system. The sample is inserted into the inner chamber filled with liquid nitrogen. The dewar is fragile and expensive, pay attention.

9 TP III Electron Paramagnetic Resonance

Figure 11: MiniScope MS 400 EPR spectrometer.

Figure 12: Starting appearance of the MiniScope software.

Figure 13: MiniScope software ready.

10 TP III Electron Paramagnetic Resonance

Tasks

General remark: make sure to save results as direct signal, ﬁrst and second integration as well as the corresponding MW frequency (note the frequency before and after each single experiment). The frequency is not saved in the program.

1. Tune and match the system with a DPPH sample in a 5 mm NMR tube, i.e. ﬁnd its cavity resonance frequency.

Calculate B0,res out of the resonance condition for a free electron and for DPPH (g = 2.0036) for the found cavity resonance frequency. Measure the EPR spectrum of DPPH for an appropriate measuring window around the calculated B0,res,DPPH .

What is the diﬀerence between real B0,res and your calculated B0,res?

Is therefore a B0 calibration necessary? If so, calculate the calibration value for B0 and re-plot the measured DPPH spectra. What is the signal to noise ratio (SNR)? Compare your DPPH spectrum to the Literature. What will the EPR line of low concentrated DPPH-solution look like?

2. What influence do the following parameters have (for each at least 5 different settings) MW intensity B-field modulation magnitude

3. Change the settings such to see saturation (electronic and sample-signal).

4. Find g1, g2 and g3 of CuCl2 · 2 H2O. Compare to literature.

5. Find g⊥ and gk for polycrystalline CuSO4 · 5 H2O. Compare to literature.

What would mono-crystalline CuSO4 · 5 H2O look like?

6. Determine the EPR spectra of an unknown single crystal. Measure the spectrum for 10◦ steps in a range of at least 180◦ . Calculate the g-factors and zero-ﬁeld splitting. Which material is the crystal made of?

Figure 14: Quartz-glass dewar.

11 TP III Electron Paramagnetic Resonance

7. Tune and match the cavity for a sample of pure H2O and Ethanol. What are the reasons for the complications? What is the maximal volume you can use in a capillary?

8. Measure the EPR-spectrum of

a) in pure water dissolved CuCl2 · 2 H2O,

b) in pure water dissolved CuSO4 · 5 H2O, c) in methanole dissolved DPPH.

9. Find the g-value for a 40 mM TEMPOL solution (H2O) in a capillary and compare it to the literature value .

10. Find the signal to number-of-radicals calibration by diluting the 40 mM TEMPOL solution with H2O appropriately. For each concentration measurement use one droplet of 10 µl in liquid nitrogen (lN). Measure each concentration for at least 4 diﬀerent settings (MW attenuation), while all other settings are unchanged (to make sure to have the right calibration for possible low-signal and high signal measurements). Why with lN and not at RT? What is therefore the number of radicals in the DPPH sample from above?

11. Read the attached paper, with focus on the EPR-part. Repeat the described UV radical production with pyruvic acid (PA) (best to scan every irradiation-time-step with diﬀerent settings, to ensure a good result for both, the start- and saturation-point). Plot the radical concentration (mM) over irradiation time and ﬁt it.

12. Repeat the radical production with a 1:1 (volumetric) mixture of PA with ethanol and PA with water.

5 Remarks

The written part must answer all preparational questions and include all measurements and their evaluation. If not a theoretical value, every value has an error and needs to be noted. Make sure to do a scientiﬁc presentation of your images, plots, results, errors and their source (especially regarding errors, error calculation, error propagation, ﬁt and correct citing of sources).

6 Literature research

There is much literature on this topic, introductions can be found under

A. Carrington and A. McLachlan Introduction to Magnetic Resonance, Harper & Row, 1969 A. Melissinos Experiments in Modern Physics, Academic Press, 1966 N. M. Atherthon, Principles of Electron Spin Resonance, Wiley, 1993 S. Blundell, Magnetism in condensed matter, Oxford University Press, 2001

12 TP III Electron Paramagnetic Resonance

M. Junk, Assessing the Functional Structure of Molecular Transporters by EPR Spec- troscopy, Chapter 2, Springer, 2012

13 Article

pubs.acs.org/JPCC

Photoinduced Nonpersistent Radicals as Polarizing Agents for X‑Nuclei Dissolution Dynamic Nuclear Polarization † ∥ ⊥ † † # ‡ Andrea Capozzi,*, Jean-Noel̈ Hyacinthe, , Tian Cheng, Tim R. Eichhorn, , Giovanni Boero, § † † Christophe Roussel, Jacques J. van der Klink, and Arnaud Comment*, † ‡ § Institute of Physics of Biological Systems, Institute of Microengineering, and Section of Chemistry and Chemical Engineering, Institute of Chemical Sciences and Engineering, EPFL, Lausanne, Switzerland ∥ School of Health Sciences − Geneva, University of Applied Sciences and Arts Western Switzerland (HES-SO), Geneva, Switzerland ⊥ Image Guided Interventions Laboratory, Faculty of Medicine, University of Geneva, Geneva, Switzerland # Sample Environment and Polarized Targets Group, PSI, Villigen, Switzerland

*S Supporting Information

ABSTRACT: Hyperpolarization via dissolution dynamic nuclear polarization (DNP) is a versatile method to dramatically enhance the liquid-state NMR signal of X-nuclei and can be used for performing metabolic and molecular imaging. It was recently demonstrated that instead of incorporating persistent radicals as source of unpaired electron spins, required for DNP, nonpersistent radicals can be photoinduced in frozen beads of neat pyruvic acid (PA), the most common substrate for metabolic imaging. In the present work, it is shown that the same radicals can be created in frozen solutions containing a fraction of PA in addition to 13C- or 6Li-labeled salts or 129Xe nuclei. The use of these nonpersistent radicals prevents the loss of a substantial part of the polarization during the transfer of hyperpolarized solutions into iron-shielded high-ﬁeld MRI scanners. It is also demonstrated that UV-irradiated d4-PA yields nonpersistent radicals exhibiting similarities with the most eﬃcient and widely used persistent trityl radicals.

