Molecular Beams

MOLECULAR BEAMS

Studies involving coherent beams of atoms and n10lecules have had

a profound influence on the development of modern physics. An10ng their technological dividends are masers, lasers and atomic clocks

by O. R Frisch

he atomic physicist, like every be reached by atoms that had passed in written in 1738 by the Swiss mathema Tbody else, needs air for breathing, straight lines through the constrictions. tician Daniel Bernoulli. He pointed out but otherwise it is in his way. At In essence Dunoyer's apparatus was that bombardment by fast molecules normal atmospheric pressure each air the prototype of all modern molecular would exert a pressure on the walls of molecule is involved in nearly a billion beam devices. The three necessary ele a container, and that the pressure would collisions a second. Only in gases at a ments of any such device are (1) a rise in proportion to the number of much lower pressure-in what is loosely source of molecules (usually called an molecules per unit volume, or the den called a vacuum-do molecules attain a oven, even though it is not always sity of the gas (as Robert Boyle had ob reasonable degree of privacy. We mea heated ) with an aperture through served some 80 years earlier). Bernoulli sure such low pressures in units called which the molecules can escape into a also explained the rise of pressure with "torr," after the Italian investigator vacuum, (2) a second, collimating ap increasing temperature by assuming that Evangelista Torricelli, who devised a erture, which forms the beam by stop the molecules move faster when the gas method for obtaining fairly good vac ping or deflecting all the molecules that is heated. Of course Bernoulli was right uums in 16 43. (Atmospheric pressure is do not pass directly through it, and (3) in every respect, but his suggestions about 760 torr.) More than 200 years a method for detecting the effects of raised at least as many questions as they later, in 1855, the German glassblower various influences on the beam. In con answered. Heinrich Geissler used Torricelli's meth trast to beams of subatomic particles, Gas molecules move as fast as rifle od to evacuate glass tubes down to which can be accelerated to speeds ap bullets; why, then, does it take minutes about .01 torr; this achievement led di proaching the speed of light, molecular for a smell to spread from one end of rectly to the discovery of cathode rays, beams normally travel at "thermal" a room to another? \lVe now know the which were eventually identified as speeds: a mile a second or so. The answer is that the molecules keep col beams of free electrons by J. J. Thomson generic term "molecular beam" applies liding and therefore make little head in 1897. At such a pressure electrons both to groups of atom s (for example way. In fact, from the rate at which a can travel many inches, but at the same neutral hydrogen, each molecule of smell does spread-to put it more gen pressure the mean free path of an air which consists of two hydrogen atoms ) erally, the rate at which one gas dif molecule between collisions is only and to single atoms (for example so fuses into another-it is possible to cal about a millimeter. To achieve a coher dium vapor). culate how often the molecules in a ent beam of any kind of atom or mole Molecular beams can be made to pass gas collide every second. Such calcula cule one needs a much better vacuum through magnetic or electric fields; they tions were not made until the middle of at least .0000l (10-") torr or better. can be directed at various obstacles, the 19th century and only then did Techniques for obtaining very low such as crystals or gas molecules; they physicists begin to regard atoms and pressures were developed rapidly dur can be illuminated with intense light, molecules as real and not merely as fic ing the first decade of this century, and bombarded with electrons or crossed tions convenient for chemists. Several in 1911 the French physicist L. Du by other molecular beams. From such methods were devised for computing noyer succeeded in producing the experiments much has been learned the speed of molecules and their mean world's first molecular beam; he demon about the basic structure of matter that free path between collisions. In general strated that in a good vacuum the so could not have been learned in other the results agreed and supported the dium atoms that comprised his beam ways. The invention of masers and la new kinetic theory of gases, which was moved in straight lines. Dunoyer's ele sers arose out of a molecular-beam ex firmly established during the latter half gantly simple apparatus consisted of an periment, and a beam of cesium or hy of the 19th century by James Clerk evacuated glass tube with two con drogen atoms in an "atomic clock" is the Maxwell of Britain, Ludwig Boltzmann strictions [see top illustration on page most accurate timekeeper ever built. of Austria and other theorists. Nonethe 60]. When he heated some sodium in less, the evidence for the kinetic theory one of the end compartments, a dark de The Stern-Gerlach Experiment remained indirect. posit of sodium appeared in the com The first direct measurement of the partment at the opposite end. The de The idea that gas molecules are in speed of gas molecules was made in posit covered only the area that could rapid motion goes back to a treatise 1920 by Otto Stern, who was then

© 1965 SCIENTIFIC AMERICAN, INC working at the University of Frankfurt; it was also the first measurement in which a molecular beam was employed. Instead of an oven Stern used a hot platinum wire plated with silver and positioned in the middle of an evacuated bell jar. Silver atoms evaporated from the wire in all directions, but only those that passed through a narrow slit were able to reach a glass detector plate, on which they left a deposit in the form of a narrow line. As long as the apparatus remained stationary the speed of the atoms did not matter; fast or slow, they all followed the same straight path. But Stern had mounted the main parts of his apparatus-the silver-plated wire, the slit and the glass plate-on a vertical axis that could be made to rotate at some 1,500 turns per minute inside the bell jar [see bottom illustra tion on next page J. At that time, with vacuum techniques still in their infancy, this was quite a tour de force. When the apparatus was rotating, the deposit was formed in a slightly different place, because now the glass plate was moving TYPICAL MOLECULAR·BEAM APPARATUS , photographed in the laboratory of Nor as the atoms were heading for it. From man F. Ramsey at Harvard University, is used for studies of nuclear magnetic resonance. the amount of shift Stern was able to A beam of hydrogen molecules enters the apparatus at the far end, passes through several estimate the average speed of the atoms magnetic fields and is detected at the near end (see schematic diagram at bottom of page 62). at 580 meters per second; this was with in a few percent of the figure predicted by the kinetic theory of gases. More over, he observed that the deposit formed during rotation was a bit fuzzy; this showed that not an the silver atoms had the same speed, which Maxwell had foreseen. About a year later Stern published, together with his colleague Walther Gerlach, the result of another molecu lar-beam experiment. This experiment had a profound influence on the devel opment of quantum mechanics; it sup ported the rather incredible conclusion, drawn from spectroscopic observations, that the orbit of an electron moving about the nucleus of an atom in a mag netic field must always be at right angles to the lines of force in the field, al though the electron can orbit in one direction or the other. Such "spatial quantization" was completely at odds with classical physics, according, to which a magnetic field would merely make the electron's orbit precess, or wobble, around a fieldline rather like a spinning top, without imposing any limit on its orientation. In thinking about the Stern-Gerlach experiment it is helpful to treat the atom as though it were a small bar magnet with a magnetic moment /1-. SOURCE·END VIEW of the molecular·beam resonance apparatus shown above reveals the {The magnetic moment of an orbiting end of the first, or A-field, deflecting magnet. The beam of hydrogen molecules passes charge e, traveling with the speed v on between the two semicircular·shaped pole pieces at the center. The source equipment and a circle of radius r, is given by the equa- the bulkheads separating the various chambers have been removed for easy viewing.

