Cross-Effect Dynamic Nuclear Polarization Explained: Polarization

Publication Date (Web): January 15, 2019. Copyright © 2019 American Chemical Society. *E-mail: [email protected]. Cite this:J. Phys. Chem. Lett. 20...
0 downloads 0 Views 2MB Size
Perspective Cite This: J. Phys. Chem. Lett. 2019, 10, 548−558

pubs.acs.org/JPCL

Cross-Effect Dynamic Nuclear Polarization Explained: Polarization, Depolarization, and Oversaturation Asif Equbal,† Alisa Leavesley,† Sheetal Kumar Jain,† and Songi Han*,†,‡ †

Department of Chemistry and Biochemistry, University of California, Santa Barbara, Santa Barbara, California 93106, United States Department of Chemical Engineering, University of California, Santa Barbara, Santa Barbara, California 93106, United States



J. Phys. Chem. Lett. Downloaded from pubs.acs.org by IOWA STATE UNIV on 01/28/19. For personal use only.

S Supporting Information *

ABSTRACT: The scope of this Perspective is to analytically describe NMR hyperpolarization by the three-spin cross effect (CE) dynamic nuclear polarization (DNP) using an effective Hamiltonian concept. We apply, for the first time, the bimodal operator-based Floquet theory in the Zeemaninteraction frame for two and three coupled spins to derive the known interaction Hamiltonian for CE-DNP. With a unified understanding of CEDNP, and supported by empirical observation of the state of electron spin polarization under the given experimental conditions, we explain diverse manifestations of CE from oversaturation, enhanced hyperpolarization by broad-band saturation, to nuclear spin depolarization under magic-angle spinning.

N

understanding of the CE-DNP theory, and supported by experimental measurement of the state of electron spin polarization under the given experimental conditions, we describe three different nuances of CE-DNP: (i) oversaturation in CE-DNP resulting in a decreased DNP enhancement above a threshold microwave (μw) power, (ii) improved CE-DNP performance by adiabatic chirp-pulse train saturation of electron spins, and (iii) nuclear spin depolarization causing a loss in NMR signal under magic-angle spinning (MAS) in the absence of μw irradiation. All three phenomena are governed by the same CE Hamiltonian for polarization transfer in an e−e−n three-spin system that, however, manifest themselves drastically differently depending on the polarization difference between the donor electron spins (ΔPe) and the acceptor nuclear spin (Pn) ensemble in frequency and time. In the magnetic resonance literature, the optimum DNP condition and its mechanism have been depicted mostly using a degeneracy-of-state or product-state picture, which is cumbersome beyond the two-spin case.14,15,17 Using an operator-based formalism, we show an alternative, simpler derivation of the effective three-spin coupling Hamiltonian leading to CE, where the polarization differences between the three involved spins are the main quantities modulated by the effective Hamiltonian. The framework is very similar to the ones used for elucidating ssNMR experiments under MAS, which requires calculating the effective (or time-averaged) Hamiltonian using the average Hamiltonian theory (AHT)18,19 or Floquet theory20−23 in the appropriate frame. The density operator formalism presented here is an effective, straightfor-

uclear magnetic resonance spectroscopy (NMR) is intrinsically a low-sensitivity technique because of weak Zeeman interactions resulting from a low gyromagnetic ratio (γ) that leads to a relatively low nuclear spin polarization, even at the highest magnetic field available today. Transferring spin polarization from a sensitive (high γ) spin to an insensitive (low γ) spin is a popular approach in NMR spectroscopy to augment its sensitivity. Dynamic nuclear polarization (DNP) is a technique that enables the transfer of polarization from electron spins to nuclear spins (γe ≫ γn) and has become one of the most exciting add-ons to solid-state NMR (ssNMR).1−8 In principle, DNP enhancement can be equal to the ratio γe/γn, but experimentally, signal amplification corresponding to this limit is not achieved, while the polarization transfer efficiency varies dramatically with the sample system and experimental conditions. Of the various DNP mechanisms in insulating solids, cross effect (CE) DNP is more efficient at higher magnetic field compared to the solid effect (SE). This is attributed to the lower power requirement of CE that relies on allowed spin transitions. A major research focus in DNP has been the optimization of the radicals used as polarizing agents and the experimental conditions (temperature, glassingsolvent) to achieve greater CE efficiency.9−11 In this Perspective, we present a generalized and extensible theoretical framework for describing the polarization transfer according to the CE-DNP mechanism12−16 from first principles that eases the hurdle of simulating experimental systems and predicting new CE-DNP modalities. We describe the effective Hamiltonian operational in CE-DNP and show that the evolution of the density operator under the effective Hamiltonian can result in nuclear spin polarization enhancement or depletion, depending on the prevailing density operator and steadystate polarization of the electron spins. With a unified © XXXX American Chemical Society

Received: September 14, 2018 Accepted: January 15, 2019 Published: January 15, 2019 548

DOI: 10.1021/acs.jpclett.8b02834 J. Phys. Chem. Lett. 2019, 10, 548−558

The Journal of Physical Chemistry Letters

Perspective

time-averaging, while all other terms have effectively no contribution, owing to the secular approximation valid at high magnetic field. In a homo-γ coupled two-spin system, where the Larmor frequencies of the two spins are close in magnitude, all ZQ terms (diagonal and off-diagonal components) remain effective. The modulation frequencies of the various coupling terms ZQ/SQ/DQ of the Hamiltonian (eq 1) in the Zeeman frame are depicted for each dipolar term in Scheme 1, including the complex conjugate (c.c.) terms for I−S. This ensures that the net Hamiltonian is Hermitian. It is important to emphasize here that for any coherent transfer of polarization to occur, there must exist an effective off-diagonal, ZQ (flip-flop) or DQ (flip-flip), operator in the Zeeman basis. In the case of a hetero-γ spin system, however, only the diagonal, zz part of the dipolar term survives the timeaveraging, requiring RF irradiation for the generation of offdiagonal ZQ or DQ terms. One approach to achieve polarization transfer is to transform the zz dipolar term (eq 1) to effectively act as an off-diagonal ZQ or DQ operator upon RF irradiation. Irradiation on both spins is a must for the IzSz coupling term to create coherence on both spins (ZQ or

We describe the effective Hamiltonian operational in CE-DNP and show that the evolution of the density operator under the effective Hamiltonian can result in nuclear spin polarization enhancement or depletion, depending on the prevailing density operator and steady-state polarization of the electron spins. ward, and versatile tool to model and rationalize currently available experimental observations of CE-DNP effects and has the benefit of being extended to describe multifrequency and time-modulated DNP experiments in the future. Theory for Polarization Transfer Mechanism. We first pedagogically present a theoretical framework to describe the polarization transfer mechanism using two- and three-spin model systems. In general, any transfer of magnetization or polarization between spins is possible in the presence of a coupling interaction between the spins, either direct or indirect. Quantum mechanically, the spin−spin dipolar coupling Hamiltonian can be constituted as a sum of zeroquantum (ZQ), single-quantum (SQ), and double-quantum (DQ) operators (eq 1). In the (rotating) frame of Zeeman interactions, these terms are modulated by frequencies that depend on the Zeeman interaction strengths of the individual spins, I and S (here, Larmor frequencies ω0I and ω0S). In the high B0 field limit, the zz or “ZQ-diagonal” term remains unmodulated, while the ZQ-off-diagonal term is modulated by the difference of the two Zeeman frequencies, ω0I − ω0S. The SQ terms are modulated by either ω0I or ω0S, depending on which spin (I or S) has coherence, and the DQoff-diagonal terms are modulated by the sum of the two Zeeman frequencies, ω0I + ω0S. A IS≠IÖÖÖÖÖÖÖ zSzÆ ´ω ÖÖÖÖÖÖÖ

Ĥdipolar =

ZQ,diagonal(zz)

+

e−i(ω1I Ix + ω1SSx)t

The Hartman−Hahn cross-polarization (CP) between heteroγ spins under MAS is the best known example in this category, and the effects of broad-band irradiation on both spins and the resultant Hamiltonian are well-known.24 Another approach to achieve polarization transfer is to transform the SQ dipolar operator to generate the coherent ZQ or DQ terms. Because the SQ dipolar operators consist of pseudosecular dipolar terms, they require irradiation on only one spin for coherence e−iω1SSxt