I. INTRODUCTION radicals can be produced in concentrations suitable for 1 performing DNP by means of low-temperature (77 K) UV- Since its development, only slightly more than a decade ago, 16 hyperpolarization via dynamic nuclear polarization (DNP) has irradiation of neat pyruvic acid (PA) frozen beads. The rapidly become a well-established and widespread technique for original work also reported that the nonpersistent radicals can enhancing the nuclear magnetic resonance (NMR) signal of be induced in water:PA 1:1 (v/v) mixtures. In this article, we low-gamma nuclei that exhibit long liquid-state longitudinal extend this method and demonstrate that other compounds 13 6 129 15 2−7 besides PA can be hyperpolarized using these photoinduced relaxation time (T1) such as C, Li, Xe, and N. The principle of DNP, known from more than half a century,8 is to nonpersistent radicals. embed radicals, which hold unpaired electron spins, within a glassy sample and transfer the high electron spin polarization II. EXPERIMENTAL METHODS − obtained by both lowering the sample temperature (1 4K) a. Samples Preparation. ESR Samples. By means of a and applying an external magnetic field of 3.35−7 T to the micropipet (1−20 μL range), three 8.0 ± 0.5 μL drops of neat nuclear spins of the sample. This is done by shining microwaves or diluted (in H O, ethanol, or THF) PA or its deuterated form slightly off-resonance from the center of the electron spin 2 (d4-PA) were poured, one by one, inside a synthetic quartz resonance (ESR) line of the radicals. It has been demonstrated fl that the nuclear spin order achieved by DNP in the solid state dewar (Wilmad, 150 mL Suprasil dewar ask type WG-850-B- Q) filled with liquid nitrogen. Samples were irradiated for up to can be maintained through a rapid dissolution method ° producing liquid-state hyperpolarized (HP) solutions.9 The 70 min (turning the dewar of 90 every 10 min) using a 1024 mW maximum output power 365 nm LED array (Hamamatsu lifetime of the HP state is however limited by T1, which is reduced by the presence of the routinely used persistent LC-L5). Once the irradiation process was completed, the dewar − radicals.10 13 Moreover, for medical applications, the radicals was inserted into the ESR spectrometer cavity for measure- need to be removed from the final HP solutions prior to ments. injection. An alternative to filtering or artificial scavenging of the paramagnetic impurities14,15 would be to use nonpersistent Received: July 29, 2015 radicals that are stable at low temperature but recombine Revised: September 8, 2015 during the dissolution step. It was recently shown that such Published: September 8, 2015

