FELIX BLOCH AND MAGNETIC RESONANCE

Erwin L. Hahn

Physics DepartmentUniversity of California

Berkeley, California 94720

Presented at the Felix Bloch Memorial Symposium*American Physical Society

April 25, 1984Washington D.C.

Among the versatile and fundamental contri-butions of Felix Bloch to physics, the science ofmagnetic resonance in condensed matter, alsointroduced independently by the Purcell Harvardgroup, was essentially the crowning achievementof his career. The background for the earlyinvestigations of the Stanford group in what wasthen called "nuclear induction" must be viewedin the context of Bloch's early scientific career.In his early youth it was unheard of to think ofmaking an assured living as a theoretical physi-cist, so a proper career to choose was engineer-ing, and he became a student of engineering inthe Institute of Technology in Zurich. He quicklybecame bored of these studies, and against theadvice of his professors switched to physics. Hewas aware that only fanatics do theoreticalphysics and was told that he would probablystarve at it if he was mediocre. As a youngphysicist, eager to capitalize on all the newphysics which could be solved by the new quan-tum mechanics, Bloch made a grand tour ofstudies at prominent centers of European physicsbefore World War II. He interacted with verita-ble Who's Who of 20th century physicists - Bohr,Schrodinger, Pauli, Debye, Heisenberg, Kram-ers, Fokker, Ehrenfest and Fermi. Very soonafterward in the early 1930's, Bloch acquired aneminent stature of his own in the physics of sol-ids. Beginning in 1928 to the time of his deathin 1983, his papers ranged through a remarkablerange of subjects. Leaving out his magneticresonance publications, they included radiation

*The author is grateful to Carson Jeffries andMartin Packard for access to source materialconcerning F. Bloch.

damping in quantum mechanics, the periodicityof electrons in crystals and metals, susceptibilityand conductivity in metals, ferromagnetism,charged particle stopping power, quantum elec-tro-dynamics, x-ray spectra, x-ray and Comptonscattering, Auger effect, beta decay and super-conductivity. What was most important inBloch's future development was his decision toleave Europe in 1934 and to begin his career atStanford where he eventually began to do exper-imental physics. The seed of his magnetic reso-nance or "nuclear induction" idea stemmed fromhis interest in the neutron. I quote from Bloch'swords in his magnetic resonance Nobel Lecture,as follows: "The idea that a neutral elementaryparticle should possess an intrinsic magneticmoment had a particular fascination for me — Itseemed important to furnish a direct experimen-tal proof for the existence of the magneticmoment of the free neutron." In 1936 Blochpublished a paper on the magnetic scattering ofneutrons in which he suggested that a beam ofslow neutrons in passing through iron wouldexperience a highly localized interaction with theiron atoms. He predicted that neutron spinswould become polarized as well as scattered,leading to the present day methods of neutronmagnetic scattering. His thinking about mag-netic resonance began by his conception of aresonance depolarization experiment in which apolarized neutron beam was passed throughspace with spins parallel to a strong constantmagnetic field H.

As acknowledged in 1938 by Rabi and col-laborators, Bloch apparently, along with Gorter,had the magnetic resonance principle in mind,with Bloch interested in carrying out neutronspin resonance. Of course Rabi first accom-plished the experiment in 1938 with a molecular

82 Bulletin of Magnetic Resonance

I a H S" •.a

* . ^ *

J

l£c

% C C . <c f ^

%** \J^ *>* /•»--•« A 4 .1* i.f < l,f ,s*i

- * •

+ 4

• - . *

Figure 1. Calculations of nuclear Boltzmann factors from original notes of F. Bloch.

Vol. 7, No. 2/3 83

«W*Pi- - ft .

{vll )

N

it.

"\"Li:

h S

Figure 2. First estimates of the nuclear induction signal from original notes of F. Bloch.

Figure 3. F. Bloch. H. Staub, and W. Hansen viewing a homemade oscilloscope in the early 50's.