− + generation (i.e., I ±Sz ⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯→ I ±̃ Sz̃ + I ±̃ S ̃ + I ±̃ S ̃ ); hence, rotating frame

irradiation on that single spin suffices to transfer polarization between the two spins.25 The pseudosecular SQ term plays an important role in EPR experiments and DNP. In fact, μwinduced SE-DNP falls into this category; see the Supporting Information, section A1 for more details. Passive coupling can also generate polarization transfer between two uncoupled spins with the aid of a third spin that is coupled to both of these uncoupled spins.26,27 Following the general principle described here, irradiation on both spins (donor as well as acceptor of polarization) is needed to achieve polarization transfer between hetero-γ spins, unless there is a strong pseudosecular SQ interaction (e.g., hyperfine coupling) between the hetero-γ that can be exploited. In the case of DNP, irradiation on both electron and nuclear spins with the required power cannot be easily performed with the currently available hardware. However, if the pseudosecular (SQ) hyperfine interaction between the electron and nuclear spins is strong, polarization transfer from an electron spin to a nuclear spin by irradiation on just the

ZQ,off‐diagonal

C + + ωISD(I −Sz + IzS −)ÖÆ IS (I Sz + IzS ) + ´ω ÖÖÖÖÖÖÖÖÖÖÖÖÖÖÖÖÖÖÖÖÖÖÖÖÖÖÖÖÖÖÖÖÖÖÖÖÖÖÖÖÖÖÖÖ Ö≠ÖÖÖÖÖÖÖÖÖÖÖÖÖÖÖÖÖÖÖÖÖÖÖÖÖÖÖÖÖÖÖÖÖÖÖÖÖÖÖÖÖÖÖÖ SQ,off‐diagonal

DQ,off‐diagonal

+

rotating frame

B + − + ´ω + I −S +)ÖÆ IS(I S ≠ÖÖÖÖÖÖÖÖÖÖÖÖÖÖÖÖÖÖÖÖ ÖÖÖÖÖÖÖÖÖÖÖÖÖÖÖÖÖÖÖÖÖ

E + + ωISF (I −S −)ÖÆ + ´ω IS(I S ) + ÖÖÖÖÖÖÖÖÖÖÖÖÖÖÖÖÖÖÖÖÖÖÖÖÖÖÖ Ö≠ÖÖÖÖÖÖÖÖÖÖÖÖÖÖÖÖÖÖÖÖÖÖÖÖÖÖÖ



DQ), specifically, IzSz ⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯→ Iz̃ Sz̃ + I +̃ S ̃ + I +̃ S ̃ + c.c..

(1)

In general, any term in the Hamiltonian with a large net nonzero frequency modulation will be averaged out to the first order under the secular approximation. In the case of a heteroγ coupled two-spin system, without external perturbation, only the diagonal (zz) part of the ZQ dipolar term survives upon

Scheme 1. Modulation of Various Dipolar Terms (Alphabets) in the Zeeman Rotating Frame of Individual Spins Constituting the I−S Dipolar Hamiltonian Shown in eq 1

549

DOI: 10.1021/acs.jpclett.8b02834 J. Phys. Chem. Lett. 2019, 10, 548−558

The Journal of Physical Chemistry Letters

Perspective

ρ0 = p1 S1z + p2 S2z = ΣpS z14 + ΔpS z23

electron spin is effective. Advantages of this mechanism are that the polarization to nuclear spin is transferred along the direction of the B0 field and that the transfer rate is limited by T1 rather than T1ρ of the nuclear spin. The derivation of the effective Hamiltonian and mechanism of polarization transfer from S (electron) to I (nuclear) spins, as operational in CEDNP upon μw irradiation of only the S spins, is the core subject to be presented here, followed by its implementation to rationalize and simulate experimental observation of oversaturation,28 broad-band μw saturation,29,30 and depolarization,31−33 as described above. Polarization Transfer in Two-Spin Case. We start with a twospin case (denoted as 1 and 2), where two spins are homo-γ dipolar coupled. In the context of DNP, this is to describe the polarization exchange and transfer between electron spins with similar frequencies (with differences smaller than their coupling strength) in the presence or absence of μw irradiation. This applies when many electron spins with a broad frequency-distribution are coupled, or under MAS conditions when electron spin frequencies level-cross at some rotor-position. We start with this description because the framework established for the two spin- 1 homo-γ (dipolar) 2 coupled case forms the theoretical basis for describing many spin cases, including the three-spin case of two electron spins and one nuclear spin. In the absence of external-perturbation, the Hamiltonian is the sum of Zeeman and coupling interactions: Ĥ = ω01S1z + ω02S2z + Ĥ dipolar. In the rotating frame of the two Zeeman interactions, transformed using the propagator U = e−i(ω01S1z + ω02S2z)t, the Hamiltonian is timemodulated and takes the form

with Σp and Δp being the sum and difference of p1 and p2, respectively. The effective Hamiltonian, which evolves this initial state-of-the-system, is Ù Ĥeff = ω12B (S1+S2− + S1−S2+) = 2ω12B Sx23

Σp Δp + cos(ω12B t ) 2 2 Σp Δp and ⟨S2z(t )⟩ = − cos(ω12B t ) 2 2

⟨S1z(t )⟩ =

(5)

Equation 5 shows that the polarizations of coupled spins 1 and 2 oscillate around a constant value (average of their polarization, Σp/2) with an amplitude proportional to half of the difference of the polarization (Δp/2) and rate, ω12/2π, proportional to the magnitude of the effective coupling Hamiltonian (eq 4). This shows that net polarization oscillation/transfer is observed if spins 1 and 2 have unequal polarizations, which can be achieved by selective saturation and/or with thermal polarization of a spin system with a large g (or chemical shift) anisotropy. Figure 1a maps the polarization transferred from S1 to S2 spins as a function of the dipolar evolution time and the polarization difference, p1 − p2. Simply relying on eqs 4 and 5, we show that the polarization on spins S1 and S2 oscillates with frequency corresponding to the dipolar coupling strength, ωB12, while the maximum polarization transferred depends linearly on the polarization difference

= ω12AS1zS2z + ω12B (S1+S2−e i(ω01− ω02)t + S1−S2+e−i(ω01− ω02)t ) + ω12C (S1zS2+e iω02t + S1+S2z e iω01t ) + ω12D(S1zS2−e−iω02t + S1−S2z e−iω01t ) + ω12E S1+S2+e i(ω01+ ω02)t + ω12F S1−S2−e−i(ω01+ ω02)t (2)

This Hamiltonian is equivalent to the one presented in eq 1 (for two S spins), but in the rotating frame of the two Zeeman interactions. The first-order effective (time-averaged) HamilτÙ Ù 1 tonian, Ĥ = ∫ Ĥ dt , can then be obtained by integrating 0

the Hamiltonian over a time period characterized by the frequency of each term.18 The SQ and DQ terms are always modulated by large frequencies, leading to an average of zero in the first-order Hamiltonian, and therefore can be ignored here. Only the ZQ term can have a nonzero average, depending on the value of ω01 − ω02. When ω01 ≠ ω02, the time-averaged Hamiltonian is Ĥ̃ eff = ωA12S1zS2z. This zz term only causes broadening of the individual spin’s signal, but cannot change the state of the spin system. When ω01 = ω02, the average is H̃̂ = (ωA12S1zS2z + ωB12(S+1 S−2 + S−1 S+2 )). Here, only the off-diagonal ZQ (flip-flop) (eq 1) part of this Hamiltonian survives in the rotating frame of the two Zeeman interactions, which in turn can lead to polarization transfer, as will be demonstrated next. We first define operators: 1 1 Sx23 = 2 (S1+S2− + S1−S2+), Sy23 = 2i (S1+S2− − S1−S2+), 1

(4)

In other words, the initial density operator represents spin states in which spins 1 and 2 are prepared as z magnetization, with polarizations of magnitude p1 and p2, respectively. Using a 23 23 simple calculation, it can be seen that S23 x , Sy , and Sz act as 34 1 “fictitious” spin- operators that satisfy the commutation 2 23 23 relation [S23 j , Sk ] = iεjkl Sl (where j, k, l are the elements of the set {x, y, z} in all possible permutations and ε denotes the Levi-Civita symbol). Only the second part of the initial density operator (eq 3), the off-diagonal ZQ term S23 z , can evolve under the effective Hamiltonian (eq 4). The first part of the initial density operator, S14 z , remains unaffected as it commutes 23 14 with the effective Hamiltonian, i.e. [H̃̂ eff, S14 z ] = [Sx , Sz ] = 0. The modulation of the individual spin polarization under the effective Hamiltonian can be seen by calculating the trace, ⟨Siz(t)⟩, of the Siz operator, here shown for ⟨S1z(t)⟩ and ⟨S2z(t)⟩.