Carbon DNP Samples. 2.25 M sodium [1-13C]acetate 14.1 T rodent MRI scanner (Varian, USA)17 and the other solutions were prepared in H2O:ethanol 1:1 (v/v) before operating at 7 T and coupled to a 9.4 T rodent MRI scanner 18,19 13 adding an amount of PA or d4-PA corresponding to 30% of the (Varian, USA). The 5 T polarizer was used for the C and total volume to obtain a final 1.5 M acetate concentration. The 129Xe DNP experiments whereas the 6Li DNP experiments UV-irradiation procedure was performed as described above for were performed using the 7 T polarizer. 1 h on about 25 frozen beads (∼200 μL of solution). An Samples containing sodium [1-13C]acetate were polarized at additional sample containing nitroxyl radicals as paramagnetic 1.50 ± 0.05 and 1.15 ± 0.05 K by shining microwaves at a centers was prepared by adding 50 mM of TEMPOL to a frequency ranging from 139.50 to 140.50 GHz with a nominal H2O:ethanol 1:1 (v/v) solution containing 1.5 M sodium output power of 55 mW (ELVA-1, St. Petersburg, Russia). The 13 [1- C]acetate. 13C polarization time evolution was monitored using a Xenon DNP Samples. Following methods already described 20 ° 5 homemade NMR setup, applying a 2 radio-frequency (rf) in an earlier publication, liquid xenon was embedded in a 2- pulse at 53.437 MHz every 5 min. Once the maximum methyl-1-pentanol:PA (or d4-PA) mixture (with the acid ° fi polarization was reached, a 20 rf pulse was applied, and once corresponding to 10% of the nal volume) to yield 5 M the microwaves were switched off and after waiting for a xenon samples. Once frozen in liquid nitrogen, the samples sufficiently long time for complete relaxation of the sample were extracted from the coldfinger and placed into the quartz ° fi nuclear spin magnetization, an identical 20 rf pulse was applied dewar lled with liquid nitrogen in order to perform 1 h of UV- to measure the reference signal corresponding to the nuclear irradiation. Lithium DNP Samples. 6 Boltzmann polarization. The DNP enhancement was obtained 4.5 M LiCl solutions were prepared by computing the ratio between the two NMR signal integrals. in H2O:ethanol 1:1 (v/v) before adding an amount of PA Once polarized, the samples were dissolved and automatically corresponding to 30% of the total volume to obtain a final 3 M μ transferred into an injection pump (equipped with a rf probe lithium concentration. About 200 L of sample (25 frozen tuned at 150.25 MHz) located inside a nonactively shielded beads) were irradiated at low temperature with UV light for 1 h. 14.1 T rodent MRI scanner. The setup and procedures were Natural abundance xenon gas was purchased from Carbagas, 19,20 similar to the previously described ones, with the notable Lausanne, Switzerland; all the other chemicals were obtained differences that the length of the plastic tube for the transfer of from Sigma-Aldrich, Buchs, Switzerland. the HP solution was 10 m instead of 5 m and the solution b. Low-Temperature X-Band ESR Methods. An X-band transfer time was set to 5 s instead of 2 s. spectrometer (EMX, Bruker Biospin, Rheinstetten, Germany) The 1H NMR signal was recorded as well while polarizing was used for all ESR experiments. The tail of the quartz dewar, the [1-13C]acetate samples. The 1H signal evolution and DNP filled with liquid nitrogen, was placed inside the resonator enhancement were measured as described above, with the cavity of the spectrometer. A series of reference ESR signals ff fi ± μ di erence that the NMR hardware components (coil, lter, and arising from three 8.0 0.5 L frozen beads of ethanol fi containing TEMPO (2,2,6,6-tetramethylpiperidoxyl) radical at power ampli er) were adapted to 212.5 MHz and the bottom various known concentrations (between 25 and 100 mM) were of the inset supporting the sample was entirely made of PTFE to avoid any signal contamination from surrounding protons. used to calibrate the radical concentration as a function of 129 ± signal integral (see Supporting Information Figure S1). The Samples containing Xe were polarized at 1.50 0.05 K by same parameters were kept for all ESR measurements, i.e., shining microwaves at a frequency corresponding to the center of the magnetic field sweep: 338 mT; sweep range: 30 optimal DNP conditions for each radical (139.875 GHz for mT;sweeptime:20s;modulationfrequency:6kHz; UV-irradiated PA, 139.925 GHz for UV-irradiated d4-PA, and 140.3 GHz for TEMPOL; see Supporting Information). The modulation amplitude: 2 G; microwave output power: 0.063 129 mW. Parameters were optimized in order to work in the linear Xe polarization time evolution and DNP enhancement were range of the detector diode of the ESR spectrometer and to measured using a procedure identical to the one used for the 13 ff avoid spurious line-broadening effects. C experiments, with the di erence that the NMR frequency c. Sample Transfer into the DNP Polarizer. Once the was set to 58.787 MHz. low-temperature UV-irradiation was completed, the frozen The 6Li samples were polarized at 1.50 ± 0.05 K in the 7 T beads were quickly transferred from the coldfinger into a DNP setup. The microwave frequency was set to the value polystyrene box containing liquid nitrogen. Afterward, the corresponding to the maximum DNP enhancement (196.75 handling of the sample was standard: a polytetrafluoroethylene GHz; see Supporting Information) and the nominal output (PTFE) sample cup was precooled in liquid nitrogen, and the power was 55 mW (ELVA-1, St. Petersburg, Russia). The 6Li beads were placed inside the cup before it was rapidly inserted polarization time evolution was monitored, using a homemade into the cryostat prefilled with liquid helium (see ref 20 for NMR setup, by applying a 5° rf pulse at 43.973 MHz every 5 6 details). The handling of the sample is not problematic since min. Because of the prohibitively long T1, the solid-state Li the radicals are stable at liquid nitrogen temperature (77 K). A DNP enhancement was not evaluated. Once dissolved, the 6Li precise evaluation of their temporal degradation as a function of T1 and maximum liquid-state polarization were evaluated using temperature has however not yet been performed. It must a rf probe (tuned at 58.89 MHz) located around a custom- nevertheless be noted that it was possible to perform DNP made injection pump, as described in a previous publica- − experiments on irradiated samples stored for months in liquid tion.18 20 The HP 6Li signal decay was monitored by applying nitrogen without observing consistent differences in maximum a10° rf pulse every 10 s. The thermally polarized NMR signal 13C polarization or build-up time constant compared to samples was measured using the same 10° pulse (16 averages with a prepared the same day. repetition time TR = 5T1). The DNP enhancement was d. Solid-State and Liquid-State DNP Methods. Solid- obtained by computing the ratio between the signal integral of state DNP measurements were performed using two different the first measured spectrum and the thermally polarized custom-built polarizers: one operating at 5 T and coupled to a spectrum.

22633 DOI: 10.1021/acs.jpcc.5b07315 J. Phys. Chem. C 2015, 119, 22632−22639 The Journal of Physical Chemistry C Article

Figure 1. Integrated X-band spectrum of UV-irradiated neat PA measured at 77 K as a function of the irradiation time in arbitrary units (A) and normalized to 1 (B).

Figure 2. (A) Radical concentration generated by low-temperature (77 K) UV-irradiation as a function of the illumination time for frozen beads of neat PA (black diamonds), PA:ethanol 1:1 (v/v) (red circles), and PA:THF 1:1 (v/v). (B) Radical concentration generated after 60 min low- temperature (77 K) UV-irradiation as a function of the PA concentration in the frozen beads of ethanol (red circles) and THF (blue triangles) solutions. Dashed lines serve as guide for the eyes.