© 1965 SCIENTIFIC AMERICAN, INC SOURCE COLLIMATOR OBSERVATION CHAMBER ergy could have only the two extreme CHAMBER CHAMBER values -/�H and +I-'-Hand nothing in between. For nearly a decade before the Stern J Gerlach experiment the classical con cept of randomly oriented electron or bits had been in conflict with the Zeeman effect, the fact (first noted by the Dutch physicist Pieter Zeeman in 1896) that the lines in a spectrum are often split into two or more lines when TO PUMP the emitting atom is placed in a mag netic field. At first the reason for the FIRST MOLECULAR-BEAM APPARATUS was constructed by L. Dunoyer in 1911. When remarkable sharpness of these spectral he heated some sodium in one of the end compartments of an evacuated glass tube (left), a dark deposit of sodium appeared in the compartment at the opposite end (right)_ The lines was unknown. The first step toward deposit covered only the area that could be reached by sodium atoms that had passed an understanding of this phenomenon in straight lines through the two constrictions. The tube was about eight inches long. came in 1900 with the introduction of Max Planck's quantum postulate, which made the rather startling assumption tion I-'- =evr/2c, c being the speed of sider a large number of such magnets, that light is emitted in quanta, or dis light.) When such a magnet is inserted oriented at random, then the cosine of crete parcels of energy. In 1913 Niels into a magnetic field H at an angle IX IXis equally likely to have any value Bohr made the further assumption that with respect to the field lines, its energy between -1 and + 1; accordingly the the energy content of atoms must also content is changed by an amount equal magnetic energy could have any value be quantized, and that the emission of to I�H times the cosine of IX[see bottom between -I-'-Hand +I-'-H. If spatial spectral lines is caused by the atom's illustration on opposite page J. If we con- quantization existed, however, the en- transition from one energy state to an other. According to Bohr's model the energy difference between the two states is carried away by the emitted BELL JAR light quantum and thus determines the wavelength of the emitted light. The sharpness of the energy states gum'an tees the sharpness of the spectral lines. If atoms were oriented at random in a magnetic field,their energy content would, as we have seen, be "blurred" by an amount equal to twice I-'-H.The Zeeman effect, however, demonstrated that spectral lines are not in fact blurred COLLIMATING SLIT ...", but instead are split into two or more sharp lines. An analysis of the Zeeman splitting of the spectral lines of silver showed that in their normal state the -==.:::;--'------..... - i :--i .-...../.".,' ---3>' �II silver atoms should be capable of only ----'\ two orientations, corresponding in terms SOURCE)D of magnetic energy to the values -I-'-H and +/�H. Now, Stern argued as follows: If a GLASS DETECTOR PLA TE magnetic atom were placed in an in homogeneous magnetic field (a field whose intensity varies from point to point), then the atom's energy content would also vary. In order to move the atom from point to point work would have to be done. In other words, the magnetic field would exert a force on the atom as a whole; the force would of course depend on the orientation of the atom with respect to the magnetic FIRST DIRECT MEASUREMENT of the speed of gas molecules was made in 1920 by Otto field. If the atoms were oriented ran Stern; this was also the first measurement in which a molecular beam was employed. The domly, the force would vary continu main parts of the apparatus (shown from above in this schematic view) were mounted on a vertical axis that could be made to rotate at some 1,500 turns per minute inside an evacu ously from -p.G to +I-'-G,G being the ated bell jar. When the apparatus was rotating, a deposit was formed in a slightly different field gradient, or the change in field in place on the glass detector plate than when the apparatus was stationary. From the amount tensity per centimeter. If, on the other of sbift Stern estimated the average speed of the atoms. He also observed that the deposit hand, spatial quantization obtained, the formed during rotation was fuzzy; this showed that not all the atoms had the same speed. force would always be either -I-'-G or

© 1965 SCIENTIFIC AMERICAN, INC +,uG. By making the force large enough to cause observable deflections in a beam of silver atoms, Stern reasoned that it might be possible to decide ex perimentally how atoms orient them selves in a magnetic field. Quite by chance Gerlach had been measuring the gradient of a magnetic field near the sharp edge of a magnet. Stern computed that such a magnet should deflect silver atoms enough for his purpose, so they joined forces. Their experiment was a complete suc cess: the fine line of silver that had been deposited in the absence of the mag cb netic field was distorted into a fuzzy, asymmetric loop when the magnet was turned on [see illustration on page 63]. The fuzziness was caused by the Max wellian distribution of the speeds of the silver atoms; the asymmetry, by the fact that the field gradient was greater � ,� near the knife-edged magnetic pole than away from it. The crucial findingwas that the deposit was a loop and not a J, � � blur: near the middle there was practi J, J, ! cally no silver. If the silver atoms had T •• • . J, • been oriented at random in the magnetic 1" � ! • field-as classical physics demanded • • J, T there should have been no fewer atoms ! f with small deflections than atoms with J, . large ones. The experimental result was ·1 . • in perfect agreement with the concept • J, 1 of spatial quantization. • • Stern could foresee many other po tential uses for molecular beams, and DEVELOPMENT OF ELECTRON THEORY is reflected in these three drawings, which in 1923, when he became professor and represent three different concepts of the behavior of the electron in a magnetic field. Accord· head of the department of physical ing to classical, or pre.quantum·mechanical, physics (top), the magnetic field would merely chemistry at the University of Ham make the electron's orbit precess, or wobble, around a field line, without imposing any burg, he set out to develop the method limitation on its orientation. According to the Bohr model of the hydrogen atom (middle), systematically. During the next 10 the electron's orbit in a magnetic field must always be at right angles to the field lines, with years he and his students published a the electron orbiting in one direction or the other. This concept, called "spatial quantiza series of 30 papers reporting many im tion," was supported by the observed Zeeman splittiug of spectral lines and also by the pOltant findings-a massive demonstra Stern·Gerlach molecular·beam experiment. Since 1925 the spin of the electron has been tion of the power of intelligent planning. known to contribute to the total angular momentum of the atom that undergoes spatial I had the good fortune to work with quantization. In the case of hydrogen or silver atoms in their ground state (bottom) there is no orbital momentum; hence electron spin is aligned parallel or antiparallel to magnetic field. Stern from 1930 to 1933, and it was wonderful training. Our whole cam paign had been outlined in the first of those 30 papers; in due course most of the original problems were attacked and solved, as well as some problems that had not been foreseen but that grew out of our work. Nothing was left to chance or conjecture. Every piece of apparatus was designed with care and its behavior was checked out in advance; in this way any shortcoming wa s quickly revealed and, if possible, remedied before the piece was put into use. With this sort MAGNET ANALOGY is helpful in thinking about the Stern·Gerlach experiment; accord ing to the analogy, the atom is treated as though it were a small bar magnet with a magnetic of training Stern's students were well moment ,u. When such a magnet is inserted into a magnetic field H at an angle 0: with re prepared to go out in the world and spect to the field lines, its energy content is changed by an amount equal to ,uH times the start their own successful schools of cosine of 0:.If we consider a large number of such magnets, oriented at random, then the work, and some of them did. cosine of 0:is equally likely to have any value between -1 and +1; accordingly the energy One of the experiments Stern had could have any value between -,uH and + ,uH. If spatial quantization existed, however, proposed in that first paper was a test the energy could have only the two extreme values - ,uH and +,uH and nothing in between.

c d

INHOMOGENEOUS MAGNETIC FIELDS, used for deflecting the angle of incidence and the strength of the field between and beams of magnetic atoms and molecules, can be set up in several outside the pole pieces. In d an inhomogeneous field similar to that a ways. The pole·piece arrangement in is similar to that used in the in b is set up by two cylindrical conductors through which a strong original Stern·Geriach experiment. The arrangement in b is a later electric current is passed in opposite directions; the absence of version for which the field can be more readily calculated from the iron in this arrangement facilitates the calculation of the field. In dimensions of the pole pieces. In c the beam is directed through all four examples a pencil.like beam of atoms is shown being split the edge of a homogeneous field; the deflection depends only on into two component beams having opposite magnetic moments.

SOURCE SEPARATION MAIN CHAMBER CHAMBER CHAMBER !-�------i COLLIMATING SLIT OIL ( MANOMETER SOURCE L ------�===:tl ==�� FLOW METER

TANKS

HYDROGEN DEUTERIUM RADIO·FREQUENCY LOOPS

MOLECULAR·BEAM RESONANCE APPARATUS, used for mea The central region also contains two loops, through which an oscil suring the nuclear magnetic moments of hydrogen molecules, is lating current from a small radio·frequency generator is passed. shown in this schematic diagram. The apparatus contains two in The two inhomogeneous fields are so arranged that the atoms homogeneous magnetic fields: a polarizer (an A field) to select selected and somewhat spread out by the A field are brought molecules of a particular orientation and an analyzer (a B field) to together again by the B field; thus the detector is able to record a test for the reoriented molecules. The transitions from one orien good fraction of the original beam. Pressures inside the chambers tation to the other take place in an intermediate third field (a C are kept sufficiently low for most molecules to travel the entire field), which must be both homogeneous and accurately known. length of the apparatus without being subjected to collisions.