Ù ̂ − ω01S1z − ω02S2z Ĥ = U†HU

τ

(3)

The exchange of polarization between two homo-γ spins (with chemical shift difference smaller than their coupling strength) is an important phenomenon that leads, for example, to large spectral diffusion among dipolar coupled electron spins and is in turn critical in the context of nuclear spin polarization enhancement and depolarization under MAS.

1

Sz14 = 2 (S1z + S2z), and Sz23 = 2 (S1z − S2z) and consider the initial density operator to be 550

DOI: 10.1021/acs.jpclett.8b02834 J. Phys. Chem. Lett. 2019, 10, 548−558

The Journal of Physical Chemistry Letters

Perspective

= 1, 2) spins, respectively (see SI section A2). The dipolar Hamiltonian terms modulated by the large frequencies (∼electron spin Larmor frequency) will be ignored, as they do not contribute to the effective S to I polarization transfer. We include only the ISi−SQ and S1S2−ZQ coupling terms that are modulated by lower frequencies (typically sub GHz). These terms will play an important role in CE-DNP transfer, as will be explained later. The truncated rotating frame Hamiltonian defined by the propagator, U = e−i(ωo1S1z + ωo2S2z + ωo3I3z)t becomes Ù Ĥ = ω12AS1zS2z + ω12B (S1+S2−e i(ω01− ω02)t + S1−S2+e−i(ω01− ω02)t ) +



(ωi3AI3z + ωiC3 I3+e iω03t + ωi3DI3−e−iω03t )Siz (6)

i = 1,2

Note that the flip-flop (off-diagonal) part of the homo-γ S1− S2 coupling is modulated by a frequency equal to the difference between the two electron spin’s precession frequencies (as shown earlier in eq 2), Δ12 = (ω01 − ω02), and the pseudosecular part of the heteronuclear ISi coupling (also known as the hyperfine term in electron−nuclear coupling) modulated by the I spin’s Larmor frequency, ω03. If Δ12 and ω03 are very large compared to the strength of the corresponding coupling strengths, ω12 and ωi3 (i = 1 or 2), the frequency-modulated terms can be ignored to first-order in the Hamiltonian owing to secular approximation. However, the cross-terms (higher-order Hamiltonian) of these modulated terms can cause polarization transfer among two coupled S to I spins, which can be obtained using perturbation theory. The Hamiltonian in eq 6 is time-dependent, modulated periodically with characteristic frequencies Δ12 and ω03, but does not commute with itself at different times. In pursuit of getting physical insights using such a time-dependent, noncommutating Hamiltonian, an effective time-independent Hamiltonian needs to be computed. This is done using a special operator-based perturbation analysis of the AHT or the Floquet theory, commonly employed in ssNMR under MAS. Both of these theories exploit the periodicity of the timedependent Hamiltonian to solve for higher-order perturbation corrections. Although AHT and Floquet theory approaches are equivalent for the case discussed here, we choose the Floquet theory as it has an advantage over AHT in solving timedependent Hamiltonian multiple incommensurate frequencies. The Floquet formalism is well-documented in the ssNMR literature,22,23,36,37 and therefore we emphasize the standard analytic equations to find the effective Hamiltonian for DNP (see SI section A2 for details). Using ladder, FnΔ and Fkω03, and number,37 NΔ and Nω03, operators, the bimodal (two frequencies, here Δ12 and ω03) Floquet Hamiltonian of eq 6 can be represented as

Figure 1. (a) Polarization oscillations between two homonuclear coupled spins (5 kHz) without any external perturbation, as a function of coupling evolution time and the polarization difference between the two spins. The contour maps the polarization on spin 2. (b) Trace along time axis to show polarization of spin 2 (blue) and spin 1 (red) vs time. (c) Trace along y axis to show polarization transferred from 1 to 2 as a function of initial polarization difference.

between the spins, Δp. This is illustrated in the traces extracted from the contours of Figure 1a, shown in Figure 11b,c (the script for this calculation can be found in the Supporting Information). The exchange of polarization between two homo-γ spins (with chemical shift difference smaller than their coupling strength) is an important phenomenon that leads, e.g., to large spectral diffusion among dipolar coupled electron spins and is in turn critical in the context of nuclear spin polarization enhancement and depolarization under MAS, which will be discussed later. The equivalent mechanism in nuclear spins is important for diffusion of nuclear polarization from core to bulk spins.14,17 Polarization Transfer in Three-Spin Case. Next, we consider the three-spin case. There are three subcategories: (i) all three are hetero-γ spins (I, N, S), (ii) all three are homo-γ spins (S1, S2, S3), (iii) and two are homo-γ spins (S) and one hetero-γ spin (I). The first case is a simple extension of second-order CP, e.g., in 1H−13C−15N PAIN-CP.26 The second case is similar to the two-spin case discussed above. However, unlike in the hetero-γ spin case, the three homo-γ spin case cannot be considered as a mere sum of two homo-γ spins. This is because of the noncommutativity of the homo-γ coupling Hamiltonian, i.e., [HS1S2, HS1S3] ≠ 0, where HSiSj is the dipolar coupling Hamiltonian between homogeneous spins i and j. This leads to dipolar truncation effects, a phenomenon discussed extensively in the ssNMR literature.35 The third case, in the scenario where the two S spins are the electron spins and I the nuclear spin, is the most important one for DNP and will be discussed here. The rotating frame dipolar Hamiltonian is time-modulated by the frequencies ω0i, ω03, (ω0i ± ω03), and (ω01 ± ω02), where ω0i and ω03 refer to the Larmor frequencies of I and Si (i



H̃F =



∑ ∑ n =−∞ k =−∞

H̃ (n , k)FΔn Fωk 03 + Δ12 NΔ + ω03Nω03

(7)

where n and k are the Fourier indices representing the frequency multiplicity. The indices n and k in the above Hamiltonian can be truncated to [−1, 1], as the Hamiltonian (in eq 6) simply has ±1 multiples of the two characteristic frequencies, Δ12 and ω03. The relevant Floquet Hamiltonian terms are then 551

DOI: 10.1021/acs.jpclett.8b02834 J. Phys. Chem. Lett. 2019, 10, 548−558

The Journal of Physical Chemistry Letters H̃ (0,0) =

Perspective

ρ0 = p1 S1z + p2 S2z + p3 I3z

A ̃ (1,0) = ω12B S1+S2− , ωi3AI3zSiz + ´ω 12 S≠1ÖÖÖÖÖÖÖÖÖÖ zS2zÆ , H ´ÖÖÖÖÖÖÖÖÖÖ≠ÖÖÖÖÖÖÖÖÖÖÆ ÖÖÖÖÖÖÖÖÖÖ = 1,2 Ö≠ÖÖÖÖÖÖÖÖÖÖÖÖÖÖÖÆ S1S2 ,(ZQ) S1S2 ,(ZZ) ´iÖÖÖÖÖÖÖÖÖÖÖÖÖÖ



0.5(S1z − S2z) − I3z 2 0.5(S1z − S2z) + I3z + (Δps + p3 ) 2 (S1z + S2z) + Σps 2

= (Δps − p3 )

SiI3 ,(ZZ) B − + H̃ (−1,0) = ´ω S2ÖÆ , H̃ (0,1) = 12S ÖÖÖÖÖÖÖÖÖ Ö≠1ÖÖÖÖÖÖÖÖÖ S1S2 ,(ZQ)



ωiC3 I3+Siz ,

= 1,2 ≠ÖÖÖÖÖÖÖÖÖÖÖÖÖÖÆ ´iÖÖÖÖÖÖÖÖÖÖÖÖÖÖ SiI3 ,(SQ)

̃ (0, −1)

H

=



ωi3DI1−Siz

= 1,2 Ö≠ÖÖÖÖÖÖÖÖÖÖÖÖÖÖÖÆ ´iÖÖÖÖÖÖÖÖÖÖÖÖÖÖ

where Δps and Σps are the difference (p1 − p2) and sum (p1 + p2) of polarization of the two S spins, respectively. The initial density operator can be written as a sum of three terms with coefficients representing (a) difference of Δps and p3, (b) sum of Δps and p3, and (c) Σps. Using a simple calculation, it can be seen that only the first part of the density operator in eq 11 can evolve under the CE Hamiltonian (eq 10). This is because the latter two operators, [(S1z − S2z) + I3z] and [S1z + S2z], commute with the effective Hamiltonian operator (S+1 S−2 I−3 + c.c.) and therefore do not evolve under the CE Hamiltonian. Just as the effective two-spin flip-flop operator (eq 4) modulates Δp between two homogeneous spins, the threespin flip-flop-flip operator (eq 10) modulates Δps − p3. Hence, this three-spin effective Hamiltonian yields a perfect archetype for CE-DNP

