Dynamic Nuclear Polarization (DNP) at 9.4 Teslawith a 30mW microwave source

Kent R. Thurber, Wai-Ming Yau, and Robert Tycko

Laboratory of Chemical Physics, National Institute of Diabetes and Digestive and Kidney Diseases, National Institutes of Health, Bethesda, MD 20892-0520

Why Dynamic Nuclear Polarization (DNP)?

1.10 to >100x signal increase

Big benefit to often signal to noise limited NMR experiments

Why Dynamical Nuclear Polarization now?

Aβ 14-23 amino acid sequence: HQKLVFFAED2.7 mg of sample, uniform 13C and 15N labeled at V18, A21Pre-formed fibrils [4] hydrated with 25ul of 160uM DyEDTA in glycerol/water (3:2 ratio by volume).

2D 13C-13C spectrum with 500 ms 13C spin diffusion mixing period

Since DNP tends to work best at low temperatures, the ability to do MAS at 25 K leads to the idea of adding DNP to CryoMAS.

Cross peaks between A21 Cα and V18 Cα and Cβindicate close contact of A21 and V18.

25 K, 6.7 kHz spin, 2048 total scans for 2D, 1H T1 = 4 sec, repeat rate 6 sec, time required = 3.5 hours each

Dynamic Nuclear Polarization: Transfer Electron spin polarization to Nuclear spin to increase NMR signal.

Electrons have 600 times larger magnetic moment, hence larger polarization than nuclei.In a 9.4 Tesla magnetic field at 25 K,

Polarization1H nuclei 400 MHz 4 x 10-4

= 0.02 KelvinElectrons 264 GHz 0.25

= 13 Kelvin

Electrons have significant polarization in this temperature/field range.Need to apply magnetic field at electron spin resonance (264 GHz)

264 GHz (1mm wavelength) is an awkward frequency –high for microwaves, low for optical.

DNP Mechanism at high field

Some possible methods for DNP:

1. Apply microwaves at sum or difference of electron and nuclear frequency [Solid state effect] Forbidden transition: one photon to change two spin states,

only possible because the electron-nuclear interaction perturbs the eigenstates. This perturbation weakens at high field because the energy difference of the

single spin states gets larger relative to the electron-nuclear interaction.

2. Cross polarization, or adiabatic passage effects.

3. Use multiple electron spins close together so electron-electron interactions can be significant.

Conserve energy with nuclear spin flip by having electron-electron dipolar coupling [Thermal mixing]or electron-electron g-factor difference [Cross effect] = NMR frequency.

Design of DNP/ESR spectrometer for 9.4 Tesla:

DNP with a frequency-tunable CW microwave source. Based on solid-state microwave source: 264 GHz, ~10 GHz tunable range

Small, simple, and relatively inexpensiveNo requirement for swept field magnetLower power (30mW) than gyrotron or vacuum electronics

Detect ESR with no change to equipment setup.

NMR magnet with optical plate for 264 GHz microwave control and transmission

Quasi-optical transmission: (quasi- because 1mm wavelength is finite relative to beam cross-section size.)Metal waveguides at the fundamental mode are too lossy at this frequency. Instead, use mirrors, wire grids (as polarizers), and oversize (multi-mode) corrugated waveguides.

Polarization control to allow all of the power to be transmittedin the circular polarization that can be absorbed by the electron spins.

Spectrometer setup: Optical plate(shown not under NMR magnet)

132

5

4

1. Microwave source with horn.

2. Polarizer (wire grid)

3. Interferometer for polarization control.

4. Mirror to reflect microwaves into vertical corrugated waveguide.

5. Detector for ESR with horn.

4. Mirror to reflect microwaves into vertical corrugated waveguide.

6. Corrugated waveguide

Corrugated waveguide leads to bottom window of vacuum cryostat

4

6

Bottom of NMR magnet

Optical plate(under NMR magnet)

NMR probe without radiation shield, next to its vacuum jacket

He-cooled NMR probe (non-spinning)

Vacuum window for entry of microwaves on bottom of cryostat(Plastic, HDPE)

7. Brass tube to convey microwaves to sample.

8. Modulation coil for AC magnetic field in z-axis for ESR detection.

9. Copper parts hold sample container.

10. RF coil lead. RF coil itself is barely visible at center of modulation coil.

Sample is at center of RF coil. Sample is sealed in a Teflon cup with a sapphire rod attached to Cu for thermal conductivity.

Close-up of NMR probe end

8

7

910

Electron spin dopants based on TEMPO nitroxide

NO•

NH2

N•O

OOH

N O•

NH

N•O

OOH

N O•

N O•OH

N

O

4-amino-TEMPO TOTAPOL“Double-Tempo”

K.-N. Hu, C. Song, H. Yu, T.M. Swager, and R.G. Griffin, JCP 128, 052302 (2008)

DOTOPA-TEMPO“Triple-Tempo”

Wai-Ming Yau

Kan Hu

ESR detection by microwave power absorption

4-Amino TEMPO(free radical)

14N nucleus (S=1) near electron causes three level hyperfine splitting

ESR signal is detected as the derivative of the microwave absorption.