© 1965 SCIENTIFIC AMERICAN, INC of the suggestion made in 1924 by Louis de Broglie to the effect that a beam of particles should somehow have associated with it a train of waves with a wavelength inversely proportional to the momentum of each particle. This was the germinal idea of modern wave mechanics. Two years later Erwin Schrodinger showed that the energy states of a hydrogen atom could be de rived by working out what the De Bro glie waves of an electron would do in RESULT OF STERN·GERLACH EXPERIMENT, which measured the deflectionof a beam of silver atoms in an inhomogeneous magnetic field, is shown in this drawing, made from the electric fieldof a hydrogen nucleus, their original 1921 paper. The fine line of silver that had been deposited in the absence of and further developments followed in the field was distorted into a fuzzy, asymmetric loop when the magnet was turned on quick time. (left) (right). The fuzziness was caused by the Maxwellian distribution of the speeds of silver atoms; the asymmetry, by the fact that the field gradient was greater near the knife.edged Diffraction Experiments magnetic pole than away from it. The crucial finding was that the deposit was a loop and not a blur: near the middle there was practically no silver. If the atoms had been oriented Oddly enough there was no direct at random in tbe field, there should have been no fewer atoms with small deflections than experimental evidence of De Broglie's atoms with large ones. Result was in perfect agreement with concept of spatial quantization. wave hypothesis until 1927, when C. J. Davisson and Lester H. Germer of the beam, but now the atoms that struck have a longer wavelength than the fast Bell Telephone Laboratories found that the partitions did not stick and so had er ones. beams of free electrons produced char to be removed continually with fast In a later experiment Stern and I acteristic diffraction patterns on being pumps. The beam struck a freshly managed to sharpen the test by select reflected from a metallic crystal. Almost cleaved slice of lithium fluoride, from ing atoms with a well-definedvelocity. simultaneously George P. Thomson of which some of the atoms were reflected, This was accomplished by successively Britain observed the diffraction of elec as light is from a mirror, and some dif passing the beam through two rapidly trons that had passed through a thin fracted; the rows of fluorine atoms act rotating toothed wheels; atoms that had metallic foil. Even after these discover ed as a diffraction grating. Atoms that passed through one of the slots in the ies a skeptic might have questioned had exchanged energy with the crystal first wheel could pass through the sec whether particles other than electrons would leave it in random directions. ond wheel only if they had the right could be diffracted. He might well have To detect the atoms, Stern had de velocity [see bottom illustration on page been doubtful, for example, that De veloped, with Friedrich Knauer, an ex 65]. When this monochromatic beam Broglie's idea would apply to beams of tremely sensitive "hot wire" manometer, struck the crystal surface, the atoms atoms or molecules. based on a principle first put forward in were diffracted within a much narrower Stern felt sure that heavy atoms such 1906 by the German physicist M. von range of angles. Since we knew both the as those of silver would not do for a Pirani. The manometer could detect wavelength and the momentum of the diffraction test; their wavelength and very small changes in pressures as low diffracted atoms (momentum equals hence their angle of diffraction would as 10.8 torr. It was connected to a mass times velocity) it was possible to be inconveniently small. Moreover, small tube that was pointed at the crys compute Planck's constant from the ex heavy atoms would usually condense on tal in the direction from which the periment (Planck's constant equals mo the crystal instead of being reflected; atoms were to be detected. There was mentum times wavelength). The result even if they did not actually stick, they some difficulty at first in making this was within 1 percent of the anticipated would gain or lose energy on making detecting tube movable, so the appara value. A discrepancy of 3 percent, contact with the crystal, thus spoiling tus was later arranged in such a way which worried us at first, was cleared the diffraction pattern. Therefore he that the crystal could be tilted in order up when, having looked in vain for decided to use helium atoms and hydro to direct the reflected and diffracted other sources of error, we checked the gen molecules for the beams; both of atoms into the tube [see illustration on number of teeth on the wheels and these gases condense only at very low next page]. found that the lathe on which they had temperatures. The crystal surface had The graph on page 65 shows a typi been cut was incorrectly labeled: it had to be hard and had to contain only light cal result obtained when Stern, together cut 408 rather than 400 teeth! The re atoms in order to make an exchange of with Immanuel Estermann, conducted a maining discrepancy of 1 percent was energy with the impinging particles less diffraction experiment using a beam of felt to lie well within the limits of ex likely. Lithium fluoride was chosen be helium atoms. The sharp peak at tilting perimental error. cause it could easily be grown in inch angle zero is caused by the reflected Stern's experiments with helium sized crystals and cleaved into smooth atom s (sometimes called the specular atoms had proved beyond doubt that slices. beam). As the tilting angle is increased, beams of complex particles could be Of course the whole technique was the number of detected atoms first diffracted and would exhibit the wave different from the earlier molecular drops sharply, but then it rises again properties De Broglie had foretold. Hy beam experiments. The source of atoms when the diffracted atoms begin to drogen molecules, each consisting of or molecules was now a glass bulb, from enter the detector. The diffracted atoms two hydrogen atoms, yielded similar re which the gas passed through a capil are spread over a range of angles be sults, although the diffraction maxima lary (to control the flow) into the main cause of the Maxwellian distribution of were weaker. Tests with heavier atoms, part of the apparatus. Partitions with their velocity; in accordance with De both in Stern's laboratory and else holes were again used to define the Broglie's hypothesis, the slower atoms where, showed no diffraction,as expect-

" / "'/ " " DIFFRACTION ...... -20'/ ·1 PATTERN \ 1+20° \ 1 1 \\ /

\ I1 \ ------j --- /- --- f' � �I�o� .. \ o LITHIUM FLUORIDL CRYS1AL�

+200

DIFFRACTION OF HELIUM ATOMS by a crystal surface of tive manometer. The graph at far right shows a typical result. The lithium fluoride was acbieved by Stern and bis colleagues in 1929. sharp peak at tilting angle zero is caused by tbe reflected atoms Their apparatus was arranged in such a way that the crystal could (sometimes called the specular beam). As the tilting angle is in be tilted in order to direct the reflected and diffracted atoms into creased, the number of detected atoms first drops sharply but then a small tube (upper right), which was connected to a very sensi- rises again when the diffracted atoms begin to enter detector. These