21 III. RESULTS AND DISCUSSION a less polar one such as THF (H2O relative polarity = 0.207) diminishes it (9.0 ± 0.5 mM). Samples prepared by diluting PA Similar X-band ESR spectra were obtained for all UV-irradiated ± samples, independently of their composition, suggesting that in equal volume with water (27.0 1.5 mM of radical concentration after 1 h UV-irradiation) exhibited a behavior the signal arises from an electron located on a PA molecular 16 orbital. Moreover, it was observed that the broadening of the similar to the samples containing ethanol. spectral line width is independent of the radical concentration. A central point for DNP applications is the determination of the minimum PA to solvent ratio necessary to obtain a A representative example is reported for neat PA in Figure 1 ﬃ where ESR spectra measured at 77 K as a function of UV- su cient radical concentration following UV-irradiation, in irradiation time are presented. A spectrum was measured after particular if PA is not the target substrate to be hyperpolarized. each consecutive 10 min of UV-irradiation and reported in To elucidate this point, the radical concentration was measured, arbitrary units (panel A) and normalized to 1 (panel B). The after 1 h of UV-irradiation, as a function of the PA dilution in comparison between panels A and B clearly demonstrates that ethanol and THF solutions (Figure 2B). For both solvents, it the line shape is independent of the radical concentration. was observed that the radical concentration is independent of Although the nature of the generated radicals is solvent- the PA dilution for any PA concentration above 10% of the independent (see Figure S2), the solvent plays a role in the total volume, corresponding to a minimum of 1.4 M. A radical yield. The radical concentration as a function of the UV- dramatic reduction of the radical concentration was however irradiation time is shown in Figure 2A for neat PA, PA:ethanol observed in solutions containing only 1% of PA (0.14 M). 1:1 (v/v), and PA:THF 1:1 (v/v) frozen beads. In all cases, a In addition to UV-irradiated natural abundance PA, it is also plateau was reached after about 1 h of UV-irradiation. herein proposed to use UV-irradiated d4-PA as DNP polarizing Tentatively, we suggest that a polar solvent such as ethanol agent because it has been shown that narrower ESR line 21 13 22 (H2O relative polarity = 0.654) increases the radical yield radicals lead to higher C polarization. The integrated X- (25.0 ± 1.5 mM) compared to neat PA (15.0 ± 0.7 mM), while band ESR spectrum of UV-irradiated frozen beads of neat PA

22634 DOI: 10.1021/acs.jpcc.5b07315 J. Phys. Chem. C 2015, 119, 22632−22639 The Journal of Physical Chemistry C Article

Figure 3. (A) Integrated X-band ESR spectrum of UV-irradiated neat natural abundance PA measured at 77 K in 8.0 ± 0.5 μL frozen beads after 60 min UV-irradiation. Inset: structural formula of natural abundance PA. (B) Integrated X-band ESR spectrum of UV-irradiated neat d4-PA measured ± μ at 77 K in 8.0 0.5 L frozen beads after 60 min UV-irradiation. Inset: structural formula of d4-PA.

Figure 4. Solid-state 129Xe polarization build-up curves measured at 5 T and 1.5 K in 5 M xenon samples prepared in (A) a 2-methyl-1-pentanol/PA ﬁ ﬁ mixture (10% of PA in the nal volume) and (B) a 2-methyl-1-pentanol/d4-PA mixture (10% of d4-PA in the nal volume).

6 6 Figure 5. (A) Solid-state Li DNP build-up curve measured at 7 T and 1.5 K in a 3 M LiCl sample prepared in H2O:ethanol 1:1 (v/v) with 30% UV-irradiated PA. (B) Liquid-state 6Li signal decay measured in an injection pump placed inside a 9.4 T rodent MRI scanner.

ﬁ 23 and d4-PA are shown in Figure 3. Experimental data were tted EASYSPIN, considering a spin 1/2 radical with isotropic using the PEPPER routine of the MATLAB-based software hyperﬁne coupling to the methyl protons or deuterons for

