Contact: Paul Preuss, [email protected]

A way to acquire chemical information with magnetic fields a million times weaker than those used in typical nuclear magnetic resonance (NMR) spectroscopy has been developed by a team led by John Clarke and Alexander Pines. Clarke is a professor of physics and Pines a professor of chemistry at UC Berkeley, and both are members of Berkeley Lab’s Materials Sciences Division.

NMR and its near relative, magnetic resonance imaging (MRI), are essential tools of scientific research and medical diagnosis. Yet NMR is often limited to samples that can be placed inside the bore of a big, high-field magnet. Because very strong magnetic fields of a tesla or more (a tesla is some 20,000 times the Earth’s magnetic field strength) must be exquisitely adjusted to reduce variations in intensity, NMR apparatus is expensive and cumbersome.

The secret of low-field success, says Robert McDermott, a graduate student in Clarke’s laboratory and an author of a recent paper in Science describing the method, is to use a superconducting quantum interference device, or SQUID, the most sensitive magnetic field detector ever devised, along with a technique called prepolarization that aligns spinning nuclei.

Pushing the state of the art

SQUIDs have been used in NMR measurements since the 1980s, but mostly for solid samples at extremely low temperatures. To analyze liquids at room temperature, the Clarke-Pines team heated their samples in an insulated chamber surrounded by the SQUID’s pick-up coils. The SQUID itself — a tiny loop of superconductor interrupted at two points by weak links, called Josephson junctions — operated in a bath of liquid helium.

A superconducting quantum interference device, or SQUID, is a tiny loop of superconducting material interrupted by narrow gaps called Josephson junctions.

Other recent SQUID experiments done at room temperature have employed magnetic fields of several thousandths of a tesla (milliteslas). The Clarke-Pines team’s measurement field was less than two millionths of a tesla (microteslas), a small fraction of the Earth’s magnetic field strength.

All NMR depends on the fact that some kinds of spinning nuclei generate their own magnetic fields. These can be lined up by an external magnetic field, then knocked off axis by a burst of radio waves. The rate at which each kind “wobbles” (precesses) is unique; for example, a hydrogen nucleus precesses four times faster than a carbon-13 nucleus. A detector can pinpoint the type of element by tuning to its precession frequency, known as the Larmor frequency.

Lines in an NMR spectrum reveal more than just different elements. Nearby electrons can alter precession frequency, causing a “chemical shift” — moving the signal or splitting it into separate lines in an NMR spectrum. Chemical shifts point to particular compounds, as in the arrangement of hydrogen and carbon atoms in alcohols.

“Chemical shift grows linearly with field strength,” says Andreas Trabesinger, a postgraduate fellow in the Pines laboratory and another of the Science paper’s authors, explaining another reason why NMR uses strong magnets. “The higher the field, the higher the Larmor frequency, and the stronger the signal.”

Detectors tuned to Larmor frequencies are not the only way to distinguish nuclear magnetic signals. Such detectors report the frequency of change in magnetic flux (the number of magnetic field lines through a surface), while SQUIDs can detect magnetic flux directly, sensing the magnetic field generated by even a slowly precessing nucleus. The resulting signal is weak but extremely sharp: the lower the magnetic field, the narrower the NMR line, yielding a signal-to-noise ratio far superior to that of high-field NMR.

“SQUIDs are frequency-independent,” says McDermott. “To achieve low-field NMR, we realized we could play this trick of operating with an untuned detector.”

An untuned SQUID is not the whole answer to measuring NMR with extremely weak fields. To detect NMR signals at all there must be net polarization, an overall magnetization resulting from the existence of a few more “spin-up” nuclei than “spin-down” nuclei. Even in high fields, spin-up nuclei hold the majority by only about 1 in 100,000; the weaker the magnetic field, the weaker the polarization.

Some spinning nuclei have magnetic moments — not unlike toy bar magnets — and orient themselves with the field lines of an external magnetic field.