ed. Such atoms are always reflected was first published by Samuel A. Goud ed in designing a wave equation that diffusely, as light is from white paper, smit and George E. Uhlenbeck in 1925, was remarkably simple, and he was sometimes with a slight enhancement in a number of puzzling spectrographic thrilled to find that the electron, if it the specular direction, as from slightly facts fell neatly into place. In some obeyed his equation, had to spin just as shiny paper. Not much work has been atoms the magnetic moment was ap Goudsmit and Uhlenbeck had predicted done in this area, but such studies may parently due to a combination of elec it would. The magnetic moment of the be of some usa in the future. tron spin and electron orbital motion, atom was also given by Dirac's equation. Certain anomalies were found in and in 1923 Alfred Lande had already What about the electron's big brother, Stern's diffraction experiments that in devised an empirical formula for com the proton? It was 1,836 times heavier dicated a very weak attraction between puting the magnetic moment of these than the electron, but Dirac's theory the helium atoms and the surface of atoms to account for the observed Zee imposed no weight limits. Indeed, there the crystal. In addition slight irregulari man splitting of the spectral lines of the was no reason why Dirac's theory ties in the shape of the reflected beam, atoms in a magnetic field. Stern had should not apply to the proton as well; found later by Stern and Estermann, induced a couple of his students to if it did, the spin, or angular mo offered a clue to the nature of submi verify Lande's formula for the elements mentum, of the proton would be the croscopic crystal imperfections. In both bismuth and thallium, but on the whole same as that of the electron but its cases further research is now dormant, interest in the magnetic moments of magnetic moment would be 1,836 times but perhaps it is not defunct. atoms seemed to be on the wane. smaller than that of the electron. In 19 26, however, Stern pointed out Here was a fresh challenge for Stern. Magnet.Moment Experiments that the molecular-beam method ought The diffraction experiments had given to be capable of detecting and perhaps him experience with hydrogen beams, By far the most fertile line of work measuring magnetic moments 1,000 and the hydrogen molecule was par in molecular beams has grown out of times smaller than any previously ticularly suited for this new test: its two the original Stern-Gerlach experiment measured. One might expect such electrons were known to spin in op and deals with the deflection of mag magnetic moments to be created by the posite directions, but its protons usually netic atoms in inhomogeneous magnetic rotation of molecules or to be intrinsic had their spins pointing in the same fields. The interpretation of the Stern to atomic nuclei. Interest in such small direction. Hence the hydrogen molecule Gerlach experiment changed soon after moments quickened two years later, would possess a magnetic moment twice it was completed; the magnetism of the when P. A. M. Dirac published his that of a single proton. The sensitivity silver atoms was now believed to be celebrated mathematical theory of the of the Stern-Gerlach experiment would caused not by the orbital motion of its electron. Dirac had tried to improve on have to be improved about a thousand outermost electron, as in Bohr's model Schriidinger's wave equation, which was fold, but Stern was confident this could of the hydrogen atom, but by the rota nonrelativistic (that is, valid only for be done. The inhomogeneous magnetic tion of the electron around its own axis. particles moving fairly slowly compared field was made narrower (to provide a When the idea of a spinning electron with the speed of light). Dirac succeed- higher magnetic gradient) and longer

© 1965 SCIENTIFIC AMERICAN, INC (para-hydrogen) are limited to even mul extra magnetic moment of the proton. tiples of the rotation quantum. In Stern's measurement was the firs t clear SPECULAR these molecules the magnetic moments indication that the proton is a much BEAM of the protons would cancel and on more complex particle than the electron. The next great advance in the mea 2' being cooled enough the molecules « w would stop rotating altogether. surement of magnetic moments by co Normally hydrogen is a three-to-one means of molecular beams was made by 0 w mixture of ortho- and para-hydrogen, 1. 1. Rabi, who had worked with Stern I- U but by cooling hydrogen with a suitable in 1931 and had then returned to the «cr: LL catalyst-for example charcoal-one can U.S. to start his own school at Columbia LL get practically pure para-hydrogen. A University. To understand Rabi's con 0 LL beam of cold para-hydrogen, having tribution requires a bit of backtracking. 0 neither a net proton spin nor a molecu If one were to plot the energy of a >- I- lar rotation, would be practically non sodium atom, say, as a function of mag iii magnetic, as we were able to verify. It netic-field intensity, the result would z w be two straight lines, corresponding to I- could be deHected in an inhomogeneous Z magnetic field only on being slightly the parallel and antiparallel orientations warmed up, and from such measure of the atom's spinning electrons with r ments we were able to determine how respect to the magnetic field [see illus much of the molecule's magnetic mo trationa t left on next page J. The fore o L-____L- ____L- ____� __ � ment is actually caused by the rotation going statement is not, however, quite -20 -10 0 +10 +20 of the molecule as a whole. By compar correct; we have neglected the fact that ANGLE OF DIFFRACT ION (DEGREES) ing these deflections with those observed the nucleus of the sodium atom has a when we used beams of ordinary hydro spin and hence a magnetic moment of experiments proved beyond doubt that gen (which is 75 percent ortho-hydro its own. In fact, the spin of the nucleus beams of complex particles (helium atoms gen) we were finally able to compute the is greater than that of the electron; and later hydrogen molecules) could be dif fracted and would exhibit the wave proper magnetic moment of the proton. as a result the nucleus is capable of a ties foretold by Louis de Broglie in 1924_ To our great surprise the proton more than just two orientations in turned out to be two and a half times magnetic field. The angular momentum more magnetic than we had expected! of any particle is always equal to the (so that the molecules would be sub This meant that the proton did not obey product of the quantum unit of spin jected to deHection for a longer time). Dirac's equation and was in fact not times 1, where 1 is one of the numbers The deHection still amounted to only a simply the electron's big brother but in the series 0, 1/ 2, 1, 3/2, 2 and so on. few hundredths of a millimeter. Very something far more complex. Today we A system with spin 1 has 21 + 1 allowed fine beams had to be used, as well have plenty of other evidence of the orientations in a magnetic field. For as very fine detector slits. The latter proton's complexity. vVe now visualize example, the electron, with 1 equal to requirement in turn meant that I had the proton as a beehive of activity, with 1/2, has just two orientations. The to build new manometers with much mesons of various kinds forming and sodium nucleus, on the other hand, has smaller volumes, so that pressure disappearing near it all the time; the a spin equal to 3/2 and so has four equilibrium would be reached within a motion of these mesons accounts for the allowed orientations. Thus the two lines period of not more than a minute. The measurements were complicated by the fact that the magnetic moment of a hydrogen molecule is not simply the sum of the two proton moments. The molecule itself also rotates, its two protons swinging around like two dancers doing a waltz, and this rotation gives rise to an additional magnetic moment. By cooling the "oven" from which the hydrogen entered the ap paratus it was possible to reduce the molecular rotation but not to stop it completely. The rules of quantum me chanics compel those molecules of hy drogen that have both protons spin ning in the same direction (called ortho hydrogen) to rotate with an angular momentum that is an odd multiple of the quantum of rotation. Thus even at the lowest temperatures these mole REFINEMENT OF DIFFRACTION TEST was achieved in a later experiment by Stern and cules have to rotate, since one quantum the author by successively passing the helium beam through two rapidly rotating toothed of angular momentum would always wheels; atoms that had passed through one of the slots in the first wheel could pass through remain. Hydrogen molecules whose two the second wheel only if they had the right velocity. When this monochromatic beam strnck protons spin in opposite directions the crystal surface, the atoms were diffracted within a much narrower range of angles.

3, i 1, 2, 4 4

I > l') 0::: w Z w

5, 6, 7, 8 6 7 I 8

.----. ---_... -,------J oL ----,-, --,----,----J Lo 20,000 40,000 o 500 1,000 1,500 INTENSITY OF MAGNETIC FIELD (OERSTEDS) INTENSITY OF MAGNETIC FIELD (OERSTEDS)