22635 DOI: 10.1021/acs.jpcc.5b07315 J. Phys. Chem. C 2015, 119, 22632−22639 The Journal of Physical Chemistry C Article natural abundance or deuterated PA, respectively. We deduced that PA and d4-PA have the same g-tensor and dipolar broadening but that irradiated d4-PA has a 6.57-fold smaller hyperfine coupling constant than its protonated counterpart (see Supporting Information). This is in good agreement with 1 2 γ the proportion between H and H gyromagnetic ratios ( 1H/ γ 24 2H = 6.51). Moreover, as expected, the radical yield was ± identical for both PA and d4-PA (16.0 1.0 mM after 1 h of UV-irradiation). In the following, we report experimental results demonstrat- ffi ing that UV-irradiated PA and d4-PA can be used as e cient polarizing agent for enhancing the polarization of 129Xe as well as 6Li and 13C in LiCl and sodium [1-13C]acetate frozen solutions, respectively. As already mentioned, 129Xe DNP experiments were performed at 5 T and 1.5 K.17 A solid-state 129Xe polarization of 5.0 ± 0.2% was measured after 3 h, about 1.5 times higher than what was obtained using nitroxyl radicals as polarizing agent.5 A representative 129Xe polarization build- up curve is shown in Figure 4A. The lithium sample was polarized for 5 h at 1.5 K in our original (now set to 7 T) polarizer and subsequently dissolved prior to transfer into the 9.4 T rodent MRI scanner (Varian/ Magnex, Palo Alto, CA) coupled to the polarizer.18 A liquid- state polarization of 4.0 ± 0.5% was measured in the infusion pump placed inside the scanner bore following the protocol described in a former publication.19 Even though the polarization value was lower than what is reported in the literature for samples prepared with nitroxyl radicals,4 a remarkably long relaxation time of 519 ± 25 s was determined, 6 equivalent to the room-temperature Li T1 measured in thermally polarized 6LiCl aqueous solutions at 9.4 T.25 Results are reported in Figure 5. To investigate in more details the DNP properties of the UV- induced radicals, sodium [1-13C]acetate samples were studied. Our analysis relies on the measurement of the 1H and 13C DNP microwave spectra performed for the three samples (Figure 6). The sample containing TEMPOL (Figure 6A) exhibits an essentially identical spectrum for both nuclei, similar to what has been found for the nitroxyl radical porphyrexide.26 At each irradiation frequency, both 1H and 13C reach the same spin temperature in the stationary state (and it is expected that all nuclei would do the same). This is very different from the behavior observed with trityl radicals.9 The types of radicals leading to identical microwave spectra for all nuclei are usually referred to as “broad-line” radicals whereas the others are called “narrow-line” radicals. In a comparative study of five radicals, it was found that replacement of protons in the solvent by deuterons leads to a 2- or 3-fold improvement of 13C−DNP for “broad-line” radicals, while the reverse effect was found for “ ” 12 Figure 6. 13C and 1H DNP microwave spectra measured at 5 T and narrow-line radicals. It has also been demonstrated in 13 aqueous sodium [1-13C]acetate samples doped with TEMPO 1.5 K in 1.5 M sodium [1- C]acetate samples prepared in that both proton and 13C spin temperatures decrease and water:ethanol 1:1 (v/v) mixtures containing TEMPOL (A), UV- irradiated PA (B), or UV-irradiated d -PA (C). The simulated ESR remain essentially identical to one another with increasing 4 2 spectrum for each of the three radical types is superimposed (solid degree of deuteration. As a consequence, the composition of gray lines). The vertical gray dashed lines represent the center of the sample presented in Figure 6A is not optimal for a “broad- gravity of the ESR spectra. line” radical, and a doubling of the enhancement is expected upon deuteration. Further improvement could also be obtained by optimizing the water/ethanol ratio as well as the radical The 1H microwave spectrum shown in Figure 6B is slightly concentration and by increasing the acetate concentration.2 broader than the corresponding 13C spectrum, but the DNP The choice of a protonated matrix for the present measure- maxima appear at the same frequencies for both nuclei. It is ments was made to allow a fair comparison, based on the therefore likely that the type of radicals present in this sample physical DNP mechanism, between radicals with different ESR exhibits a “broad-line” behavior, for which deuteration could line width. improve the maximum 13C polarization. From a theoretical

22636 DOI: 10.1021/acs.jpcc.5b07315 J. Phys. Chem. C 2015, 119, 22632−22639 The Journal of Physical Chemistry C Article