84 Bulletin of Magnetic Resonance

beam. Bloch thereafter acknowledged Rabi'smagnetic resonance principle to flip the neutronspin, and in 1940 in collaboration with LuisAlvarez at Berkeley, the neutron moment wasmeasured to an accuracy of 1%. This accuracywas determined by flip coil measurements, whichcould not be better than 0.4%. Bloch's firstnotion, which he quickly abandoned as too awk-ward, was to improve the accuracy of this meas-urement by calibrating the magnetic field by amolecular beam technique. Bloch began to won-der whether it might be possible to measure themagnetic resonance of nuclei in ordinary con-densed matter? He was unaware that Gorterhad failed in 1936 and in 1942 to carry out suchan experiment in a crystal. This is not surpris-ing since Bloch was not a conscientious surveyorof the literature. He preferred to rediscover andwork things out ior himself - a marked charac-teristic of his independent personality. Nor washe aware of the experiments contemplated orinitially carried out on nuclear adsorption atHarvard.

Bloch began from first simple principles tocalculate the voltage that would be induced in aninductance by the precessing macroscopicmagnetization due to protons in 1 mL of water atroom temperature, subjected to a strong field Hand a perpendicular r.f. field H,. (See Figures 1and 2). He stated to me personally and to oth-ers, on querying him about his train of thoughtthen, that he was amazed how such a simplecalculation, taking into account a small Boltz-mann factor (of the order 10" *), could give suchlarge voltage signals of the order a millivolt fromnuclear induction, well above amplifier noise. Herecalled, having confirmed ample nuclear signalsize, that the worry of long thermal equilibriumrelaxation times might cause failure to see anysignal whatsoever. As I proceed further on, thisconcern about relaxation will be brought out.

The first attempt to measure nuclear induc-tion took place in the fall of 1945, after Blochreturned to Stanford from wartime service at theRadio Research Laboratory in Cambridge, Mas-sachusetts, where he worked on radar scatteringand theory of the magnetron. With his studentMartin Packard and colleague W. W. Hansen, anapparatus was assembled, with Bloch working onthe magnet and Packard and Hansen assemblingthe radio transmitter and receiver. In thosedays the physics apparatus at Stanford was in asorrowfully primitive state with none of the ele-gant equipment Bloch and Hansen were accus-tomed to during their research sojourns else-where during the war. The Stanford physicsbasement was cluttered with antique x-ray

Figure 4. M. Packard, R. Sands and F. Bloch.

apparatus -- and that was about it For thenuclear induction experiment they used a ratherprimitive lecture demonstration magnet (Figure5) with current supplied by the small Stanfordcyclotron.

They first failed repeatedly in their searchfor resonance signals at the field supposedlyadjusted to the correct value of 1826 gauss forproton resonance at 7.76 MC. In fact, returningto the worry about thermal equilibrium, thesample of water was allowed to sit in the mag-netic field all day to make certain that enoughmacroscopic magnetization would build up toprovide a signal before making a search. Littledid they know that they had to wait only about 3seconds for protons to relax in water. Finallysomething happened after they switched off themagnet one day (sometime after Christmas,1945) to see what was wrong. A signal blipappeared on the scope unexpectedly. They sawtheir first adiabatic fast passage signal by drop-ping the magnetic field through the resonancecondition w = YH. It was quickly confirmedthat the signal could be improved by using aparamagnetic iron nitrate solution to shorten therelaxation time. Now nuclear induction wasestablished as a reality. After this success Blochgave a colloquium lecture while Martin Packardvaliantly tried to reproduce the experiment onthe lecture table in front of the audience, but he

Vol. 7, No. 2/3 85

Figure 5. Lecture demonstration magnet used toobtain the first nuclear induction signal.

couldn't find the signal — the spins were still shyabout exposing themselves to the public, havingbeen left alone in their incoherent privacy for aneternity.