ENERGY OF A SODIUM ATOM, plotted as a function of mag nucleus is very small. In a weak magnetic fie ld the interaction of netic-field intensity (le/t), appears to increase along two straight the electron spin and the magnetic moment of the nucleus gives lines, corresponding to the parallel and antiparallel orientations of rise to an even more complicated "hyperfine" structure (right). The the atom's spinning electrons with respect to the magnetic field. In arrows indicate the transitions between the closely spaced energy actuality the magnetic moment of the nucleus also makes a minute states that can be studied by the molecular-beam resonance method. contribution to the total magnetic energy of the atom. Thus the two Transition a is comparatively easy to study because the energy dif lines on the graph are actually each a set of four lines, the lines ference and hence the frequency is much lower than for the other being very close together because the magnetic moment of the transitions. Transition c is suitable for use in an "atomic clock." on our graph are actually each a set of known for some time before Rabi's ex view was difficult because spectral lines four lines, the lines being very close periments, through the study of spec are often not sharp enough to allow the together because the magnetic moment troscopy. As the first crude spectro clear resolution of hyperfine structure. of the nucleus is very small and makes scopes were improved late in the 19th Molecular beams were able to help only a minute contribution to the total century it was found that some of the here in two different ways. In the first magnetic energy of the atom. spectral lines were complex. For in place they could be used to produce In a weak magnetic field the situation stance, the yellow sodium line was sharper spectral lines. In a flame or an is further complicated by the fact that found to be a pair of lines differingin electric discharge the atoms that emit both the electron and the nucleus may wavelength by about .1 percent; similar light are flying about with speeds up to be thought to precess in a complex fine structure was found in many other a mile a second or so. When an observer manner. Such pre-quantum-mechanical lines. The standard explanation, given at rest receives light from a moving ways of speaking are of little use, but by Goudsmit and Uhlenbeck in 1925, source, he receives more wave crests per quantum mechanics enables us to work was that the electron can spin either in second if the source is moving toward out the allowed energy values [see the direction of its orbit or in the op him and fewer wave crests if the source illustmtion at Tight above J. At zero posite direction, the two possible states is moving away. This phenomenon field there are just two energy values, possessing energies that differ by about (known as the Doppler effect after corresponding to the electron spin .002 electron volt. Christian Doppler, who first predicted being aligned parallel or antiparallel to As the spectroscopist's tools became it in 1842) causes a spread in wave the nuclear spin. These two states have more delicate, each of the sodium lines lengths amounting to several parts in a net spins of I equals + 1/2 and I equals was itself seen as a group of several million and obscures some details of -1/2 respectively. For sodium this lines, and many other spectral lines hyperfine structure. If the source con works out to two and one net spins revealed a similar "hyperfine"structure . sists of atoms in a beam, however, and respectively; thus a weak field splits the Sometimes this structure is merely an the light that is emitted is observed ap energy values into five and three lines, isotope effect: the element in question proximately at right angles to the direc corresponding to fiveand three allowed consists of several isotopes whose spec tion of the beam, the Doppler broaden nuclear orientations. As the intensity of tral lines do not exactly coincide. Usual ing of spectral lines becomes much the field increases, there is a gradual ly, however, the hyperfine structure is smaller. This trick was used successfully transition to a situation in which the caused by the interaction of the electron in 1928 by the Soviet investigators L. mutual interactions of the electron spin spin and the magnetic moment of the Dobrezov and A. Terenin; in fact, they and nuclear spin become insignificant. nucleus, as I have explained above. were the first to observe and measure All the foregoing phenomena were Complete confirmation of the latter the hyperfine structure of the yellow

© 1965 SCIENTIFIC AMERICAN, INC sodium lines. Up to that time hyperfine netic moments to an accuracy many the A field were brought together again structure had been observed only in times greater than before. The only two by the B field; thus the detector was heavier elements, where it is more pro quantities that had to be measured able to record a good fraction of the nounced and where the Doppler broad were a radio frequency (number of original beam. ening is less. A number of other oscillations per unit of time) and a The detector used by Rabi in much spectroscopists have since used molecu magnetic-field intensity. Time measure of his work deserves a few words. Irving lat·-beam sources, excited either by elec ments are among the most accurate Langmuir-the American pioneer in vac tron bombardment or by intense light. in physics, and a homogeneous mag uum physics-had discovered in 1923 Very sharp absorption lines can also be netic fieldcould be measured, with that when alkali atoms strike a white observed by studying the light that has care, to a few parts in a million. Rabi's hot tungsten wire, they tend to lose an passed through a molecular beam. apparatus still contained two inhomo electron to the tungsten and to depart What Rabi did was something quite geneous fields, but their precise field from the wire as positive ions. John B. different: he used molecular beams to gradient no longer mattered. They mere Taylor, working with Stern, demon detect transitions between the closely ly served as a polarizer (an A field) to strated in 1929 that this phenomenon spaced energy states that result from select molecules of a particular orien provided a convenient method for de the action of a magnetic field on an tation and as an analyzer (a B field) to tecting alkali atoms quantitatively: one atom or a molecule. The idea of study test for the reoriented molecules [see simply had to collect the ions on a posi ing such transitions had occurred to bottom illustration on page 62]. The tively charged metal plate and measure others; the Dutch physicist C. J. Gorter transitions from one orientation to an the current they produced. had tried it, but without success. To other took place in an intermediate third A tungsten wire can also detect a cause the transitions was quite easy: a field (a C field), which had to be both beam of alkali halide molecules, and small shortwave radio transmitter pro homogeneous and accurately known. Rabi used such a beam (consisting of duces quanta of the required energy. The center region also contained a loop potassium chloride molecules) in his first To detect the transitions was much through which an oscillating current test of the new resonance method. He more difficult, however, and Rabi real from a small radio-frequency generator deliberately chose a molecule contain ized that molecular beams were ideally was passed. The two inhomogeneous ing an even number of electrons, with suited for this. fields were so arranged that the atoms their spins aligned in such a way that selected and somewhat spread out by their magnetic moments cancel one an- Magnetic-Resonance Experiments

Stern had once posed the question of whether potassium atoms, aligned in a particular direction in a magnetic field D (by splitting the beam in a Stern C Gerlach field and then blocking one of o . the two halves), would occasionally flip _._ .... over into the opposite orientation if 90 they were passed through a region in o which the field direction changed rap o ;:::::- idly. By testing potassium atoms in a zw U second Stern-Gerlach field Emilio Segre cr: w and I had found in 1932 that such flip 80 over processes did indeed occur. Several eo. theorists took an interest in our work � - I- fields in which the atoms experienced 70 u (f) variations only as a result of their rap zw I- id motion through the field. Rabi was Z o the first to use an oscillating radio frequency field to obtain "resonance,"

that is, transitions in a narrow range of _•... . 60 frequencies only. I still remember the thrill of see ing the first nuclear-magnetic-resonance I curve, published by Rabi and three of his students (Jerrold R. Zacharias, Sid ______-L _____ ney Millman and Polykarp Kusch) in 50 � j,_�__ --,- ___ The Physical Review of February 15, 110 112 114 116 118 120 CURRENT IN MAGNET (AMPERES) J�� 1938 [see illustration at Tight]. Their very first attempt showed a sharp res FIRST NUCLEAR-MAGNETIC-RESONANCE CURVE, observed at radio frequency with onance peak; its width was less than a beam of potassium chloride molecules, was puhlished hy I. I. Rahi, Jerrold R. Zacharias, 2 percent of its height. Clearly here was Sidney Millman and Polykarp Kusch in The Physical Review of Fehruary IS, 1938. Their a method for measuring nuclear mag- first aUempt showed a sharp resonance peak, with a width Jess than 2 percent of its height.