The microwaves pass through the sample, are reflected back through the sample, and down to the microwave optics. When the microwaves are circularly polarized, the returning microwaves end up at the detector.The modulation coil provides an AC (6100 Hz) z-axis magnetic field (2 Gauss = 6 MHz). The signal from the microwave detector is fed into a lock-in amplifier, and the AC absorption signal (= the derivative of the absorption) is detected.

-25

-20

-15

-10

-5

0

5

10

15

264 264.2 264.4 264.6 264.8

ESR absorption derivative signal6ul of 10mM 4-Amino TEMPO in toluene, room temperature,

microwave polarization is absorbed circular

in phase lockin signalout of phase lockin signal

ESR

abs

orpt

ion

deriv

ativ

e [a

rb. u

nits

]

Microwave frequency [GHz]

-0.2

-0.1

0

0.1

0.2

0.3

0.4

0.5

264 264.2 264.4 264.6 264.8

ESR lineshape from integration of derivative absorption signal

4-Amino TEMPO sampleIn

tegr

al o

f abs

orpt

ion

deriv

ativ

e [a

rb. u

nits

]

Microwave frequency [GHz]

1H NMR signal enhancement by DNP

-10

0

10

20

30

40

0 20 40 60 80 100

1H NMR echo using DNP with field modulation, and without any DNP

16K, 10ul sample, 40mM 4-amino TEMPO25/75 mol% glycerol/water

1H echo, using DNP with field modulation1H echo, no DNP

1H N

MR

ech

o [a

rb. u

nits

]

Time [us]

264 GHz CW microwave irradiation. Field modulation used during DNP:

20 Gauss rms = 56 MHz for electron spins at 6100Hz

-1

0

1

2

3

4

0 20 40 60 80 100

x10 expanded y scale

Time [us]

DNP enhancement of 1H NMR, and ESRas a function of microwave frequency

-20

-10

0

10

20

30

40

263.6 263.8 264 264.2 264.4 264.6 264.8 265 265.2

Dynamical Nuclear Polarization (DNP) at 16 K

10ul sample (glycerol/water 25/75 mol% doped with 40mM 4-amino TEMPO)

Signal enhancement relative to thermal polarization at 16 K

Sign

al s

ize

rela

tive

to th

erm

al =

1

Microwave frequency [GHz]

-50

0

50

100

150

200

250

263.6 263.8 264 264.2 264.4 264.6 264.8 265 265.2

Electron Spin Resonance

Same sample and conditions as for DNP

ES

R a

bsor

ptio

n lin

esha

pe(in

tegr

al, a

rb. u

nits

)

Microwave frequency [GHz]

16K, 10ul samples of 25/75 mol% glycerol/water

DNP microwave frequency dependence

DNP lineshape is not strongly dependent on having multiple Tempo radicals linked together.Signal increase from DNP is strongly affected.

-80

-60

-40

-20

0

20

40

60

80

263 263.5 264 264.5 265 265.5

"Triple-Tempo" 20mM

Totapol "Double-Tempo" 20mM

4-amino-Tempo 40mMD

NP

1H

Sig

nal r

atio

to th

erm

al

Microwave Frequency [GHz]

40mM 4-amino-Tempo, 16K, 10ul sample, 25/75 mol% glycerol/water

Time dependence of DNP polarizationand T1 (thermal polarization)

Polarization time for DNP is important, not just signal increase.Not good to have a big signal that takes forever to polarize.Our conditions show DNP polarization time comparable to T1 (thermal). Time constants vary strongly with dopant, its concentration, and temperature.

160mM 4-amino-Tempo, 35K, 10ul sample, 25/75 mol% glycerol/water

0

100

200

300

400

500

600

700

800

0

2

4

6

8

10

12

14

0 100 200 300 400 500 600

TDNP

= DNP Polarization buildup(microwaves on)

T1n

= recovery after saturation(microwaves off = no DNP)

DN

P 1

H e

cho

inte

gral

[ar

b. u

nits

]

1H echo integral [arb. units]

Time [s]

~130 s

0

10

20

30

40

50

60

70

80

0

0.5

1

1.5

2

2.5

3

3.5

4

0 5 10 15 20 25

DN

P 1

H e

cho

inte

gral

[ar

b. u

nits

]

1H echo integral [arb. units]

Time [s]

~3 s

Temp. Dopant SignalIncrease

BuildupTime, TDNP (s)

Signal/Noise in fixed time= Signal/Sqrt(TDNP)

80 K TripleTEMPO 30mM 10x 1.4 14Totapol 20mM 6x 8.3 4.1TEMPO 40mM 11x 50 1.8

35 K TripleTEMPO 30mM 26x 2.4 73Totapol 20mM 11x 18.5 13TEMPO 40mM 13x 90 3.6

16 K TripleTEMPO 30mM 59x 4.5 265Totapol 20mM 26x 41 52TEMPO 40mM 29x 126 15

Multiple linked Tempo radicals improve DNP

20mM Totapol “Double-Tempo”, 35K, 10ul sample, 25/75 mol% glycerol/water

Same data normalizedshowing similarity of frequency dependence

Bz field modulation at kHz frequency improves DNP significantly (up to 2.3x signal)

Slow field modulation has no effect. Timescale for benefit does not depend on size of field modulation or the microwave power level.

