I. I. RABI Columbia University, ¹m York, N. Y. (Received March 1, 1937)
The nonadiabatic transitions which a system with angular momentum J
makes in a magnetic field which is rotating about an axis inclined with respect to the field are calculated. It is shown that the effects depend on the sign of the magnetic moment of the system. We therefore have an absolute method for measuring the sign and magnitude of the moment of any system. Applications to the magnetic moment of the neutron, the rotational moment of molecules, and the nuclear moment of atoms with no extra-nuclear angular momentum are discussed.
N a previous paper' the effect of a rapidly depolarization caused by the nonadiabatic transi- - - varying magnetic field on an oriented atom tions to be described below. Another example is possessing nuclear spin and extra-nuclear angular the sign of the moment arising from molecular momentum. It appeared that it was possible to rotation which results in a positive contribution deduce the sign of the magnetic moment of the from the motion of the nuclei about the centroid nucleus from the nature of the nonadiabatic and a negative contribution from the electrons. transitions which occur if the field rotates an The following considerations should make it appreciable amount in the time of a Larmor possible to make the same sort of observations rotation. This effect was applied experimentally' with simple systems as are made in the Einstein- with the method of atomic beams to measure the de Haas and Barnett experiments: namely, the sign of the proton, deuteron, K", etc. The magnitude and sign of the gyromagnetic ratio. evaluation of the sign was possible because the Consider a simple system such as a neutron experiment decided whether the h. f.s. level was with magnetic moment p = — gpo J, where g is the normal or inverted. Since the sign of the electronic Lande g factor, Jis the total angular moment moment is known to be negative a normal level due to all causes. If g is positive the total moment meant positive nuclear moment and an inverted is negative as in the spinning electron. If g is level negative moment. negative the moment is positive. In a magnetic Clearly it is desirable to find another effect field II the system precesses with the Larmor which will make it possible to find the sign of the frequency v=gpolI/k. If g is positive the pre- nuclear moment in cases where the normal state cession is in the positive direction and if negative of the atom is one in which there is no electronic in the negative direction. We shall now consider angular momentum as in the alkaline earths. our system initially quantized with magnetic Spectroscopic methods where applicable will quantum number m in a field II which is constant yield this information, but there are numerous in magnitude but rotates with a frequency cu/2~ important instances in which molecular and about some direction which is at an angle 8 with atomic beam methods are the only ones available. respect to the direction of the field. For example, it would be very desirable to This problem was solved by Giittinger' for the measure the sign of the moment of the neutron particular case when the angle is ~/2. He found directly. Although it would be very difficult to that transitions will occur to other magnetic apply atomic beam methods to the neutron, the levels with quantum number m' when a&/Z~ is of polarization effect of magnetized iron suggested the order of magnitude of v. The transition by Bloch may possibly be useful in this con- probabilities in this case do not depend on the nection as a device for measuring the degree of direction of the field. It will be shown that in the more general case the direction of rotation ' Rabi, Phys. Rev. 49, 324 (1936). introduces an asymmetry into the problem and ' Kellogg, Rabi and Zacharias, Phys. Rev. 50, 472 (1936); Torrey and Rabi, Phys, Rev. 51, 379A (1937) P. Guttinger, Zeits. f. Physik '73, 169 (1931); E. Major- Millman and Zacharias, Phys. Rev. 51, 380A (1937). ana, Nuovo Cimento 9, 43 (1932). 52 SPACE QUANTI ZAT ION 653
as a, consequence with the same and p« the ~
« ~ ~ ~ The quantities A and 8 are determined from the transition probabilities will be different depend- initial conditions the normalization and con- ing on whether g is positive or negative. The dition ~C «~'+ [C «~'=1. If P and Pe are the Majorana and Guttinger arrangements do not vectors which correspond to m=+-,', — ', re- —,
possess this property. spectively, in the direction of H we obtain4