Figure 7. 13C DNP build-up curves recorded in a 5 T/1.15 K polarizer (left panel) and room-temperature relaxation curves measured, after dissolution and transfer, in an injection pump placed inside a 14.1 T MR scanner (right panel). The samples consisted in frozen 1.5 M sodium 13 [1- C]acetate samples prepared in water/ethanol mixtures containing UV-irradiated PA (red squares), UV-irradiated d4-PA (blue circles), or TEMPOL (green triangles). Colored dashed lines prolong the NMR signal obtained at the end of the DNP process to visually underline the 13 polarization losses during the transfer; arrows in the right panel indicate the measured C T1 of each sample in the 14.1 T MR scanner. standpoint, the sample corresponding to Figure 6C is the most shielding structure of the scanner. The 13C polarization build- interesting. Compared to what was shown in refs 9 and 26, this up measured in the 5 T/1.15 K polarizer is displayed in the left- is an intermediate case for which the two microwave spectra are hand panel of Figure 7 (the microwave frequency was set to the clearly distinct, although the proton spectrum is not well optimal frequency determined by the 13C microwave spectrum resolved. The ratio between the largest 1H and 13C enhance- shown in Figure 6). The central panel of the figure pictures the ments is about 3, which is smaller than the one reported in ref 9 unknown fate of the polarization during the 5 s that it takes to for trityl radicals. At the frequencies corresponding to the transfer the solution from the polarizer into the infusion pump extrema of the 13C microwave spectrum of Figure 6C, the 1H placed inside the 14.1 T scanner bore,20 where the acetate enhancement is rather low because most of the “cooling power” liquid-state 13C polarization decay was monitored (Figure 4, delivered by the microwaves is spent on polarizing the low- right panel).19 The essential result is that the polarization gamma nuclei, in particular 13C. In this case, deuteration is not created by the two nonpersistent radicals was almost entirely expected to improve the 13C polarization. The difference in 13C maintained throughout the transfer across the complex field enhancement between the samples of Figure 6B,C are most path between the polarizer and the scanner (about 5% relative likely linked to the two competing effects highlighted in ref 12. difference between solid-state and liquid-state 13C polarization); A similar behavior was found for 129Xe when the UV-irradiated conversely, the sample containing TEMPOL suffered a clear d -PA radical was used: instead of the 5.0 ± 0.2% polarization polarization loss of around half of its solid-state value. A 4 13 mentioned above for the protonated radical, a value of 9.5 ± correlation is that the acetate liquid-state C T1 measured in 0.4% was recorded with its deuterated form (see Figure 4B). It the 14.1 T scanner for the sample containing nitroxyl radicals is ± is therefore expected that all low-gamma nuclei will follow this much shorter (16.2 0.1 s) than for the other two samples in trend. which the UV-induced radicals quenched at the time of The acetate samples were also used for investigating one dissolution. The dissolved UV-irradiated samples indeed crucial point of the dissolution-DNP technique: the transfer of exhibited a T1 value nearly identical to the value measured the HP solution from the polarizer to the NMR/MRI system. for melted nonirradiated PA beads in a 600 MHz high- ± To minimize the transfer time, the dissolved sample can be resolution system (45 1 s). driven through a plastic tube from one instrument to the other using compressed He gas.1,9,20,27 An ill-defined combination of IV. CONCLUSIONS several liquid-state relaxation mechanisms may cause loss of The herein presented results demonstrate that radicals created polarization along the way and it is generally believed that the by UV-irradiation of PA exemplify a low-cost and versatile way path should not pass through regions of very low or very for polarizing X-nuclei in molecules that can be codissolved, rapidly changing magnetic field values (the so-called adiabatic together with a fraction of PA, in a medium with a polarity condition). The above-mentioned issue has been recently comparable to that of water. Moreover, the self-quenching studied in some details for 1H and 13C.28 As the UV-induced property of these radicals upon dissolution offers a promising radicals are entirely quenched when the temperature of the solution for simplifying the HP liquid transfer procedure from sample is raised during the dissolution process, there is a the polarizer to a high-field MRI scanner with minimal disappearance of the main relaxation mechanism causing polarization losses. An additional advantage of the automatic nuclear polarization losses, namely paramagnetic relaxation. elimination of the potentially toxic radical species during In the herein reported hyperpolarized 13C experiments, the dissolution could in principle be to alleviate the quality control HP solution was transferred into the bore of a nonactively procedure required in human MRI/MRS. shielded 14.1 T rodent MRI scanner (Varian/Magnex, Palo Even though at the present stage of the work the radical Alto, CA) from the polarizer located outside of the massive iron concentration obtained after UV-irradiation of PA is still rather