(8)

SiI3 ,(SQ)

We apply the van Vleck perturbation theory21,23,36 to get the effective Hamiltonian in the rotating frame. Such a treatment is valid when the condition nΔ12 + kω03 ≠ 0 is satisfied (i.e., avoiding resonance or accidental degeneracy conditions). The effective Floquet Hamiltonian with up to second order of the van Vleck expansion21,23,36 is given by H̃ eff =

1 [H̃ (n0 − n , k 0− k) , H̃ (n , k)] 2 n ,k n,k nΔ12 + kω03 0 , k 0 Ö≠ÖÖÖÖÖÖÖÖÖÖÖÖÖÆ 0 0 ´nÖÖÖÖÖÖÖÖÖÖÖÖ ´ ÖÖÖÖÖÖÖÖÖÖÖÖÖÖÖÖÖÖÖÖÖÖÖÖÖÖÖÖÖÖÖÖÖÖÖÖÖÖ ≠2ÖÖÖÖÖÖÖÖÖÖÖÖÖÖÖÖÖÖÖÖÖÖÖÖÖÖÖÖÖÖÖÖÖÖÖÖÖÖÆ 1

∑ H̃ (n , k ) − ∑ ∑ 0

H̃ eff

0

H̃ eff

(9)

where n0 and k0 are the Fourier indices and will sum over only those values where the resonance condition n0Δ12 + k0ω03 = 0 is satisfied, in both first and second order. The first-order term of the effective Hamiltonian is zero in our case. If the ω03 and Δ12 frequencies are significantly larger than the coupling strengths, the main contribution to the effective Hamiltonian comes from the second-order term, which is simply the commutator between different first-order terms. The summation of Fourier indices in the second order physically means that the net modulation frequency of the commutator of two terms is the sum of frequencies of the individual terms. Of all the possible hetero cross-terms, [S1zS2z, SizI3z], [S1zS2z, SizI±3 ], [S±1 S∓2 , SizI3z], and [S±1 S∓2 , SizI±], only the last cross-term, i.e., the commutator between the S1S2−ZQ (H̃ (±1,0)) and the SiI3− SQ (H̃ (0,∓1)) coupling term, can lead to a three-spin polarization transfer operator, as will be further discussed. Adding all the possible resonance conditions, (n0, k0) = (±1, ∓1), the second-order effective Hamiltonian operational for polarization transfer between three spins, termed henceforth as the CE Hamiltonian, then becomes

CE H̃ eff =

C ) + −− ω12B (ω13C − ω23 S1 S2 I3 + c.c. ω03

ωe1e2(ωe1n − ωe2n) ω0n

(S1+S2−I3−) + c.c.

(12)

According to the effective CE Hamiltonian, the rate of the polarization transfer is directly proportional to the strength of the e−e coupling, ωe1e2, and the difference of the two e−n (hyperfine) couplings, ωe1n − ωe2n, while inversely proportional to the nuclear Larmor frequency, ω0n (see SI section B2 for further discussion). The CE theory has been presented in the literature using different approaches. A common approach is the diagonalization13 of the electron spin Hamiltonian using a unitary transformation, as recently described by Thankamony et al.,8 showing that the hyperfine-coupling term under a specific unitary transformation yields a Hamiltonian similar to eq 12. Note that this Hamiltonian gives us information only about the rate of CE transfer, but not the amount of transfer, which is proportional to the term Δps − p3. Therefore, large CE enhancement of nuclei ideally requires a large polarization difference, Δps, between the two electron spins that fulfill the CE condition compared to the nuclear spin polarization, pI. Without μw perturbation, Δps − pI is small because the constants, pi (i = 1, 2, 3), are simply proportional to the Boltzmann polarization. Therefore, in practice, efficient saturation of part of the electron paramagnetic resonance (EPR) spectrum is required for CE-driven DNP to generate a large electron spin polarization difference that can lead to large nuclear spin polarization enhancement. To illustrate this subtlety of the CE Hamiltonian, Figure 2 maps the polarizations, ⟨Siz(t)⟩, of the three-spin system as a function of time for two different cases: (a) Δps > pI and (b) Δps < pI. It can be clearly seen that for a large initial Δps in Sspins (blue and green), a large polarization is transferred to the I-spin (red) (Figure 2a). The oscillations of all involved spins have the same modulation rate, as the underlying Hamiltonian is the same. It should be noted that these oscillations or transfer rates are independent of the initial density operator of the system and instead depend on the strength of the effective CE Hamiltonian. In the limit where the polarization difference

ÄÅ −1 ÅÅÅÅ [H̃ (1,0) , H̃ (0, −1)] [H̃ (−1,0) , H̃ (0,1)] 2 ̃ Heff = + ÅÅ 2 ÅÅÅÇ Δ12 −Δ12 ÑÉ (0,1) (−1,0) (0, −1) ̃ ̃ ̃ , H̃ (1,0)] ÑÑÑÑ [H , H ] [H + + ÑÑ ÑÑ ω03 −ω03 ÑÖ =

(11)

(10)

As mentioned, this second-order cross-term is resonant and will be dominant only if the resonance condition, n0Δ12 + k0ω03 = 0, is satisfied, i.e., Δ12 = ω03 because n0 = −k0. This entails that the Larmor frequency difference between the two coupled S spins must equal the Larmor frequency of the third spin. In order to understand how this effective Hamiltonian causes oscillations in the polarization, let us consider the following initial density operator: 552

DOI: 10.1021/acs.jpclett.8b02834 J. Phys. Chem. Lett. 2019, 10, 548−558

The Journal of Physical Chemistry Letters

Perspective

of polarization transfer, yielding NMR signal enhancement by CE-DNP. In the limit of thermal mixing of a strongly coupled S spin bath (high concentration) or high μw power, the CEDNP efficiency decreases, as will be demonstrated experimentally in the next section. The various polarization transfer mechanisms are summarized in Table 1. The CE-DNP mechanism falls into the threespin case, where the effective Hamiltonian of the form S+1 S−2 I± originates from the cross-term or commutator between the S1S2−ZQ (flip-flop) and SiI−SQ (pseudosecular) coupling terms. In general, a Hamiltonian of the form ∏iI±i evolves a density matrix term of the form ∑i±piIiz. An important difference between an effective CE and SE-DNP Hamiltonian (SI section A1) is that the effective SE Hamiltonian is generated only in the presence of off-resonant μw irradiation, whereas the CE Hamiltonian is operational with or without μw irradiation; the CE Hamiltonian is “intrinsically” operational by the dynamics of the given spin system that satisfies the required resonance condition.



EXPERIMENTAL SECTION The μw-amplitude (νμw 1 ) currently available for high magnetic field DNP is tiny (sub-MHz) compared to the strengths of the g-anisotropy (∼0.8 GHz for nitroxide radicals at 7 T), hyperfine-coupling (a few MHz), and e−e coupling, i.e. including dipolar (∼24 MHz for 1.3 nm) and exchange (∼40 MHz for AMUPOL)38 interactions. Hence, the electron to nuclear spin polarization transfer rate in CE-DNP depends mainly on the e−e and e−n coupling strengths and ω0n.15 The role of μw irradiation is essentially to saturate (or depolarize) a portion of the EPR line (inhomogeneously broadened), where the degree of its saturation (electron-depolarization) is dependent on μw power as well as the intrinsic spin dynamics.39 Specifically, the extent of electron spin saturation depends on electron spectral diffusion (eSD), which equilibrates the polarization via the e−e dipolar coupled spin network and cross-relaxation, besides on T1e. The eSD mechanism broadens the width of the hole burnt in the EPR line by μw irradiation, and a new quasi steady-state electron spin polarization is reached that is different from the Boltzmann polarization under the Zeeman Hamiltonian. The cumulative polarization difference (ΔPe) generated by μw saturation between the electron spins (satisfying the CE conditions) across the EPR line, and the evolution of this quasi steady-state under the CE Hamiltonian, leads to the enhanced polarization of the coupled nuclear spins. The total enhancement is the weighted sum of enhancements from different electron spin pairs satisfying the CE conditions. Different crystallite orientations lead to different CE enhancements as the essential interactions are anisotropic in nature. Polarization of the bulk-nuclei is achieved via the spin-diffusion mechanism, similar to those operating in coupled electron spins.14 Using this mechanistic understanding of CE-DNP, we will next illustrate the experimental nuances of select CE-DNP features.