It will suffice to consider only the case where J=-,' since the solution of more general problems 4-=(2 ) '(pe -'V -«+ 0 «), - depends only on the solution of this simple case. A= (2q) '*( — Pe—'-V-«+qy «), The Schrodinger equation for this case is p=sin 0, a=1 —cos 0, y=1+cos 0, ig=XP, o.'+ p' = 2n, y'+ p' = 2y. ~= g(««p/2)(~«H, +~«Hp+~pHp). If P(0) =f (0) at t=0 we start with the system If we set ', and quantized in the direction of H with m= —. (2) A+B =P/(2a) &,
we obtain from (1)
(+P«l !e+Pp)« dC«/dt = ( ig««p/2h) —[IipC«+(H« iHp) C «—], E b ) E b ) (3) dC «/dt = ( —ig««p/2k)L —HpC «+(FF«+DEFI«) C«j. Utilizing these values of A and and setting 8 n = (p«'+4@'+4b' — 4pph) l we obtain for the proba-
t- With the substitutions bility amplitudes II~ —II sin 8 cos q, II2 —B sin 8 sin q, 1 n FI3 —H cos 8, C« — . e &'"'"& P cos (4) (2cx)' 2
g— —cos P=c, g— upH II sin 8=b, Ztp y=cot, (2nb+2Pa Pp«) n— 2 k 2k n sin —' 2 t, which represents a field H constant in magnitude and precessing about the s direction with angular C ~ — e'"'" n cos —t velocity co, we have (2n) 2
d C«/dt = iaC« "'C
ibe '— «, —- Z (2Pb —2nu+up«) sin n —t . n 2 dC «/dt=iaC « ibe' 'C«— Therefore The probability of finding the system in a state with m= — ' —, with respect to IX is therefore
d'C«/dt'+i pp C«+ (a'+ b' —(pa) C—« —0. (6) —~A"4 ~'= —pe'"'C«+pC dt F'(«, «) ~ «~' (11) (4~v)' The solution of this familiar equation is with the values given in Eq. (12) C« — Ae'»'+B—e'~ p' n ++P1 ++P2 g&P2t sin' —t. (12) g&r«& t A giP1t+ n2 (7) P« ——(p«/2)+-', (p«'+4a'+4b' —4ppu)', In terms of the Larmor frequency v =g««pH/k, the
— pp — (p«/2) '(p«'y4a'y4b-' —4ppa)'*. — 4 Dirac, Quantum Mechanics, p. 70. 654 I. I. RAB I
angle 8 and the frequency of rotation r= a&/2n. difference and a knowledge of the sense of rota- tion one can infer the sign of the moment. The sin' 8r' magnitude may be inferred from the absolute sin2 ~$ v'+v' —2vr cos 6 value of the fields and their direction and the angular velocity of its rotation. X (v'+ v' —2vr cos 8) l. (13) It is unnecessary to consider the details of the realization of these rotating fields and the de- This result reduces to Guttinger's formula tection of these transitions since similar con- when 6 = 7r/2. For other values of 6 it is apparent ditions have already been obtained in the experi- that the transition probability for given 8, II ments cited above. and co will be quite different depending on whether g is positive or negative since v appears To generalize these results we apply the general linearly in the result. Expressed in another way result of 1Vlajorana' ' for any value of J we may say that I'(-, , =, ) depends on the direc- P(, , ) —(cos 'n)'~(J+m)!(1+m')! —,
tion of rotation of the field for a given sign and
magnitude of g. X (J—m)!( —m')! j (17) Since rotating fields are usually realized by — 9J ( 1) v(tan 1~) 2v m+m' allowing the system to pass through a field which x Z— v!(v — =0 m+m')!(J —m —v)!(J'—m' —v)! changes in direction from point to point, the total change 1.n direction is fixed. If we set where the value of the parameter n is given by y = 27rrt we obtain from (15) setting v/r = q Sln &0! = I (-'„ &(:, -&) =- sin' 8 s1n J. 1+q' —2q cos 8 and depends only on g and not on m or It may be of interest to note that in cases X(1+g' —2g cos 6)'. (14) where we have a coupled system such as a nuclear spin coupled to molecular rotation, in If we set q=cos8 which can be arranged by which the coupling is weak and g small, the field suitably varying the magnitude and 'direction of required for these transitions can be such that the the field, component systems are completely decoupled. As P&;, q) =sin' (12' sin 8), (15) was shown by Motz and Rose, ' each system would then make these transitions independently. which can be made as close to unity as one pleases In particular, if the moments of the two systems by arranging experimental conditions so that are opposite in sign it should be possible to 2~) q )7r. If one were then to leave everything arrange conditions so that only one of these two unchanged but reversed the direction of rotation systems makes these transitions. In this case it we would get should be possible by means of the focusing methods developed in this laboratory to measure P&~ ~& = sin' 8 sin' P — (1+3 cos' 8) ', (16) directly one moment in the presence of another f+3 cos' 0 2 (rotational and nuclear) with only slight interference. a very much smaller quantity. It is clear therefore that from this qualitative ' Motz and Rose, Phys. Rev. SO, 348 (1936).