10 Gauss field modulation is equivalent to 28 MHz for the electrons

Dependence of DNP on field modulation frequency

Increase in DNP with kHz field modulation

6080

100120140160180200220

0 1000 2000 3000 4000 5000 6000

20 Gauss rms modulation10 Gauss rms modulation0.4 Microwave Power, 10 G rms modulation40 Gauss rms modulation

1H S

igna

l [a

rb. u

nits

]

Field modulation frequency [Hz]

0

0.2

0.4

0.6

0.8

1

1.2

0 500 1000 1500 2000 2500 3000

20 Gauss rms modulation10 Gauss rms modulation0.4 Microwave Power, 10 G rms modulation

Nor

mal

ized

1H

sig

nal i

ncre

ase

[arb

. uni

ts]

Field modulation frequency [Hz]

Simulation: Bloch equations with relaxation, sinusoidal modulation of magnetic field, and spectral diffusion.

Electron polarization averaged over complete modulation.

Slow electron spectral diffusion = narrow hole saturated in electron line by microwaves

0

0.2

0.4

0.6

0.8

1

-60 -40 -20 0 20 40 60

10 G rms

20 G rms

5 G rms

0.1 G rms

Aver

age

pola

rizat

ion

Field offset [G]

Modulation amplitude

Parameters: 10 kHz modulation frequency, T1e = 2 ms, T2e = 4 us, B1 = 0.03 G, spectral diffusion = 1017 Hz2/s

0

0.2

0.4

0.6

0.8

1

1.2

1.4

0 1000 2000 3000 4000 5000 6000

Elec

tron

spin

Sat

urat

ion

Inte

gral

Modulation Frequency (Hz)

Electron spin saturation regionTotal electron spin saturation = DNPas a function of modulation frequency

0

0.2

0.4

0.6

0.8

1

1.2

40mM 4-amino TEMPO, 35 K

Triple TEMPO, 35 K

Totalpol, 35 K

Simulation for 40mM 4-amino TEMPO

0 2000 4000 6000

1 H D

NP

enha

ncem

ent

norm

aliz

ed

Field modulation frequency [Hz]

No adjustable parameters for normalized frequency dependence.Parameters: some extrapolated from 140 GHz.

(Farrar, et al., J. of Chem. Phys. 2001.)Simulation has a good fit considering extrapolation of

parameters.Experiment: 135% increase (DNP more than double)

with field modulationSimulation: 200% increase (DNP triples)

What does simulation indicate will help DNP?

Modulate field or frequency faster than 1/T1e

Comparison between Experiment and Simulation

Parameters: T1e = 2 ms, T2e = 4 us, B1 = 0.03 G, spectral diffusion = 1017 Hz2/s

Large B1, modulation amplitude. Large spectral diffusion, long T1eT2e does not matter in this range

What about MAS? It involves g-anisotropy modulation at kHz frequencies.

Microwave power dependenceof DNP signal

400x increase in signal/noise in fixed time at 7K with DNP compared to 80K without DNP

100x increase in signal/noise in fixed time at 25K where we can definitely do MAS

0

100

200

300

400

500

0 20 40 60 80 100

Signal per unit timenormalized to 80K without DNP

Sign

al p

er u

nit t

ime

= 1H

ech

o in

tegr

al /

Sqr

t(Tbu

ildup

)Temperature (K)

0

20

40

60

80

100

120

0 0.2 0.4 0.6 0.8 1 1.2

20mM "Triple-Tempo"20mM Totapol "Double-Tempo"160mM 4-amino-Tempo320mM 4-amino-Tempo

DN

P 1

H S

igna

l (a

rb. u

nits

)

Microwave Power(normalized to max power ~30mW)

35K, 10ul sample, 25/75 mol% glycerol/water 30mM “Triple-Tempo”, 10ul sample, 25/75 mol% glycerol/water, pH 3 acetate buffer

Temperature dependenceof DNP signal

Conclusions

Improving Biomolecular Solids NMR by Increasing Signal to Noise:

With solid-state source at 264 GHz, Dynamic Nuclear Polarization (DNP) can provide at least 30x increased signal at 35 K,

on top of the benefit of low temperature (~1/T)

Goal: combine DNP with low-temperature MAS at T ~ 25 K.“Biomolecular solid state NMR with magic-angle spinning at 25 K,”JMR vol 195 (2008) 179.

Thanks to:Bernie HowderKan Hu Low-temperature MAS probe