© 1965 SCIENTIFIC AMERICAN, INC other. The only net magnetic moment mains unbeaten, however, when it ditional spectral lines in both the infra of such a molecule is due to its nuclear comes to measuring the magnetic mo red and the ultraviolet regions of the spin and molecular rotation (as in the ments of rare species of nuclei-for hydrogen spectrum; the existence of case of hydrogen). As a consequence instance short-lived radioactive nuclei, these lines was verifieda short time very long and fine beams had to be used of which only minute amounts can be later. Schrodinger's wave equation gave in order for the beam to be split appre obtained at a time. Moreover, many a new meaning to Bohr's model, remov ciably in an inhomogeneous A field. On properties of free atoms and molecules, ing its arbitrariness and some of its in the other hand, the C field could be predicted by quantum theory and consistencies. What is more, Schrodin made quite strong without getting ex roughly tested by the methods of opti ger's equations made it clear how Bohr's cessively high resonance frequencies; as cal spectroscopy, have been investigated hydrogen model could be applied to we have seen, in a strong magnetic field in much finer detail by the molecular other atoms. Dirac modifiedthe wave the mutual interaction of the nuclear beam resonance method. Foremost equation to bring it into accord with spin moments and the molecular-rota among these latter applications was the relativity theory and found that the spin tion moments is negligible. Lamb-Retherford experiment, which of the electron followed naturally from Today, 27 years after Rabi's first was performed at Columbia in 1947. his modification. He was able to com resonance experiments, perhaps a dozen pute the fine structure of the spectral laboratories are still using his method to The Lamb-Retherford Experiment lines of hydrogen without any ad hoc determine nuclear magnetic moments, assumptions, but it was difficult to de and there is no sign that its applications The spectrum of the hydrogen atom cide whether or not the theory really have been exhausted. In recent years has been the touchstone of atomic phys agreed with the facts. The fine struc some competitors have appeared; for ics from the very beginning. The regu ture of hydrogen's spectral lines is par example, it is now possible, thanks to larity with which the lines in the spec ticularly fine,and the lines are strongly methods devised by Felix Bloch at trum are spaced had been noted by broadened by the Doppler effect(be Stanford University and Edward M. the Swiss schoolteacher Johann Jakob cause hydrogen atoms are so light and Purcell at Harvard University, to mea Balmer in 1885. Bohr's first model of hence fast). For nearly 20 years the sure nuclear magnetic moments in sol the hydrogen atom accounted for evidence for Dirac's predictions re ids. The molecular-beam method re- Balmer's observations and predicted ad- mained inconclusive. Some spectrosco-

15' ------.---

00 2P3/2

UJ1- 10 -' 0 > z r 0 > 0::: l') 1- 0::: U w w -' 2S 1/2 Zw � >- l') 0::: 2S ' 2 2P ' 2P' 2 w / /2i / i 5 Zw 5

; ; �€�e�==J: I 8 ______...... ___. ____ .. __ __---' o ___ o

______INTENSITY OF MAGNETIC FIELO--?> O jI�' �1 I

FIRST EXC ITED STATE of the hydrogen atom, which corresponds to the second energy IN A MAGNETIC FIELD the Dirac theory level in the diagram at left, was expected (according to the Dirac theory ) to be split into predicted that the three energy levels of the three closely adjacent energy states (center ), of which the so·called S» level coincided ex first excited state of hydrogen should be actly with the P» level. Using the molecular-beam·resonance method, Willis Lamb and R. C. split further, into a total of eight states. Ac Retherford at Columbia found that these two levels did not iu fact coincide but were cording 10 quantum theory 18 of the 28 con separated by about 1,000 megacycles (right). This separation is now known as the Lamb shift. ceivabJe transitions between tbe eight states

© 1965 SCIENTIFIC AMERICAN, INC pists found small deviations from the To obtain hydrogen in the desired states would then immediately radiate theory, whereas others thought all was state, Lamb and R. C. Retherford first a photon and drop to the ground state, well. had to break up the hydrogen molecules losing its ability to release an electron When World War II ended in 1945, into atoms. This was done by using a from the metal plate. many physicists retumed to their labo tungsten oven at intense white heat. The procedure adopted by Lamb and ratories keen to take up pure research The escaping atoms were then bom Retherford was similar to that of Rabi's once more and to use the new tools the barded by electrons with enough energy first experiment. The radio frequency war had forced them to design. In par to raise them to the first excited state. was set successively at a number of ticular the development of radar had Atoms that were raised to one of the values, and each time the magnetic field produced new microwave techniques so-called P states would retum almost was varied until a dip in the electron and increased the scope of the reso at once to their ground state, emitting current indicated that a transition was nance method for studying atomic a photon, or quantum of light. The S taking place. The results were in strik transitions. At Columbia, Willis Lamb state, on the other hand, is metastable; ing disagreement with Dirac's theory; decided to use the resonance method to that is, it lasts a much longer time to explain them one had to assume that study transitions between the three long enough to let the atom complete its the S),> level did not coincide with the closely adjacent energy states into which joumey through the apparatus. More P),> level as expected but was about the Dirac theory had split the first ex over, a metastable atom is easy to de 1,000 megacycles higher. This shift cited state of the hydrogen atom. In tect: when it strikes a metal plate, it now generally called the Lamb shift-is the presence of a magnetic field these tends to liberate an electron and thus only about half the typical Doppler states should be split further, into a betrays its presence. Only one hydrogen broadening of an optical line of hydro total of eight states. According to the atom in several million was raised to gen. No wonder the spectroscopists had "selection rules" of quantum theory 18 the metastable state, but since a fairly been unable to pin it down! of the 28 conceivable transitions be broad beam could be used, there were The theoretical physicists rallied tween the eight states should be for enough atoms to perform the measure quickly from their surprise, and within bidden, leaving 10 transitions to be ob ments. To detect the transitions was a few weeks Hans Bethe produced a served, of which five were expected to also easy: an atom that made a transi complete explanation of the Lamb shift. be very feeble. tion from the S state to one of the P Actually they had not been unprepared 15,000 · la

-- - - 1 I _ ------f"' - - - - ��-r------l 2 - - - if)w - - - - - [:?:�:�2:::l==:====:===�======J- 10'000 � --- _

3 �w b � ------>------� -- 4 w :::J 5 8 0:: LL o 5,000 1------!------" is 0::« =J Ie

------� - - -, - - ",, --- o ,:::.:::- ) 1,000 2,000 3,000 o o INTENSITY OF MAGNETIC FIELD---l> INTENSITY OF MAGNETIC FIELD (GAUSS)

should be forbidden, leaving 10 transitions RESONANCES WERE OBSERVED by Lamb and Retherford at the positions marked by to be observed, of which five were expected the small open circles on this graph; according to the Dirac theory, resonances should have to be very feeble. The remaining five are been ohserved along the broken lines. To explain the discrepancy theorists were forced to shown here in the positions predicted by the pursue Dirac's calculations beyond a first approximation. The best theoretical figurefor Dirac theory (left) and in the positions the Lamb shift is now 1,057.2 megacycles, whereas experiments indicate 1,057.8. These found by Lamb and Retherford (right) . calculations also successfully predicted a slightly larger magnetic moment for the electron.

© 1965 SCIENTIFIC AMERICAN, INC 3D5/2 atom spends in the C field. Hence we 3P3{2- get a resonance whose peak, under op 3D3{2/ timum conditions, can be located to 3S1/2'-. perhaps a hundredth of its width. 3P'/2/ A typical hydrogen molecule takes n l 1 If 1 about 10-4 second to travel an inch; in a c e order to measure its frequency to within b d f 9 10 cycles per second one needs a C 11 1 field at least 10 inches long; this creates 2P3/2- a problem because it is quite difficult to