The experimenters met the challenge by using two magnetic fields. One, of about two millitesla, first prepolarizes the sample then is quickly shut off. The measurement field — at a microtesla, a thousand times weaker — is applied at right angles.

“We get the best of both worlds,” Trabesinger remarks, “high polarization from the higher field, plus narrow lines from the low measurement field.”

Abruptly turning off the polarizing field creates its own problem, however: the precessing nuclei are pushed out of phase. To correct this, the measurement field is reversed. Wandering precessions instantly reverse direction and “wobble backwards” into maximum polarization.

A final disadvantage of low-field NMR is that chemical shifts, caused by the screening effects of nearby electrons, vanish in microtesla fields. But a different kind of nuclear signature, imposed by the quantum states of shared electrons in chemical bonds, persists. This scalar coupling (or J coupling), says Trabesinger, “is intrinsic. It doesn’t depend on the external magnetic field, so it doesn’t change in a low field.”

“Information about chemical bonds can be resolved even in incredibly inhomogeneous magnetic fields,” adds McDermott. Scalar coupling opens the door to chemical sampling even when magnetic signals vary, as in living organisms or the fluid-filled porous rocks measured in oil-well logging.

Spying on bonds

Unlike a traditional tuned NMR system, an untuned SQUID is “broadband” — it can simultaneously detect different kinds of nuclei precessing at different frequencies. The Clarke-Pines team detected the nuclei of phosphorus and hydrogen (protons) in the same measurement of phosphoric acid.

Moreover, “through detection of chemical bonds, we can monitor chemical reactions as they happen,” McDermott says. At low fields the NMR spectrum of phosphoric acid and methanol mixed with water shows only a single proton peak, while the spectrum of trimethyl phosphate — the reaction product of phosphoric acid and methanol — shows the proton peak split in two by the scalar coupling of the compound’s phosphorus-hydrogen bonds. With labeled “spy nuclei,” a low-field “bond detector” might track the formation of other bonds like these.

When mixed with water, phosphoric acid and methanol show only a single proton peak in a low-field NMR spectrum, but for trimethyl phosphate — the reaction product of phosphoric acid and methanol — the proton peak is split in two by the scalar coupling of bonds between phosphorus and hydrogen.

The low-field SQUID technique could also be used in MRI. Since the precession rate of spinning nuclei depends on the strength of the external magnetic field, their spatial distribution can be plotted to create an image. MRI typically has depended on strong magnetic fields that fall off rapidly; in medical diagnosis machinery, this has meant housing the magnet in a large torus, into which the patient is slid on a table.

Not only are the big machines expensive and immobile, their high fields are dangerous to patients with pacemakers or metal prosthetic inserts. Moreover, Trabesinger remarks, “it’s not easy to convince everybody to get into these monsters. Some get a sudden attack of claustrophobia.”

A low-field detector, on the other hand, could achieve high resolution with a small field gradient, and the detector’s portable small coil could be moved over the patient. Among other advantages, this would allow patients to assume a more natural posture. “There are lots of physiological situations where lying down might not be the best position to measure,” Trabesinger says, “for example, the stomach’s position while swallowing.”

Low-field NMR will never replace traditional high-field methods, because only high fields make available the powerful diagnostic capabilities of chemical shift. In the right circumstances, however, low-field NMR could give limited but potentially very useful information about the scalar coupling of many molecules.

One application that excites both McDermott and Trabesinger is the possibility of studying the brain by combining MRI with direct detection of chemical bonds, yielding a unique way to investigate the mysteries of brain chemistry. “It would not be easy,” says McDermott, “but the new low-field SQUID detector may make this possible.”

“Liquid-state NMR and scalar couplings in microtesla magnetic fields,” by Robert McDermott, Andreas H. Trabesinger, Michael Mück, Erwin L. Hahn, Alexander Pines, and John Clarke, appeared in the 22 March 2002 issue of Science.

Additional information: