P H Y S I CAL REVI EW VOLUME 127, NUMBER 3 AUGUST 1, 1962

Millimeter Wave Molecular Beam Spectroscopy: Alkali Bromides anti Iodizes*

JAMES R. RUSK AND WALTER GORDV
Departznent of Physics, Duke University, Durham, tVorth Carolina
(Received March 22, 19(i2}

The pure rotational spectra of the alkali bromides and the alkali iodides were investigated in the 1.5 to
5.0 mm range of the microwave region. The experiment was performed by passing a beam of molecules in the
vapor state from an oven, capable of producing temperatures up to 1000'C, into an oversized section of
wave guide so that the direction of incident microwave radiation was at right angles to the flow direction
of the molecular beam. In this way linewidths less than 100 kc/sec at 100 000 Mc/sec were easily achieved.
The line frequencies were measured with a precision better than one part in 10'. The apparatus was de-
veloped as a series of modifications on an earlier experiment performed at this laboratory. Dunham's theory
for diatomic molecules was applied in interpretation of the data. In most cases, the values for B„o.„andp,
were known although less accurately, but D, and P, were found for the first time. From the latter two, ac-
curate determinations were made for the first time of cu, and co,x.. From these values, information has been
obtained about the potential functions, which suitably describe these diatomic alkali halides. In addition,
centrifugal distortion constants, isotope mass ratios, moments of inertia, and internuclear distances have
been accurately evaluated for each molecule.

INTRODUCTION high temperatures necessary for vaporization of these

'
ARL'lER it was demonstrated in this laboratory molecules. Their method is similar to ours in that it
~
'-& employs a radiation detector rather than the morc
that rotational absorption spectra of collimated
specialized molecular detector of the electric and
molecular beams at high temperature can be directly
detected in the millimeter wave region by observation magnetic resonance methods. It differs from ours in
of the dip in the detected power as the frequency of the that it does not employ molecular beams to avoid, or
radiation source is swept through the spectral fre- reduce, the considerable Doppler broadening and
collision broadening encountered by a molecular system
quency. 'We have improved this method and have used
it to measure precisely millimeter wave transitions for in thermal equilibrium at several hundred degrees
most of the alkali halides. centigrade.
Our molecular beam spectrometer complements
The electric-beam resonance method as originally
rather than overlaps the applicability of the earlier applied by Hughes did not measure rotational transi-
molecular beam resonance methods'' which employ tions, and hence did not give moments of inertia but
molecular rather than radiation detectors. The electric
rather the product of the dipole moment and the
moment of inertia. Although the product pI could be
resonance method' is applicable principally at radio
frequencies or in the long-wave microwave region where
obtained very precisely, the separate values of and JM, I
could be learned only approximately. Later, however,
the molecules are sensitive to Stark fields, whereas the
present, direct-absorption method is principally appli- Lee, Fabricand, Carlson, and Rabi" were able to
extend the method to the centimeter wave region where
cable at the high millimeter or submillimeter wave
frequencies, where the absorption coeKrients of the lowest J rotational transitions of a few of the alkali
halides could be reached and measured very precisely.
molecules become large. As is seen from the results in
this paper, complete and accurate solution for the
EXPERIMENTAL ASPECTS
spectral constants of diatomic molecules requires
measurements at a number of frequencies, including 3fillimeter 8'ave ass-d Electrortic Comportents With.
those in the millimeter wave range. the exception of the absorption cell, the spectrometer
Rotational absorption spectra of many alkali halides is very similar to the millimeter and submillimeter
have been measured earlier by Honig, Mandel, Stitch, wave gaseous spectrometer which has been used for a
and Townes4' in the centimeter wave range with a number of years in this laboratory. ~ The frequency
conventional microwave spectrometer having a specially measurements were made with a secondary standard
designed absorption cell which could be heated to the monitored by the 5 Mc/sec standard frequency broad-
cast by station WWV, in a manner described by
* This research was supported by the U. S. Air Force Ofhce of Unterberger and Smith. ' The millimeter-wave harmonic
Scienti6c Research of the Air Research and Development generator and detector are those developed and de-
Command.
' A. K. and W. Gordy, Phys. Rev. 10S, S99 (1957).
scribed by King and Gordy. The fundamental source
' H. K. Garrison
Hughes, Phys. Rev. 72, 614 (1947).
'C. A. Lee, B. P. Fabricand, R. O. Carlson, and I. I. Rabi, 'B. P, Fabricand, R. O. Carlson, C. A. Lee, and I. I. Rabi,
Phys. Rev. 91, 1.395 (1953). Phys. Rev. 91, 1403 (1955).
4 M. L.
Stitch, A. Honig, and C. H. Townes, Rev. Sci, Instr. 7 W. C.
King and W. Gordy, Phys. Rev. 90, 319 (1953).
25, 759 (1954). ' R. R. Unterberger and W. V. Smith, Rev. Sci. Instr. 19, 580
' A. Honig, M. Mandel, M. L. Stitch, and C. H. Townes, Phys, (1948).
Rev. 96, 629 (1954). 9 W. C.
King and W. Gordy, Phys. Rev. 93, 407 (1954).
817
 J. R. . lit) SK AN D W. GOODY