e keep the magnetic field sufficiently con stant over such a distance. Norman F. f Ramsey, who first worked with Rabi and then started his own school at 9 Harvard University, found an ingenious way around this difficulty.He realized 2S'/2- that the two ends were the most im 2P'/2 portant places for the comparison of EVIDENCE OF LAMB SHIFT in the second energy level of helium is visible in the spectro two wave trains, so instead of exposing gram at left, made by C. Herzberg of the National Research Council of Canada_ The spectral his molecules to radio frequency all lines designated by letters are produced by transitions between the hyperfine energy levels along their path through the C field he indicated in the chart at right. The lines designated c and d would coincide if the Lamb shift merely placed one short coil at one end (which is the splitting between the levels marked 2SYz and 2PYz) were zero_ Lines a and b and a second coil at the other. This ar are not as well separated, because their separation is the difference of the splitting in the up rangement sharpened the resonance by per and lower states, whereas the separation between c aud is the sum of the Lamb shift in d as much as if he had lengthened the C the upper and lower states_ The other spectral lines are not affected by the Lamb shift_ field by 60 percent. Moreover, small irregularities in the magnetic field no for discrepancies, since Dirac's calcu pect of the theory had to be tested, and longer mattered. lations had not been pursued beyond a change of a few parts in a million of a first approximation. The reason was the magnetic moment of the electron Atomic Clocks that every attempt to go to higher ap would have indicated that the theory proximations had produced nonsense; had broken down. Sometimes the value Ramsey's method was an important even the energy content of a free elec of precision is not so apparent. The step toward the development of atomic tron came out infinite. Lamb's experi knowledge of nuclear spins and mag clocks. Every clock (except for water ment now forced theorists to tackle netic moments has been a great help in clocks and similar primitive devices) these infinities in earnest. If the energy constructing nuclear models, that is, in depends on a frequency standard, such of a bound state of the electron came deciding what mathematical approxima as a pendulum or a balance wheel, out infinite, one had to deduct from tions are useful in tackling the appall coupled to some kind of gear that dis- it the energy of a free electron at rest, ing complexity of atomic nuclei. It is which was also infinite. That is what precisely because of this complexity Bethe did, and what others did after that no one has yet tried to predict a him with greater refinements. To com nuclear magnetic moment to anything pute the difference between two infinite like the experimenter's five-figure ac quantities seemed a somewhat unsound curacy. All the same, aiming at high procedure, but it worked remarkably precision is usually good policy and is well. A race for precision between the rewarded in the end. theorists and the experimenters ended The fundamental limit to the ac with excellent agreement: the best the curacy with which a frequency can be oretical figure for the Lamb shift is measured is the time during which an 1,057.2 megacycles, whereas the experi oscillating system can be kept under ments indicate 1,057.8 megacycles, a observation. If one were to compare difference of less than .1 percent. Both two trains of waves with nearly the groups may well be proud of their same frequency, it would be evident achievement. The theorists gained a that they can be quite accurately great deal of confidence in their tech matched for a time but that they even niques, and much has been done since tually get out of phase. The closer the to put those mathematically doubtful frequencies are, the longer it takes for subtraction procedures on a sounder the phase difference to show up. Dur footing. It might be added that these ing the transition between two energy calculations also indicated a slightly states an atomic system has a wave larger magnetic moment for the elec function that oscillates with a frequency AMMONIA MASER, invented by Charles tron, and that too has been very well that is proportional to the energy dif H. Townes in 1954, was based on a princi confirmed by atomic-beam experiments, ference between the states. Transitions ple enunciated by Einstein in 1909, namely in particular by Kusch. will be caused by an external oscillat that the presence of a light wave could cause In this kind of work the need for ing field if the two frequencies are rea an atom to lose energy rather than gain it; precision is obvious; an important as- sonably matched during the time the the light wave, on passing through such a

© 1965 SCIENTIFIC AMERICAN, INC plays the number of oscillations that well-defined atom or molecule and the frequency. Even so, a commercial have occurred. Since the frequency should not be affected by external con version has been produced and is at standard is under continuous observa ditions. Most of the transitions shown present the most accurate clock in gen tion, the difficulty I have just men in the graph at the right on page 66 eral use. tioned does not arise. But a pendulum, occur at frequencies that depend strong left to itself, will lose much of its ly on the magnetic field, but there is one Masers and Lasers energy in a few hundred oscillations; in the region where the effect of a weak there must be a driving mechanism to magnetic field is very small. All alkali In a resonance experiment of the replace the energy loss, either continu metals have such a transition, but Rabi type one can select either the ously or from time to time, and if that cesium seems to be most suitable. Its higher or the lower energy state for mechanism is not carefully designed, it atoms are heavy and will evaporate at passing through the C field. If the is liable to alter the frequency of the quite a low temperature; hence they are higher state is selected, the radio-fre standard. Clearly for high precision it slow and spend a long time in the C quency field will then cause transitions is desirable to use oscillators that con field. Their frequency is high, about to the lower state. The absorption of serve a high percentage of their original nine billion oscillations per second, radiation, which raises an atom from a energy. which provides a high fractional preci lower to a higher state, is a familiar With the advent of electronic devices sion for a given change in frequency. phenomenon, but Einstein had shown, it became possible to use standards that Cesium can be detected efficiently by as early as 1909, that the opposite oscilLtte at much higher frequencies; its ionization on contact with a hot transitions are just as likely. In the lat tuning forks that could oscillate several tungsten wire, and all its atoms are ter case the presence of a light wave hundred times a second were soon suc strictly alike because cesium has only would cause the atom to lose energy ceeded by quartz crystals capable of one stable isotope. rather than gain it; the light wave, on oscillating at frequencies 1,000 times The oscillations in a cesium clock are passing through such a material, would higher. Such crystals hang on to their made by a small microwave generator suffer "negative absorption" and get energy with remarkable tenacity, and whose frequency peak is kept adjusted, stronger, not weaker. Einstein used the for some years quartz clocks held the by a kind of electronic servomechanism, phrase "stimulated emission" for this record for precision, with errors of no to within 1 percent or so of the width phenomenon, thus setting it apart from more than a few seconds a year. They of the resonance peak. The best value the spontaneous emission by which an made it possible to detect small ir for the cesium frequency is probably atom in a high energy state will eventu regularities in the rotation of the earth, the one found by L. Essen in Britain, ally lose its energy if left to itself. which greatly interested geophysicists, namely 9,129,631,830 ± 10 oscillations Normally a light wave will be ab who began to ask for still higher preci per second; the ±10 is due mainly to sorbed rather than causing stimulated sion. That was not easy, however; no the difficulty in obtaining the astronomi emission, because in normal atomic two quartz crystals are exactly alike, cal observations by which the length of populations there are many more atoms and even a given crystal changes slight the second is measured. A cesium clock in the lower state than in the higher ly with time. is a fairly complex assembly of elec one. In an atomic-beam apparatus, how An ideal frequency standard should tronic equipment in which the cesium ever, one can create an inverted popu depend only on the properties of a beam merely serves to keep a check on lation in which the molecules in the

RESONANT CAVITY

BEAM OF AMMONIA MOLECULES

material, would suffer "negative absorption" and get stronger, not tion of the correct frequency was present, the beam had the expect weaker. The maser used an inhomogeneous electric field set up by ed effect of strengthening these oscillations_ When the pressure of four parallel metal bars to focus the higher-energy ammonia mole the ammonia was increased , the oscillations could be kept at the cules (color) toward the end of the field, whereas those in the lower same intensity with less input of microwave power ; finally the state (black) were pulled out of the beam. When the high-energy power could be switched off altogether and the beam then acted as beam entered a tuned cavity (right) in which a microwave oscilla- a generator of microwave power_ A cross section is shown at left.

© 1965 SCIENTIFIC AMERICAN, INC This was an entirely new principle. It was not simply microwave radiation I � � 1 from the molecules, which was much '\ too feeble to be detected. Yet there was \ ,0"" \i ) no need for a separate microwave gen erator after the initial stages of the � ":� \, " � experiment. It was enough to let an �\" inverted-population beam enter a prop � 1 I I erly tuned cavity; the cavity then be � gan to "ring," or oscillate electromag \ � netically, extracting energy from the / I \ " \\ beam by the process of stimulated emis " sion. Einstein's theoretical insight had 1 '\ � ,� J , waited 40 years before Townes found �\ a way to exploit it. /" Once a way had been shown it was "" # r \ quickly realized that there were many