Figure 2. Polarization oscillations between e1−e2−H spins satisfying the CE condition leading to (a) nuclear enhancement for ΔPe > Pn and (b) depolarization ΔPe < Pn. The Larmor frequency of 1H was set to 300 MHz, same as the difference of electron-spins Larmor frequencies. e1 − e2, e1 − H, and e2 − H couplings were to 20, 20, and 0 MHz, respectively.

between the two S spins is smaller than the polarization of the I spin, Δps < pI, a reverse polarization transfer would occur, i.e. the I spin will lose its polarization to the two S spins. This can lead to depolarization, as depicted in Figure 2b. The smaller the value of p1 − p2 compared to half of p3, the more the I-spin polarization be depleted to the S-spins. Experimentally, these oscillations are difficult to capture owing to (i) nuclear spin diffusion, (ii) powder averaging, and (iii) relaxation effects, which are not included in the simulations presented here. Also, continuous μw saturation of the electron spins further adds to the multifaceted complexity of electron spin dynamics. Simulations of changes in polarization in experimentally more realistic scenarios, involving powder orientation, relaxation effects, and continuous μw saturation, are demonstrated using the SpinEvolution package in the Supporting Information (section B2). The CE mechanism is the most efficient mechanism in the limit where the nuclear Larmor frequency ω0n is larger than the strength of the S1S2 coupling. Otherwise, the equalization of the electron spin polarization (or the reduction in Δp) can not be circumvented owing to the first-order S1S2−ZQ Hamiltonian itself, as discussed in the homonuclear two-spin case, diminishing the CE-DNP efficiency (see SI section B2). For this reason, CE requires an inhomogeneously broadened EPR line, so that a polarization difference between electron spin populations can be created and maintained during the course

Table 1. Summary: Polarization Transfer between Coupled Spins spin−spin coupling operator

external perturbation

example

IzSz I±Sz+ c.c. S+1 S−2 + c.c.

B1 irradiation on both I and S spins B1 irradiation on only S spin no irradiation (if ω0S1 ≈ ω0S2)

cross-polarization solid effect DNP spin diffusion

S+1 S−2 + I±Siz + c.c.

no irradiation (if ω0I ≈ ω0S1 − ω0S2)

cross effect DNP

553

DOI: 10.1021/acs.jpclett.8b02834 J. Phys. Chem. Lett. 2019, 10, 548−558

The Journal of Physical Chemistry Letters

Perspective

Figure 3. (a) Normalized DNP enhancement vs μw frequency for 120 mW power, 60 s build-up time, at 7 T and 4 K, using 40 mM 4AT in DNPjuice. (b) DNP enhancement vs μw power for fixed μw frequency, 193.7 GHz. (c) ELDOR saturation profile for four different μw powers. The probe frequency is set to 193.7 GHz, which is the optimum μw frequency for DNP.

Figure 4. (a) Normalized DNP enhancement vs μw frequency for 120 mW power at 7 T using 40 mM 4AT acquired at four temperatures. (b) DNP enhancement vs μw power for the four temperatures at their optimum μw frequencies.

and width of the electron spin saturation increase with μw increasing νμw 1 . The μw power of ν1 = 1.7 mW (orange line) is too low to measurably saturate the EPR line, and correspondingly the ε is small. In contrast, at maximum νμw 1 = 120 mW, the EPR line is dramatically oversaturated, which leads to a net lower ΔPe and therefore lower ε. Maximum ε is achieved at an intermediate νμw 1 (here, ∼25 mW), where ΔPe is maximal for this system. The ELDOR experiments clearly demonstrate that optimizing the CE efficiency can be essentially achieved by maximizing ΔPe. The saturation factor that utilizes the eSD mechanism is determined by a combined effect of electron spin relaxation rates and the e−e coupling strengths or the flip-flop transition rate. Therefore, the extent of saturation via eSD is also governed by electron spin concentrations and temperatures, as these parameters markedly influence the e−e coupling strength and/or the electron spin relaxation rates. Higher concentrations and lower temperatures generally lead to a larger saturation via eSD. To demonstrate how temperature influences the saturation factor, and in turn ϵ, we measure the normalized enhancement versus the μw frequency, as well as power. Figure 4a shows normalized ϵ versus fμw at four different temperatures [4 K (red), 10 K (green), 25 K (blue), and 80 K (orange)] of a sample containing 40 mM 4AT in DNP-juice. The normalized frequency profiles at different temperatures appear similar, with some small changes occurring only in fμw values where positive and negative maximum enhancements are observed. This is attributed to changes in the relative contribution of SE and CE to the overall DNP effect with temperature. At higher temperature, the SE contribution decreases, which reduces the difference between fμw corresponding to positive and negative maxima, but we will not focus on this effect.47 Here, we monitor how temperature

The experiments shown here were acquired with a home-built, quasi-optics-based, dual NMR/EPR spectrometer at 7 T.40−43 Details and parameters of the experimental setup are given in SI section C1. Oversaturation. The 1H DNP enhancement (ϵ) recorded as a function of μw frequency (fμw) for 40 mM 4-amino Tempo (4AT) in DNP-juice at 4 K and 7 T field is shown in Figure 3a. Here, ϵ is defined as the ratio of NMR signal obtained with and without μw irradiation. Two fμw conditions displaying maximal ϵ, 193.67 and 194.1 GHz, typical for nitroxide CE-DNP at 7 T, can be seen.44 This frequency profile is obtained with the maximum available μw power, νμw 1 = 120 mW. At the positive enhancement condition (fμw = 193.7 GHz) ε as a function of νμw 1 is shown in Figure 3b, whose shape a priori is surprising. Intuitively, ϵ is expected to increase with increased νμw 1 , but here ε reaches an optimum value at 25 mW and subsequently μw decreases with increasing νμw 1 . For instance, ϵ at ν1 = 120 mW μw is less than half of that obtained at ν1 = 25 mW. This phenomenon is a consequence of “oversaturation” of the EPR line and can be rationalized using the theoretical framework presented here. As discussed earlier, the CE enhancement is proportional to ΔPe between CE electron spin pairs across the EPR line. The quasi steady-state electron spin polarization profile, and hence ΔPe, induced by μw irradiation can be experimentally accessed by pump−probe ELDOR experiments.45,46 In Figure 3c, experimental ELDOR profiles acquired at the same conditions for the four different νμw 1 (marked with arrows in Figure 3b) are shown, with a fixed saturation pulse duration of 100 ms. The ELDOR profiles show the phenomenological exchange of polarization between electron spins across the EPR line with those at the probe frequency, i.e., 193.7 GHz in this experiment. The ELDOR profiles clearly reveal that the extent 554