22637 DOI: 10.1021/acs.jpcc.5b07315 J. Phys. Chem. C 2015, 119, 22632−22639 The Journal of Physical Chemistry C Article low to efficiently perform DNP at high magnetic field, other (4) van Heeswijk, R. B.; et al. Hyperpolarized Lithium-6 as a Sensor biocompatible solvent systems are under investigation to of Nanomolar Contrast Agents. Magn. Reson. Med. 2009, 61, 1489− improve the radical yield. Furthermore, we are directing our 1493. efforts in the research or synthesis of new molecules with the (5) Capozzi, A.; Roussel, C.; Comment, A.; Hyacinthe, J. N. Optimal same photochemical properties than PA, but with a more Glass-Forming Solvent Brings Sublimation Dynamic Nuclear Polar- isolated molecular environment at the site of the unpaired ization to Xe-129 Hyperpolarization Biomedical Imaging Standards. J. fi Phys. Chem. C 2015, 119, 5020−5025. electron. A reduced hyper ne coupling between the latter and (6) Comment, A.; et al. Hyperpolarizing Gases Via Dynamic Nuclear surrounding nuclei would produce a sharper EPR-line radical Polarization and Sublimation. Phys. Rev. Lett. 2010, 105, 8104 1−4. species. Increasing the radical yield, or even the discovery of (7) Karlsson, M.; Jensen, P. R.; Duus, J. O.; Meier, S.; Lerche, M. H. other UV-induced radicals, will provide a more generic Development of Dissolution Dnp-Mr Substrates for Metabolic improvement of the method for a widespread use of Research. Appl. Magn. Reson. 2012, 43, 223−236. nonpersistent radicals for DNP. Concerning the practical (8) Abragam, A.; Goldman, M. Principles of Dynamic Nuclear- applications of this method, it is important to note that the Polarization. Rep. Prog. Phys. 1978, 41, 395−467. liquid-state measurements were performed inside an injection (9) Wolber, J.; et al. Generating Highly Polarized Nuclear Spins in pump designed for in vivo rodent experiments, meaning that all Solution Using Dynamic Nuclear Polarization. Nucl. Instrum. Methods liquid-state values reported for 13C and 6Li nuclei correspond to Phys. Res., Sect. A 2004, 526, 173−181. the polarization levels at the time of injection, and although (10) Lumata, L.; Merritt, M.; Khemtong, C.; Ratnakar, S. J.; van Tol, these values are not the highest obtained to date, they are J.; Yu, L.; Song, L. K.; Kovacs, Z. The Efficiency of Dpph as a sufficient for performing in vivo studies with the competitive Polarising Agent for Dnp-Nmr Spectroscopy. RSC Adv. 2012, 2, 16 12812−12817. advantage that the solutions are free of radicals. (11) Lumata, L.; Merritt, M. E.; Malloy, C. R.; Sherry, A. D.; Kovacs, ■ ASSOCIATED CONTENT Z. Impact of Gd3+ on Dnp of [1-C-13]Pyruvate Doped with Trityl Ox063, Bdpa, or 4-Oxo-Tempo. J. Phys. Chem. A 2012, 116, 5129− *S Supporting Information 5138. The Supporting Information is available free of charge on the (12) Lumata, L.; Merritt, M. E.; Kovacs, Z. Influence of Deuteration ACS Publications website at DOI: 10.1021/acs.jpcc.5b07315. in the Glassing Matrix on C-13 Dynamic Nuclear Polarization. Phys. − X-band ESR spectrometer calibration curve; ESR addi- Chem. Chem. Phys. 2013, 15, 7032 7035. tional data and fitting parameters; 129Xe and 6Li DNP (13) Lumata, L.; Ratnakar, S. J.; Jindal, A.; Merritt, M.; Comment, A.; Malloy, C.; Sherry, A. D.; Kovacs, Z. Bdpa: An Efficient Polarizing microwaves spectra (PDF) Agent for Fast Dissolution Dynamic Nuclear Polarization Nmr Spectroscopy. Chem. - Eur. J. 2011, 17, 10825−10827. ■ AUTHOR INFORMATION (14) Nelson, S. J.; et al. Metabolic Imaging of Patients with Prostate Corresponding Authors Cancer Using Hyperpolarized [1-C-13]Pyruvate. Sci. Transl. Med. − *E-mail andrea.capozzi@epfl.ch. 2013, 5, 198ra108 1 10. *E-mail arnaud.comment@epfl.ch. (15) Mieville, P.; et al. Scavenging Free Radicals to Preserve Enhancement and Extend Relaxation Times in Nmr Using Dynamic Notes − fi Nuclear Polarization. Angew. Chem., Int. Ed. 2010, 49, 6182 6185. The authors declare no competing nancial interest. (16) Eichhorn, T. R.; Takado, Y.; Salameh, N.; Capozzi, A.; Cheng, T.; Hyacinthe, J. N.; Mishkovsky, M.; Roussel, C.; Comment, A. ■ ACKNOWLEDGMENTS Hyperpolarization without Persistent Radicals for in Vivo Real-Time We thank Dr. Ben van den Brandt and Dr. Patrick Hautle who Metabolic Imaging. Proc. Natl. Acad. Sci. U. S. A. 2013, 110, 18064− kindly helped us with the X-band ESR measurements at PSI. 18069. Our acknowledgment goes also to Dr. Gil Navon and Dr. Mor (17) Jannin, S.; Comment, A.; Kurdzesau, F.; Konter, J. A.; Hautle, Mishkovsky for their suggestions about 6Li experiments. We P.; van den Brandt, B.; van der Klink, J. J. A 140 Ghz Prepolarizer for also thank Dr. Hikari Yoshihara for the constructive discussions Dissolution Dynamic Nuclear Polarization. J. Chem. Phys. 2008, 128, 241102 1−4. about the nature of the UV-radical. This work was supported by (18) Cheng, T.; Capozzi, A.; Takado, Y.; Balzan, R.; Comment, A. the Swiss National Science Foundation (Grant ’ ́ Over 35% Liquid-State C-13 Polarization Obtained Via Dissolution PP00P2_157547), the Centre d Imagerie BioMedicale Dynamic Nuclear Polarization at 7 T and 1 K Using Ubiquitous (CIBM) of the UNIL, UNIGE, HUG, CHUV, EPFL, the Nitroxyl Radicals. Phys. Chem. Chem. Phys. 2013, 15, 20819−20822. Leenards and Jeantet Foundations and a SOCLE grant of the (19) Cheng, T.; Mishkovsky, M.; Bastiaansen, J. A. M.; Ouari, O.; School of Health Sciences − Geneva, University of Applied Hautle, P.; Tordo, P.; van den Brandt, B.; Comment, A. Automated Sciences and Arts Western Switzerland. Transfer and Injection of Hyperpolarized Molecules with Polarization Measurement Prior to in Vivo Nmr. NMR Biomed. 2013, 26, 1582− ■ REFERENCES 1588. (1) Ardenkjaer-Larsen, J. H.; Fridlund, B.; Gram, A.; Hansson, G.; (20) Comment, A.; van den Brandt, B.; Uffmann, K.; Kurdzesau, F.; Hansson, L.; Lerche, M. H.; Servin, R.; Thaning, M.; Golman, K. Jannin, S.; Konter, J. A.; Hautle, P.; Wenckebach, W. T.; Gruetter, R.; Increase in Signal-to-Noise Ratio of > 10,000 Times in Liquid-State van der Klink, J. J. Principles of Operation of a Dnp Prepolarizer Nmr. Proc. Natl. Acad. Sci. U. S. A. 2003, 100, 10158−10163. Coupled to a Rodent Mri Scanner. Appl. Magn. Reson. 2008, 34, 313− (2) Kurdzesau, F.; van den Brandt, B.; Comment, A.; Hautle, P.; 319. Jannin, S.; van der Klink, J. J.; Konter, J. A. Dynamic Nuclear (21) Lide, D. R. CRC Handbook of Chemistry and Physics, 90th ed.; Polarization of Small Labelled Molecules in Frozen Water-Alcohol CRC Press: Boca Raton, FL, 2009. Solutions. J. Phys. D: Appl. Phys. 2008, 41, 155506 1−10. (22) Comment, A.; van den Brandt, B.; Uffmann, K.; Kurdzesau, F.; (3) Cudalbu, C.; Comment, A.; Kurdzesau, F.; van Heeswijk, R. B.; Jannin, S.; Konter, J. A.; Hautle, P.; Wenckebach, W. T. H.; Gruetter, Uffmann, K.; Jannin, S.; Denisov, V.; Kirik, D.; Gruetter, R. Feasibility R.; van der Klink, J. J. Design and Performance of a Dnp Prepolarizer of in Vivo N-15 Mrs Detection of Hyperpolarized N-15 Labeled Coupled to a Rodent Mri Scanner. Concepts Magn. Reson., Part B 2007, Choline in Rats. Phys. Chem. Chem. Phys. 2010, 12, 5818−5823. 31B, 255−269.