When you ask, here are some of the answers you'll get : Garrett's portable gas turbine power pack and inflatable or expandable shelter give you all the comforts of a completely con trolled environ men t - whether your base is in the middle of the Sahara one week or moved to the Arctic Circle the next. A good example of these new I t works this way: field units is the MUST complex The components are designed The power pack provides elec that Garrett-AiResearch is build and produced by our two tricity, refrigeration, air heating, ing for the U. S. Army Medical AiResearch Manufacturing Divi air pressure, and water heating Service. MUST means Medical sions and Air Cruisers Division. In and pumping. Unit, Self-contained, Transport addition, we also offer total system Yo u simply connect water and able, and comprises a complete management responsibility as well fuel lines. Start the turbine which 400 bed evacuation hospital. as integrating the componen ts inflates and deflates the structures Outputs from a typical MUST into the complete system. pictured here, or set up the power pack include : 90 kw 400- expandable shelter -and you have cycle and 10 kw 60-cycle electrical a modern, well-lighted, tempera power, 20 tons of refrigeration, ture-controlled facility -fully pro 520,000 BTU /hr. air heating, tected from sun, wind, rain, snow, 60,000 BTU/hr. water heat ing. ice, heat, humidity, dirt and The heart of the power pack is insects. a small gas turbine -the only type Yo u need just five men and half of power that can give you so an hour to get the units up and many outputs from a single light Ty pical applications of the new operating. weight unit. field units are : chemical and bio How do you transport the units? Only Garrett has so much logical protection; radar installa Easy. experience in designing and apply tion, communications bases, civil They come packaged in con ing small gas turbines to this type defense shelters; and oil, mining, tainers that fit nicely on the bed of of system. and construction operations. a truck or in a helicopter sling or Please direct inquiries to in the cargo hold of an airplane. AiResearch Manufacturing Divi sion, Phoenix, Arizona. Garrett • • lS exper Ience • AiResearch Manufacturing Divisions Los Angeles · Phoenix

© 1965 SCIENTIFIC AMERICAN, INC bouncing around for a second or so be that is, unless the relative speed of the who has fore they escape again through the colliding molecules is high enough. An small hole by which they entered. The intriguing technique for giving the flask itself is in a tuned cavity and its molecules extra speed was developed the answer inside is painted with Teflon, an inert recently by Philip B. Moon of the Uni material that reflects impinging hydro versity of Birmingham. He found he gen atoms without any significant effect could cause a spindle-shaped steel rotor to EDP on their transition frequency. At first to rotate at more than a mile a second in it may seem strange that a frequency a vacuum; if a small amount of gas was economy? standard can bounce around in a bottle, admitted, those molecules that got in undergoing 100,000 collisions or so a the way of the rotor tips were slapped second, but the proton spin inside a into rapid motion, and some could EDP systems designers, builders and hydrogen atom is well protected from escape through a small hole into an users who are fighting the traditionally outside disturbances, except for mag adjacent evacuated space, producing a high cost of computing. That's who. netic ones, and Teflon is inert precisely molecular beam with a high and ap They specify system components tha t because all its valences are saturated proximately uniform speed, fast enough do the job better for less money. For example, they specify computer mag and so its molecules develop practically to make them react chemically with netic tape units from Datamec. Either no magnetic field. other molecules with which they were the 02020 or the 03030. Both are well Because the atoms stay so long in the allowed to collide. known for setting new industry stand cavity, its tuning has only a minute A different method, which gives ards in greater all-around economy: effect on the frequency that is gener beams of high intensity as well as high lower initial cost, reduced maintenance expense, greater up-time, higher per ated (less than one part in a billion). speed, is the use of a high-pressure fo rmance rel iability. Moreover, this effectbecomes less as nozzle instead of the usual oven slit. the beam is made stronger, and it can Usually the pressure in the oven is kept The 02020 is an attractively-priced unit for computer and off-line applications be minimized by tuning the cavity until so low that the molecules escape with where moderate speed performance is a change in beam intensity is seen to out much jostling, each with the veloci highly practical (data transfer rates up make no difference. Ramsey has shown ty it had in the oven. Only the slowest to 36,000 characters per second). that two such hydrogen clocks, after molecules tend to get pushed from be being adjusted independently, will keep hind, and indeed one observes fewer The 03 030 offers the same unprece time to better than one part in a tril dented economy and reliabili ty for slow molecules than one would expect heavy duty, on-line use with dig ital lion; neither will gain over the other by from Maxwell's formula. If the pressure computers and other digital EDP sys more than a thousandth of a second in is raised, more molecules get pushed to tems requiring higher data transfer 30 years! a higher speed, and eventually one gets rates (up to 60,000 characters per a gas jet that both expands and cools. second). Some Recent Innovations By the time the molecules emerge from the nozzle they are several times faster There are a number of less spectacu than normal beams, but the spread of lar uses of molecular beams. For in velocities has become less. When such stance, high-energy physicists are cur a "nozzle beam" strikes a solid surface, rently interested in using beams of po the surface is subjected to a bombard larized protons (that is, protons with ment somewhat like that to which a their spins pointing predominantly in spacecraft or the nose cone of a rocket one direction). Fast protons can be is subjected on reentering the atmo polarized by selecting those that have sphere. The laboratory study of the been deflected through a certain angle effects of such bombardment has at by, say, carbon nuclei. This is a most tracted the interest of designers of inefficient method, however, since only space vehicles. a minute fraction of the protons in the Molecular-beam research started out original beam are deflected in a given as a "family affair" more than 40 years Some 70 leading companies al ready direction. A molecular-beam method ago. For many years it was all con use Datamec computer magnetic tape can do much better, and sources of slow ducted, with a few exceptions, by Otto units in their data systems. Like to read polarized protons, ready for acceleration Stern and his students and their stu over the list of people who have the in whatever machine is available, have dents. Stern has every right to be proud answer to EDP economy ? Write Tom been developed in several laboratories. of the enterprise he started; in the im Tracy at Datamec Corporation, 345 Middlefield Road, Mountain View, Calif. Collisions between molecules can be mediate "family" alone its success has studied under controlled conditions by resulted in five Nobel prizes-to Stern letting a molecular beam pass through himself in 1943, to Rabi in 1944, to a dilute gas or, even better, by cross Lamb and Kusch in 1955 and to Townes ing it with another beam. It is possible in 1964. The applications of molecular in this way to study the chemical reac beam techniques now extend into all tions between gases and to observe the branches of physics and can no longer be single steps that are difficult to separate said to be strictly a family affair. New @)5J.7?5J.um@� when the reaction takes place in a mix projects are being undertaken in labora leadership in low-cost/high-reliability ture of gases. Of course, gases often do tories all over the world, and new sur digital magnetic tape handling not react unless they are hot enough, prises may well be just around the corner.

SCIENTIFIC AMERICAN

Announces a single-topic issue concerned with the urbanization of the world population under the title of CITIES

To be published in September, 1965

Photography, we admit, is the awkward step child of the visual arts. And understandably so. Because photography, if it is an art form, is the only one that isn't a continuous creative process. Oils, water colors, sculpture -you name it-all let you cre ate one step at a time. If you make a mistake, you learn from it right then. If you're inspired, you can take advan tage of your inspiration immediately. Yo ur technique can improve and your satisfaction can increase without interruption. But the pho tographer who must wait to see his pictures doesn't have the advantage of

A minute late?', getting better but still not pel,/ect. Now get a little lowel' and wait fol' the j' ight moment, ..

waiting period.) Small wonder there aren't more "Sunday" photographers. The one exception to all this : a Polaroid Land camera. We think it's the key to an enor mously satisfying way to express yourself. If you take black qnd white pictures, you see them in 10 seconds.' If you prefer color, one minute is all it takes. And because you see your results so fast, you soon learn the aesthetic values of fine photography. Interpretation, composition, lighting and total effectare all

First try, too fal' away and not vel'y well composed. apparent, and changeable, on the spot. The And so impersonal it could be anybody's family. challenge and the satisfaction become a little sharper with every picture you take. that continuous experience. He's practicing a stop-and-go art form. When the average per Whether you're a camera bug or not, we son takes a picture he doesn't know how it suggest that you borrow a Polaroid Land cam turns out 'til days later. Even if he's interested era and try it out. We can almost guarantee in profiting by his mistakes, it's too late. The that your second picture will be better than location and lighting are different. The mood your first. And that your third will be better is gone. The crea tive continuity is shattered. than your second. (Even getting involved in processing your own And that's what Polaroid Land photography pictures doesn't do much except shorten the is all about.