DOI: 10.1021/acs.jpclett.8b02834 J. Phys. Chem. Lett. 2019, 10, 548−558

The Journal of Physical Chemistry Letters

Perspective

influences the optimal νμw 1 in CE-DNP. This is shown in Figure 4b. At lower temperatures, where saturation via eSD is very effective (because of long T1e), optimum ϵ is reached at much 39,46 lower νμw Increase in νμw 1 . 1 beyond optimum attenuates ϵ via oversaturation. Raising the temperature decreases the time frame over which eSD occurs (shorter T1e), so that larger νμw 1 is needed to generate the maximum ΔPe and ε, as validated at at 10, 25, and 80 K. This implies that for a fixed high νμw 1 , ϵ might be higher at elevated temperatures. In the literature, the cause has sometimes been of decreasing ϵ with increased νμw 1 ascribed to sample heating.48 However, relying on the analytical theory of CE-DNP and the shape of the experimental ELDOR profile, we clearly demonstrate that such effects result from oversaturation of the EPR spectrum as first reported by Siaw et al28 and lately by Leavesly et al.49 Chirp-DNP. In general, DNP enhancement is achieved using CW μw irradiation, but recent hardware developments have shown that enhancements can be dramatically increased when using frequency-modulated pulse-trains50 (e.g., chirp) instead of CW irradiation. The benefit of broad-band pulsed μw irradiation is more notable in the regime of low spectral diffusion, i.e. at higher temperatures and/or lower electron spin concentrations, where ΔPe(chirp) > ΔPe(CW) . The use of linear chirp-pulses, in which the μw frequency varies linearly in time, can lead to broad-band electron spin saturation (i.e., larger ΔPe ⇒ larger ε), whose effect is amplified in systems where small eSD is the limiting factor. A major advancement in this area has been made by Kaminker et al.,30 who reported on the implementation of arbitrary phase-shaped pulses for DNP at 7 T, further enhancing the work of Hovav et al. on chirpDNP at 3.34 T.51 Kaminker et al. and Hovav et al. have demonstrated that in the limit of faster frequency modulation rate (δfμw/δt) than the relaxation rate, the DNP efficiency of chirp irradiation increases with increased chirp bandwidth, δfμw (or the frequency-modulation amplitude of chirp), until a threshold is reached where the oversaturation condition is reached. It is straightforward to understand that the optimal δfμw must be smaller than the ω0n, otherwise it will cause a decrease in net ΔPe between the CE electron pair and hence the DNP enhancement. Indeed, using a 40 mM 4AT radical system, we demonstrate that chirp-DNP can lead to >2-fold greater enhancement compared to conventional CW DNP (Figure 5a). This is due to larger ΔPe induced at steady-state, as confirmed by ELDOR (see SI section C2). The chirp irradiation scheme was implemented using an arbitrary waveform generation (AWG) setup in the Han Lab, as introduced by Kaminker et al.42 Notably, chirp-DNP increases not only the magnitude of ε but also its robustness with respect to the μw irradiation (carrier) frequency (see Figure 5a). Nuclear Spin Depolarization. The effective Hamiltonian used above holds true for static-samples, but not exactly for magicangle spinning (MAS) experiments, as under MAS the energies of spin are time-modulated because of anisotropic (g and dipolar) interactions. Modulation of ESR frequencies over the course of the MAS rotor period leads to multiple different energy-level crossings (diabatic transitions) and avoidedcrossings (adiabatic transitions) of the coupled e−e−n spins, as detailed by Thurber and Tycko.15 MAS therefore renders the spin-B1 (μw-induced electron transition/saturation) and spin−spin (e−e homogeneous-mixing and e−e−n CE-mixing) interactions time-dependent, also called rotor-events.16 The transitions caused at the rotor-events depends on the magnitude of the perturbations, i.e., the off-diagonal terms

Figure 5. (a) DNP frequency profile using 40 mM 4AT at 7 T and 4 K acquired using CW (red) and frequency-modulated (black) irradiation. (b) Numerically simulated 1H depolarization profile as a function of spin rate and T1e. The simulation was done for 3 spins (eeH) with electron g-tensors of nitroxide-based radical, e−e anisotropic coupling of 37 MHz, and using a 7 T field conditions.

(e.g., νμw 1 for electron spin saturation, ωee for e−e mixing, and ωeeωen for CE transitions) and the rate of the energy change at ω0n

the crossing, i.e., the diagonal terms (mainly dependent on ganisotropy and MAS rate). An analytic solution to the transition dynamics for a time-dependent Hamiltonian was first proposed by Landau52 and Zener53 for a two energy-level system. According to the Landau−Zener (LZ) model, the adiabatic transition probability 7(adiabatic or avoided crossing) = 1 − exp( −πE2 /A) (13)

The 7 at the crossing is inversely proportional to the energy-crossing-rate (A) and directly proportional to (scales up) the magnitude of perturbation (E) that mixes the two states of the system. 7 = 1 means that the transition is purely adiabatic and that the state-of-the-system completely changes from one to another. This would happen when the perturbation factor (E2) is much larger than the rate of energy change (A) at the crossing. 7 = 0 (no perturbation) would essentially mean that the two states do not mix at the energycrossing condition. Intermediate values of 7 would basically employ partial mixing of the states-of-the-system at the crossing. Thurber et al.15 first used the LZ concept to explain the CE mechanism under MAS. Using this theory, they also uncovered the phenomenon of nuclear depolarization, which leads to a decrease in nuclear spin polarization under MAS, in the absence of μw irradiation.31,32 Thurber and Tycko31 and Mentink-Vigier et al.32 showed that under MAS, electron spin pairs (partially) exchange their polarization through dipolar/ exchange coupling (similar to eq 4) at some specific rotor positions that depend on the orientation of the individual g555

DOI: 10.1021/acs.jpclett.8b02834 J. Phys. Chem. Lett. 2019, 10, 548−558

The Journal of Physical Chemistry Letters

Perspective

tensors. If the transition probability (7 ) is small because of the weak ωee, ΔPe will be reduced after crossing. Under the condition of long T1e compared to the rotor-period, the electron spins do not relax fast enough to reach their Boltzmann equilibrium. Subsequently, the rotor-positions satisfying the CE-resonance condition can cause a reverse transfer of polarization, i.e. from the nuclear spins to electron spins if ΔPe < 0.5 Pn. This depletion of the nuclear polarization is termed nuclear spin depolarization, where this effect increases with increased spinning-rate before reaching a plateau. We show in Figure 5b a numerically simulated 1H depolarization profile map, displaying increased depolarization with increased MAS spinning rate and increased T1e. This results from the effect of small ωee coupling (leading to nonadiabatic mixing, eq 4) and an operative CE Hamiltonian (eq 12) under the condition of long T1e, where the steady-state polarization difference between the electron spins drops below that of the nuclear Boltzmann polarization. Clearly, a larger ωee leading to an adiabatic e−e crossing would be a way to reduce the depletion of nuclear spin polarization (see Figure S6 for more details). Recently, Lund et al. experimentally elucidated that for similar radical systems and comparable spin concentrations, nuclear depolarization is dominantly modulated by T1e, with longer T1e leading to greater nuclear depolarization. Further, it was shown that the depolarization effect can be significantly attenuated by purposefully shortening T1e (without affecting T1n) by adding paramagneticrelaxation-agents, e.g., Gd-chelate complexes.33 Using a mixture of narrow and broad radicals with distinct isotropic g-tensors will also circumvent this conundrum. To this end, mixed radicals, which exhibit a truncated-CE54 would be the most favorable radical system to minimize depolarization under MAS, as it has been shown recently that one of the two electron spins always remains fully polarized owing to its fast electron spin T1e relaxation property (this is demonstrated in SI section B2 in greater detail). In this Perspective, we have described the polarization, depolarization, and oversaturation features of CE-DNP, aided by the theory of an effective Hamiltonian and experimentally supported by ELDOR data. We recapitulated the knowledge in the literature that the CE mechanism relies on the intrinsic dynamics of the spin system, while the role of low-power μw irradiation is merely to generate a large electron spin polarization differential by selectively saturating one of the electron spin populations. However, if the EPR line is oversaturated under conditions of high μw power and/or broad chirp-bandwidth, large spectral diffusion supervenes with increased e−e coupling (high concentration) and/or slow relaxation (low temperature) conditions, decreasing the net polarization difference. This in turn can decrease the CE-DNP enhancement, as highlighted in this Perspective. Here, we also discussed the advantage of frequency-modulated shaped-pulses (linear-chirp) that increase the net ΔPe by recruiting more

electron spins into the operation, which is beneficial especially under conditions of low spectral diffusion. Nuclear spin depolarization is another salient feature of the CE mechanism, leading to a reverse polarization transfer, i.e., from nuclear spins to electron spins under conditions of MAS-induced nonadiabatic polarization mixing (at the energy-crossings) between electron spins with long T1e. All examples presented here share the same underlying mechanism but are effective at different steady-state electron spin polarizations, as probed and verified by pump−probe ELDOR experiments under DNP conditions. Using the same perspective, it can be envisaged that mixed biradicals with different relaxation rates (a slow and a fast relaxing electron spin combination) will lead to larger ΔPe, as only one of the two electron spins can be selectively saturated, and therefore lead to higher CE enhancement, especially under MAS conditions. We have analytically derived the CE Hamiltonian using an elegant, simple, and effective operator-based Floquet theoretical framework that, in this form, has not been derived and presented before in the literature. The main advantage of using Floquet theory is for solving the multimodal time-dependent Hamiltonian, e.g. as applied to DNP under multifrequency applications or time-optimized DNP experiments, that the AHT cannot solve. With the advent of high-power μw amplifier technology,55 coherent EPR manipulation that relies on time-modulated μw pulse schemes will become increasingly viable for highly efficient DNP enhancement and electron− nuclear decoupling under DNP conditions.56,57 Given this trajectory, the real benefit of the operator-based Floquet framework will become apparent in time-dependent DNP experiments, which represent exciting future developments.



ASSOCIATED CONTENT

* Supporting Information S

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.jpclett.8b02834. Experimental and numerical results and methods, matlab code, and Spin-Evolution code (PDF)



AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. ORCID

Songi Han: 0000-0001-6489-6246 Notes

The authors declare no competing financial interest. Biographies Asif Equbal received his BS-MS degree from IISER Mohali, India in 2013. He then pursued his Ph.D. (2013−2016) from iNANO, Aarhus University, Denmark under the supervision of Prof. Niels Christian Nielsen on development of a heteronuclear spin decoupling pulse scheme under MAS. He is currently working as a post doctoral fellow in the lab of Prof. Songi Han, at University of California Santa Barbara, on development of dynamic nuclear polarization enhancement techniques.

The real benefit of the operatorbased Floquet framework will become apparent in time-dependent DNP experiments, which represent exciting future developments.

Alisa Leavesley completed her Ph.D. (2013−2018) at the University of California Santa Barbara under the supervision of Prof. Songi Han. Her Ph.D. work focused on the development of solid-state sourcebased high field EPR and DNP instrumentation and the study of radical and electron-spin clustering on DNP mechanisms and 556

DOI: 10.1021/acs.jpclett.8b02834 J. Phys. Chem. Lett. 2019, 10, 548−558

The Journal of Physical Chemistry Letters

Perspective

(8) Lilly Thankamony, A. S.; Wittmann, J. J.; Kaushik, M.; Corzilius, B. Dynamic nuclear polarization for sensitivity enhancement in modern solid-state NMR. Prog. Nucl. Magn. Reson. Spectrosc. 2017, 102−103, 120−195. (9) Hu, K.-N.; Yu, H.-h.; Swager, T. M.; Griffin, R. G. Dynamic nuclear polarization with biradicals. J. Am. Chem. Soc. 2004, 126, 10844−10845. (10) Sauvée, C.; Rosay, M.; Casano, G.; Aussenac, F.; Weber, R. T.; Ouari, O.; Tordo, P. Highly efficient, water-soluble polarizing agents for dynamic nuclear polarization at high frequency. Angew. Chem. 2013, 125, 11058−11061. (11) Sauvée , C.; Casano, G.; Abel, S.; Rockenbauer, A.; Akhmetzyanov, D.; Karoui, H.; Siri, D.; Aussenac, F.; Maas, W.; Weber, R. T.; et al. Tailoring of Polarizing Agents in the bTurea Series for Cross-Effect Dynamic Nuclear Polarization in Aqueous Media. Chem. - Eur. J. 2016, 22, 5598−5606. (12) Hwang, C. F.; Hill, D. A. Phenomenological model for the new effect in dynamic polarization. Phys. Rev. Lett. 1967, 19, 1011−1014. (13) Hu, K.-N.; Debelouchina, G. T.; Smith, A. A.; Griffin, R. G. Quantum mechanical theory of dynamic nuclear polarization in solid dielectrics. J. Chem. Phys. 2011, 134, 125105−125119. (14) Hovav, Y.; Feintuch, A.; Vega, S. Theoretical aspects of dynamic nuclear polarization in the solid state−the cross effect. J. Magn. Reson. 2012, 214, 29−41. (15) Thurber, K. R.; Tycko, R. Theory for cross effect dynamic nuclear polarization under magic-angle spinning in solid state nuclear magnetic resonance: the importance of level crossings. J. Chem. Phys. 2012, 137, 084508−084514. (16) Mentink-Vigier, F.; Akbey, U.; Oschkinat, H.; Vega, S.; Feintuch, A. Theoretical aspects of magic angle spinning-dynamic nuclear polarization. J. Magn. Reson. 2015, 258, 102−120. (17) Hovav, Y.; Feintuch, A.; Vega, S. Theoretical aspects of dynamic nuclear polarization in the solid state−the solid effect. J. Magn. Reson. 2010, 207, 176−189. (18) Haeberlen, U.; Waugh, J. Coherent averaging effects in magnetic resonance. Phys. Rev. 1968, 175, 453−467. (19) Waugh, J.; Huber, L.; Haeberlen, U. Approach to highresolution NMR in solids. Phys. Rev. Lett. 1968, 20, 180−183. (20) Shirley, J. H. Solution of the Schr̈odinger equation with a Hamiltonian periodic in time. Phys. Rev. 1965, 138, B979−987. (21) Ramesh, R.; Krishnan, M. S. Effective Hamiltonians in Floquet theory of magic angle spinning using van Vleck transformation. J. Chem. Phys. 2001, 114, 5967−5973. (22) Ernst, M.; Samoson, A.; Meier, B. H. Decoupling and recoupling using continuouswave irradiation in magic-angle-spinning solid-state NMR: a unified description using bimodal Floquet theory. J. Chem. Phys. 2005, 123, 064102−064110. (23) Leskes, M.; Madhu, P. K.; Vega, S. Floquet theory in solid-state nuclear magnetic resonance. Prog. Nucl. Magn. Reson. Spectrosc. 2010, 57, 345−380. (24) Hartmann, S.; Hahn, E. Nuclear double resonance in the rotating frame. Phys. Rev. 1962, 128, 2042−2053. (25) Jain, S. K.; Mathies, G.; Griffin, R. G. Off-resonance NOVEL. J. Chem. Phys. 2017, 147, 164201−164213. (26) Lewandowski, J. R.; De Paëpe, G.; Griffin, R. G. Proton assisted insensitive nuclei cross polarization. J. Am. Chem. Soc. 2007, 129, 728−729. (27) De Paëpe, G.; Lewandowski, J. R.; Loquet, A.; Böckmann, A.; Griffin, R. G. Proton assisted recoupling and protein structure determination. J. Chem. Phys. 2008, 129, 245101−245121. (28) Siaw, T. A.; Fehr, M.; Lund, A.; Latimer, A.; Walker, S. A.; Edwards, D. T.; Han, S.-I. Effect of electron spin dynamics on solidstate dynamic nuclear polarization performance. Phys. Chem. Chem. Phys. 2014, 16, 18694−18706. (29) Hovav, Y.; Feintuch, A.; Vega, S.; Goldfarb, D. Dynamic nuclear polarization using frequency modulation at 3.34 T. J. Magn. Reson. 2014, 238, 94−105. (30) Kaminker, I.; Han, S. Amplification of Dynamic Nuclear Polarization at 200 GHz by Arbitrary Pulse Shaping of the Electron

efficiency. She is currently working at Thomas Keating Ltd. to design custom quasi optical-based EPR and DNP instruments. Sheetal Kumar Jain received his Ph.D. from Aarhus university under the supervision of Prof. Niels Christian Nielsen in 2014. His Ph.D. was focused on development of cross-polarization methods for solidstate NMR spectroscopy and applications. In 2014, he joined the group of Prof. Robert G. Griffin at Massachusetts Institute of Technology (MIT), U.S., where he worked on developing methods for pulsed DNP. Currently, Sheetal is a postdoc in the Han group at University of California Santa Barbara. His current research is centred around DNP/NMR methods and applications. Songi Han received her Doctoral Degree in Natural Sciences (Dr.rer.nat) from Aachen University of Technology (RWTH), Germany, in 2001. She pursued her postdoctoral studies at the University of California Berkeley sponsored by the Feodor Lynen Fellowship of the Alexander von Humboldt Foundation. Dr. Han joined the faculty at University of California Santa Barbara (UCSB) in 2004, received tenure in 2010, and was promoted to full professor in 2012. She is currently a Professor in the Department of Chemistry and Biochemistry and the Department of Chemical Engineering at UCSB. She is a recipient of the 2008 Packard Fellowship for Science and Engineering, the 2010 Dreyfus-Teacher Scholar Award, the 2011 NIH Innovator Award, the 2015 Bessel Prize of the Alexander von Humboldt Foundation, and the 2018 recipient of the Biophysical Society Innovator Award. Her research group focuses on mechanistic studies of DNP as well as broadening the application scope of DNP and EPR.



ACKNOWLEDGMENTS We thank Yuanxin Li for fruitful discussion on CE-DNP. This work was supported by the National Science Foundation (CHE #1505038 to S.H.), the National Institute of Health (NIBIB #1R21EB022731 and #R21GM103477 to S.H.), and Binational Science Foundation (Grant #2014149 to S.H.). The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.