22638 DOI: 10.1021/acs.jpcc.5b07315 J. Phys. Chem. C 2015, 119, 22632−22639 The Journal of Physical Chemistry C Article

(23) Stoll, S.; Schweiger, A.; Easyspin, A. Comprehensive Software Package for Spectral Simulation and Analysis in Epr. J. Magn. Reson. 2006, 178,42−55. (24) Abragam, A. Principles of Nuclear Magnetism; Oxford University Press: New York, 1961. (25) Balzan, R.; Mishkovsky, M.; Solomon, Y.; van Heeswijk, R. B.; Gruetter, R.; Eliav, U.; Navon, G.; Comment, A. Hyperpolarized Li-6 as a Probe for Hemoglobin Oxygenation Level. Contrast Media Mol. Imaging 2015, DOI: 10.1002/cmmi.1656. (26) Borghini, M.; Udo, F. Dynamic Polarization of C-13 Nuclei in 1- Butanol. Phys. Lett. A 1973, 43,93−94. (27) Bowen, S.; Hilty, C. Rapid Sample Injection for Hyperpolarized Nmr Spectroscopy. Phys. Chem. Chem. Phys. 2010, 12, 5766−5770. (28) Milani, J.; Vuichoud, B.; Bornet, A.; Mieville, P.; Mottier, R.; Jannin, S.; Bodenhausen, G. A Magnetic Tunnel to Shelter Hyperpolarized Fluids. Rev. Sci. Instrum. 2015, 86, 024101 1−8.

22639 DOI: 10.1021/acs.jpcc.5b07315 J. Phys. Chem. C 2015, 119, 22632−22639 SUPPORTING INFORMATION

Photo-Induced non-Persistent Radicals as Polarizing Agents for

X-Nuclei Dissolution-DNP

Andrea Capozzi,*,§ Jean-Noël Hyacinthe,†,♣ Tian Cheng,§ Tim R. Eichhorn,§,Δ Giovanni Boero,Ξ

Christophe Roussel,♦ Jacques J. van der Klink,§ Arnaud Comment.*,§

§ Institute of Physics of Biological Systems, EPFL, Switzerland

†School of Health Sciences – Geneva, University of Applied Sciences and Arts Western

Switzerland.

♣Image Guided Interventions Laboratory, Faculty of Medicine, University of Geneva, Geneva,

Switzerland

Δ Sample Environment and Polarized Targets Group, PSI, Switzerland

Ξ Institute of Microengineering, EPFL, Switzerland

♦ Section of Chemistry and Chemical Engineering, Institute of Chemical Sciences and

Engineering, EPFL, Switzerland

AUTHOR INFORMATION

Corresponding Author

*Andrea Capozzi, e-mail: [email protected]; *Arnaud Comment, e-mail: [email protected]. Phone: +41 21 69 37982. Home-page: http://gr-co.epfl.ch.

X-band ESR spectrometer calibration curve

Figure S1 X-band ESR spectrometer calibration curve. The data (black circles) were measured in reference samples containing known TEMPO concentrations. The linear fit (red line; slope: 0.145±0.002) was used to estimate the radical concentration of all UV-irradiated samples.

ESR additional data and fitting parameters

The X-band ESR spectra at 77 K of neat and diluted (in H2O, ethanol, or THF) PA measured after 1 h of UV-irradiation are reported in Figure S2. It can be seen that the solvent does not affect the fundamental structure of the ESR spectrum, suggesting that the signal arises from an electron located on a PA molecular orbital. Note that the width of the spectrum is slightly modified by the presence of the different solvents.

Figure S2 Normalized integrated X-band spectrum of UV-irradiated neat PA, PA:H2O 1:1 (v/v), PA:ethanol 1:1

(v/v) and PA:THF 1:1 (v/v) measured at 77 K after 1 h of irradiation.

ESR spectra were fitted using the PEPPER routine of the MATLAB®-based software

EASYSPIN with the following parameters:1

 Neat pyruvic acid

Electron spin S = ½; g-tensor = [2.0041; 2.0037; 2.0042]; isotropic hyperfine coupling

with the molecule methyl group ACH3 = 48.00 MHz; phenomenological FWHM Gaussian

line-broadening of 1.01 mT.

 Neat deuterated pyruvic acid

Electron spin S = ½; g-tensor = [2.0041; 2.0037; 2.0042]; isotropic hyperfine coupling

with the molecule methyl group ACD3 = 7.30 MHz; phenomenological FWHM Gaussian

line-broadening of 1.01 mT.

 TEMPO in concentration of 50 mM

Electron spin S = ½; g-tensor = [2.0094; 2.0065; 2.0017]; rhombic hyperfine coupling

with 14N = [20.48; 17.68; 101.00] MHz; phenomenological FWHM Gaussian line-

broadening of 1.4 mT.

The same parameters were used for simulating the different radicals spectra at 5 T (Figure 6 of the main text).

129Xe and 6Li DNP microwave spectra

Figure 3 129Xe DNP microwave spectra measured at 5 T and 1.5 K in samples containing 5 M xenon dissolved in 2- methyl-1-pentanol: PA 10:1 (v/v) (red dots) and 5 M xenon dissolved in 2-methyl-1-pentanol: PA 10:1 (v/v) (blue dots).

6 6 Figure 4 Li DNP microwave spectra measured at 7 T and 1.5 K in 3 M LiCl sample prepared in H2O:ethanol 1:1

(v/v) with 30% UV-irradiated PA. Values reported on the vertical axis are rescaled as a function of the maximum liquid state polarization, measured in 9.4 T MR scanner after dissolution, irradiating the sample at 196.75 GHz.

AUTHOR INFORMATION

Corresponding Author

*Andrea Capozzi, e-mail: [email protected]

*Arnaud Comment, e-mail: [email protected]. Phone: +41 21 69 37982. Home-page: http://gr-co.epfl.ch.

Notes

The authors declare no competing financial interests.

REFERENCES

1. Stoll, S.; Schweiger, A., Easyspin, a Comprehensive Software Package for Spectral Simulation and Analysis in Epr. J Magn Reson 2006, 178, 42-55.