These preliminary results, reported in 1946,were followed by Bloch's famous paper onNuclear Induction in which the phenomenologicaltheory of macroscopic spin dynamics was given.It is of historical interest again to reproduce acertain page (Figure 6) of Bloch's original note-book. This shows clearly his reasoning in arriv-ing at what are now the standard dynamicalequations of nuclear magnetic resonance, and thefamous transverse T2 and longitudinal T, relax-ation times.

A few weeks after the first successful exper-iments at Stanford, Bloch received the news thata closely related discovery had been made simul-taneously at Harvard. Their discoveries wereannounced in the same issue of Physical Review,January 1946. For a brief period of time it wasthought that possibly these two investigationswere dealing with different things. Finally itwas realized the two experiments were observingthe same phenomena from different points ofview. Bloch's approach was in terms of dynamicmacroscopic voltage signals induced by preces-sion and the Faraday effect, whereas the Purcellgroup description was in optical terms of

quantum mechanical susceptibility and absorp-tion.

Many new findings developed in the years tofollow. The concept of motional narrowing inliquids developed explicitly from the Harvardgroup. Their style focused on microscopic effectsof local fields, line widths and shapes, and effectsof fluctuations of local fields upon the magneticresonance signal and its relaxation. Whennuclear induction signals in liquids were firstinvestigated at Stanford, their observed linewidths, which were predominantly caused bymagnetic field inhomogeneity, were interpretedin the beginning to be due to dipolar broadeningby neighboring spins in the sample. The germ ofmotional narrowing was of course finally evidentto the Stanford group, since the effect of longrelaxation times was included in the Bloch equa-tions. The Stanford group interest, however,was more in the measurement of spins andmoments. The Bloch method of crossed-coils wasadapted quite efficiently to the search for newresonances which W. W. Hansen had in mind asa master project, interrupted, however, by hisuntimely death. With Nicodemus and Staub,Bloch returned to his old interest in the neutron,and made a precision measurement of its mag-netic moment by observing in the same field twoprecession frequencies, that of a beam of neu-trons in vacuum, and that of protons in water.Later tritium (3H) was measured by Levinthal,and Packard worked on the precise moment of2H. Warren Proctor developed a search NMRspectrometer, and he began to measure the spinsand moments of a large number of nuclei. In thecourse of these searches it was inevitable forProctor and Yu that they should discover thechemical shift - the same nucleus in differentcompounds has a different resonance frequencybecause contributions of the bonded electrons tothe effective field at the nucleus. This shift wasalso discovered independently by Dickenson atMIT in the laboratory of Francis Bitter. Proctorand Yu observed the NMR shift of nitrogen-14 inNH4NO3 dissolved in water: one nitrogen reso-nance came from the NH4 group, shifted from asecond nitrogen resonance in the N0 3 group.This effect excited Bloch very much at first,thinking that perhaps there existed stablenuclear isomers that had slightly different prop-erties. The other possibility of course was that itcould be just a nasty diamagnetic chemical shiftwhich would not and did not excite him. After itwas confirmed as chemical in origin the shift didnot command Bloch's interest as fundamental inhis list of priority problems in physics. To con-firm whether or not the two nitrogen resonances

86 Bulletin of Magnetic Resonance

-

•mm • s »

Figure 6. Formulation of Bloch equations from original notes of F. Bloch.

were not isomers, Bloch telephoned Segre atBerkeley and asked to borrow a little 1 5 N. Ifthe two resonances were not present in ammo-nium nitrate made up with x s N, then the expla-nation would be nuclear instead of chemical.Finally the sample arrived and proved again togive two resonances. Thus the matter of chemi-cal shift was a new complicating correction factorwhich had to be taken into account in measuringmagnetic moments.

The above episode was an example of howBloch was really not enthusiastic about

analyzing the numerous side effects brought inby the many body effects of local fields in con-densed matter that could not be comprehendedwithout an involved series of empirical measure-ments. I remember distinctly the time whenDick Norberg and I were in Felix's office, pour-ing out our resonance research results to him,Norberg on metals, and I with chemical echomodulation effects. Bloch said: "Norberg, youshould be a metallurgist, and Hahn, you shouldbe a chemist!"