REFERENCES

(1) Overhauser, A. W. Polarization of nuclei in metals. Phys. Rev. 1953, 92, 411−415. (2) Carver, T. R.; Slichter, C. P. Polarization of nuclear spins in metals. Phys. Rev. 1953, 92, 212−213. (3) Wind, R.; Duijvestijn, M.; Van Der Lugt, C.; Manenschijn, A.; Vriend, J. Applications of dynamic nuclear polarization in 13C NMR in solids. Prog. Nucl. Magn. Reson. Spectrosc. 1985, 17, 33−67. (4) Ardenkjær-Larsen, J. H.; Fridlund, B.; Gram, A.; Hansson, G.; Hansson, L.; Lerche, M. H.; Servin, R.; Thaning, M.; Golman, K. Increase in signal-to-noise ratio of > 10,000 times in liquid-state NMR. Proc. Natl. Acad. Sci. U. S. A. 2003, 100, 10158−10163. (5) Rosay, M.; Zeri, A.-C.; Astrof, N. S.; Opella, S. J.; Herzfeld, J.; Griffin, R. G. Sensitivity-enhanced NMR of biological solids: Dynamic nuclear polarization of Y21M fd bacteriophage and purple membrane. J. Am. Chem. Soc. 2001, 123, 1010−1011. (6) Bajaj, V.; Farrar, C.; Hornstein, M.; Mastovsky, I.; Vieregg, J.; Bryant, J.; Elena, B.; Kreischer, K.; Temkin, R.; Griffin, R. Dynamic nuclear polarization at 9 T using a novel 250 GHz gyrotron microwave source. J. Magn. Reson. 2011, 213, 404−409. (7) Ardenkjaer-Larsen, J.-H.; Boebinger, G. S.; Comment, A.; Duckett, S.; Edison, A. S.; Engelke, F.; Griesinger, C.; Griffin, R. G.; Hilty, C.; Maeda, H.; et al. Facing and overcoming sensitivity challenges in biomolecular NMR spectroscopy. Angew. Chem., Int. Ed. 2015, 54, 9162−9185. 557

DOI: 10.1021/acs.jpclett.8b02834 J. Phys. Chem. Lett. 2019, 10, 548−558

The Journal of Physical Chemistry Letters

Perspective

Spin Saturation Profile. J. Phys. Chem. Lett. 2018, 9, 3110−3115 PMID: 29775537. . (31) Thurber, K. R.; Tycko, R. Perturbation of nuclear spin polarizations in solid state NMR of nitroxide-doped samples by magic-angle spinning without microwaves. J. Chem. Phys. 2014, 140, 184201−184211. (32) Mentink-Vigier, F.; Paul, S.; Lee, D.; Feintuch, A.; Hediger, S.; Vega, S.; De Paëpe, G. Nuclear depolarization and absolute sensitivity in magic-angle spinning cross effect dynamic nuclear polarization. Phys. Chem. Chem. Phys. 2015, 17, 21824−21836. (33) Lund, A.; Equbal, A.; Han, S. Tuning nuclear depolarization under MAS by electron T 1e. Phys. Chem. Chem. Phys. 2018, 20, 23976−23987. (34) Vega, S. Fictitious spin 1/2 operator formalism for multiple quantum NMR. J. Chem. Phys. 1978, 68, 5518−5527. (35) Bayro, M. J.; Huber, M.; Ramachandran, R.; Davenport, T. C.; Meier, B. H.; Ernst, M.; Griffin, R. G. Dipolar truncation in magicangle spinning NMR recoupling experiments. J. Chem. Phys. 2009, 130, 114506. (36) Scholz, I.; Meier, B. H.; Ernst, M. Operator-based triple-mode Floquet theory in solidstate NMR. J. Chem. Phys. 2007, 127, 204504. (37) Boender, G.; Vega, S.; De Groot, H. A physical interpretation of the Floquet description of magic angle spinning nuclear magnetic resonance spectroscopy. Mol. Phys. 1998, 95, 921−934. (38) Gast, P.; Mance, D.; Zurlo, E.; Ivanov, K.; Baldus, M.; Huber, M. A tailored multifrequency EPR approach to accurately determine the magnetic resonance parameters of dynamic nuclear polarization agents: application to AMUPol. Phys. Chem. Chem. Phys. 2017, 19, 3777−3781. (39) Hovav, Y.; Shimon, D.; Kaminker, I.; Feintuch, A.; Goldfarb, D.; Vega, S. Effects of the electron polarization on dynamic nuclear polarization in solids. Phys. Chem. Chem. Phys. 2015, 17, 6053−6065. (40) Armstrong, B. D.; Edwards, D. T.; Wylde, R. J.; Walker, S. A.; Han, S. A 200 GHz dynamic nuclear polarization spectrometer. Phys. Chem. Chem. Phys. 2010, 12, 5920−5926. (41) Siaw, T. A.; Leavesley, A.; Lund, A.; Kaminker, I.; Han, S. A versatile and modular quasi optics-based 200 GHz dual dynamic nuclear polarization and electron paramagnetic resonance instrument. J. Magn. Reson. 2016, 264, 131−153. (42) Kaminker, I.; Barnes, R.; Han, S. Arbitrary waveform modulated pulse EPR at 200 GHz. J. Magn. Reson. 2017, 279, 81−90. (43) Leavesley, A.; Kaminker, I.; Han, S.Versatile DNP Hardware with Integrated EPR. In Dynamic Nuclear Polarization Handbook, 2018; ISBN 9781119441649. (44) Hu, K.-N.; Bajaj, V. S.; Rosay, M.; Griffin, R. G. High-frequency dynamic nuclear polarization using mixtures of TEMPO and trityl radicals. J. Chem. Phys. 2007, 126, 044512−044517. (45) Hovav, Y.; Kaminker, I.; Shimon, D.; Feintuch, A.; Goldfarb, D.; Vega, S. The electron depolarization during dynamic nuclear polarization: measurements and simulations. Phys. Chem. Chem. Phys. 2015, 17, 226−244. (46) Leavesley, A.; Shimon, D.; Siaw, T. A.; Feintuch, A.; Goldfarb, D.; Vega, S.; Kaminker, I.; Han, S. Effect of electron spectral diffusion on static dynamic nuclear polarization at 7 T. Phys. Chem. Chem. Phys. 2017, 19, 3596−3605. (47) Shimon, D.; Hovav, Y.; Feintuch, A.; Goldfarb, D.; Vega, S. Dynamic nuclear polarization in the solid state: a transition between the cross effect and the solid effect. Phys. Chem. Chem. Phys. 2012, 14, 5729−5743. (48) Mathies, G.; Caporini, M. A.; Michaelis, V. K.; Liu, Y.; Hu, K.N.; Mance, D.; Zweier, J. L.; Rosay, M.; Baldus, M.; Griffin, R. G. Efficient dynamic nuclear polarization at 800 MHz/527 GHz with trityl-nitroxide biradicals. Angew. Chem., Int. Ed. 2015, 54, 11770− 11774. (49) Leavesly, A.; Jain, S.; Kamniker, I.; Zhang, H.; Rajca, S.; Rajca, A.; Han, S. Maximizing NMR signal per unit time by facilitating the e−e−n cross effect DNP rate. Phys. Chem. Chem. Phys. 2018, 20, 27646−27657.

(50) Garwood, M.; DelaBarre, L. The return of the frequency sweep: designing adiabatic pulses for contemporary NMR. J. Magn. Reson. 2001, 153, 155−177. (51) Hovav, Y.; Feintuch, A.; Vega, S.; Goldfarb, D. Dynamic nuclear polarization using frequency modulation at 3.34 T. J. Magn. Reson. 2014, 238, 94−105. (52) Landau, L. D. Zur theorie der Energieubertragung II. Z. Sowjetunion 1932, 2, 46−51. (53) Zener, C. Non-adiabatic crossing of energy levels. Proc. R. Soc. London, Ser. A 1932, 137, 696−702. (54) Equbal, A.; Li, Y.; Leavesley, A.; Huang, S.; Rajca, S.; Rajca, A.; Han, S. Truncated Cross Effect Dynamic Nuclear Polarization: An Overhauser Effect Doppelganger. J. Phys. Chem. Lett. 2018, 9, 2175− 2180. (55) Soane, A. V.; Shapiro, M. A.; Jawla, S.; Temkin, R. J. Operation of a 140-GHz gyroamplifier using a dielectric-loaded, severless confocal waveguide. IEEE Trans. Plasma Sci. 2017, 45, 2835−2840. (56) Saliba, E. P.; Sesti, E. L.; Scott, F. J.; Albert, B. J.; Choi, E. J.; Alaniva, N.; Gao, C.; Barnes, A. B. Electron decoupling with dynamic nuclear polarization in rotating solids. J. Am. Chem. Soc. 2017, 139, 6310−6313. (57) Jain, S. K.; Siaw, T. A.; Equbal, A.; Wilson, C. B.; Kaminker, I.; Han, S. Reversal of Paramagnetic Effects by Electron Spin Saturation. J. Phys. Chem. C 2018, 122, 5578−5589.

558

DOI: 10.1021/acs.jpclett.8b02834 J. Phys. Chem. Lett. 2019, 10, 548−558