Vol. 7, No. 2/3 87

Figure 7. M. Packard and R. Varian.

In 1950 Bloch and Jeffries measuredprecisely the proton magnetic moment in units ofthe nuclear magnetic by observing in the samefield the NMR and cyclotron resonance of protonsusing a small anti or inverted cyclotron techni-que. The body of data obtained by all thesemeasurements was ultimately applied to thenuclear shell model of Mayer and Jensen.

Arnold, Dharmati and Packard observed in avery uniform magnetic field, resolution of hyper-fine proton resonances in C2H5OH, an effectobserved by Gutowsky and Slichter in the steadystate, and independently by myself in terms ofmodulation beats of spin echoes. The existenceof these fine structure effects together with thechemical shift serve as the basis of high resolu-tion NMR as it exists today. The enterprise ofhigh resolution NMR instrumentation was pio-neered by the Varian Company, and now hasspread as an analytic technique throughout theworld. Very good scientists, a number of whomcame from Bloch's NMR group, worked for

Figure 8. F. Bloch in a debating mood, with E.Ginzton.

Varian in those early days which gave the com-pany the research thrust that enabled it to pro-duce the first efficient instrumentation for ana-lytic NMR chemistry.

At this point I would like to interject someparticular remarks about the Bloch equations.The Bloch equations have had wide application toa number of physical effects which do not neces-sarily involve gyromagnetic spins. They applydirectly to any two quantum level or equallyspaced quantum level system, and are particu-larly useful in predicting effects involving electricdipole laser resonance phenomena.

Numerous effects in quantum optics areanalogs of spin resonance phenomena discoveredin former days and are often predicted by Blochequations. Although considerations of propaga-tion and fluorescence are not present in NMRcavity resonance experiments, there remain ahost of similar effects in the time and frequencydomain that also appear in quantum optics.

Although the original phenomenological Blochequations work very well for fluids, in manycases they are not rigorous for all systems, par-ticularly in solids. Nevertheless, the Bloch for-mulation has stimulated new statistical investi-gations with the density matrix that are morerigorous for the particular system under investi-gation. Exceedingly useful is the property that

88 Bulletin of Magnetic Resonance

the Bloch equations enable predictions of nonli-near quantum macroscopic phenomena that noamount of fastidious quantum mechanical per-turbation theory could predict as handily.

Personally it was my good fortune and privi-lege to study under Felix Bloch as a postdoctoralstudent during the two years 1950 to 1952. Itwas in his nature to have a profound influenceon his .students. His love for physics took a .highpriority in his life, which induced him continuallyto avoid the impediments of formal rules ofbureaucratic restraints that prevented him fromdoing things for himself. He always invited oth-ers to share in his search for answers, and didnot distance himself from anyone who would joinhim in the search, regardless of his or her status.What many of his students gained from himintellectually was often merged with his adviceand counsel. Felix was a devoted family man,and not incidentally, he was also an accom-plished pianist. With his good wife Lore and thefamily, the invitations to participate in activitiesof family life, with musicals, hikes and parties,were all occasions indeed memorable, giving

positive incentive and enjoyment in learningphysics from Felix.

The legacy of contributions to science byFelix Bloch was already a monument to himwhile he was alive. Among his versatile contri-butions to physics I have emphasized his workon magnetic resonance. The application andimpact in particular of the science of NMR nowinvolves the activities of thousands of researchpeople in solid state physics, analytical chemis-try, and most recently of clinicians in the medicalworld who apply the technique for imaging thehuman body for medical diagnosis. Not longbefore his death Felix indicated to me as well asto others that the extension of magnetic reso-nance to humanitarian practical use was of greatsatisfaction to him,

Felix Bloch is among the Greats in the his-tory of science, affecting many who never methim, a living influence for them and a personalone for those of us who were fortunate to haveknown him, and who shall remember him withaffection and gratitude.

Vol. 7, No. 2/3 89