The bottom section was prepared from a solid block of

steel by hollowing it into a shell with an open top,
leaving enough room on the other three sides for drilling
OVEN
several holes the length of the oven. These hold the
loosely wound coils of molybdenum wire (0.030 in. in
diam) which were insulated from the steel by tubes of
FIG. 1. Sketch showing waveguide cell and molecular-beam high-temperature Vicor glass, or of quartz or thoria
source. M and D represent the microwave source and detector,
respectively. ceramic. The latter were needed only for the highest
temperatures. These heater coils, connected in series,
was an 8-mm EMI reflex klystron. As a multiplier are joined to Kovar seal terminal posts by means of
crystal we employed one of the special silicon crystals another pair of posts and strips of stainless steel. The
developed by Ohl, of the Sell Telephone Laboratories. thermocouple lead is mounted loosely next to the oven,
These crystals and the special treatment to enhance and it is insulated and sealed in the bottom plate with
their multiplying characteristics are described by Ohl, an epoxy resin. The Chromel-Alumel thermocouple
"
Budenstein, and Burrus. We are grateful to C. A. leads are connected to a Leeds and Northrup po-
Burrus, now at the Bell Telephone Laboratories, for tentiometer, where the temperature is read in millivolts.
kindly supplying us with the particular crystals The top section of the oven contains two more heating
employed in this work. coils and the collimating apertures, which will be
Absorption Cel/ Fig.ure 1 shows the arrangement of discussed later. The oven is relatively efficient up to
the absorption cell. It consists of a section of S-band temperatures of 700' to 800'C according to the exact
waveguide (dimensions: 2 —, ",)&1P, in. ) with a slot in condition of the radiation shields and the absence or
the narrow sides to allow the molecules from the oven presence of resistance leaks to ground.
and the collimator to pass across the cell perpendicularly Colli mator. For best collimation with maximum
to the path of radiation. As arranged in the diagram, cross-sectional beam area, long parallel tubes were
the molecular path is perpendicular to the E vector of stacked in groups. The theory of long parallel tubes for
collimating purposes has been fully discussed by
the radiation, but this is not essential. Actually, in
most experiments a cell was employed which had a slot
"
Giordmaine and Wang. The 8-in. holes were drilled
cut only in the side where the molecules entered the in the top section as shown in Fig. 2, and each of these
waveguide. They were trapped on the opposite wall of was packed with about fifty stainless-steel tubes, 4 in.
the waveguide cell which was maintained at room long, with an inner diameter of 0.015 in. and an outer
temperature. In runs of normal length, the dielectric diameter of 0.020 in. Basically, this type of aperture
can be treated just like a channel slit (a slit with a finite
loss caused by salt collected on the waveguide wall
did not become excessive before the end of the run.
"
depth). The slit or diameter of the tubing should be of
Thin mica windows used to seal the cell were placed the order of the mean free path inside the source. A
across the small millimeter waveguide at the points of typical mean free path of an alkali halide at 1 mm
input and output of the cell. These were arranged at a pressure of mercury at 1000'C is about 0.010 to 0.020 in.
distance of about one foot from the molecular path to The linewidths which we observed indicate that
prevent scattered molecules from collecting on the adequate collimation was present, and the necessary
windows. Because the large S-band guide has little large number of molecules was present in the microwave
attenuation for millimeter waves, the extra lengths Geld.
required to achieve clean windows was not costly. Precision. In our experiments we chose to compromise
To connect the small G-band waveguide (0.075X0.034 the degree of molecular beam collimation to gain in
in. ), the output of the harmonic generator, to the large number of molecules passing through the cell. This
5-band waveguide, tapered horns 12 in. in length were compromise was justified because our electronic system
was not designed to take advantage of excessively sharp
employed. These'were electroformed in our laboratory.
The chief innovation in the present experiments ov'er lines. Even so, we were able to achieve with ease a
the initial one of Garrison and Gordy' is the use of the factor of precision, 6f/f, of 10 ' as contrasted with the
slotted S-band guide rather than matched horns with 10—4 to 10 5 achieved in the earlier hot-cell measure-
an open space between them, through which the mole- ments by the Columbia University group. 4' With a
cules were sprayed. The present cell proved to be more frequency-stabilized klystron and a high-fidelity re-
broad banded and easier to use in the search for lines. ceiving system, a more highly collimated beam would
Also, it seemed to be less lossy for the shorter millimeter
be justiled. With such a system one couM achieve a
waves. precision factor of 10 7 or better, but the measurements
would be slower and more tedious. At this stage we
O~ee. The oven, enclosed by the radiation shields,
was constructed in two parts, each of stainless steel. "J. A. Giordmaine and T. C. Wang, J. Appl. Phys. 31, 463
"R. S. Ohl, P. P. Budenstein, and C. A. Burrus, Rev. Sci. Instr.
(1960).
' N. F. Ramsey, Mnlecllar Beams (Oxford University Press,
So, T65 (1959). New York, 1956), pp. 11—25.
 M ILI. I METEP WAVE MOLE CULAR. BEAM SPECTROSCOP Y

where
B.= B.—~.(e+-')+V. (s+ l)'+ (4)
D. = D.+P.(a+ 0)+
FrG. 2. The upper ov o I

II„
II„—
(5)
figure is a sectional dia-
gram of the oven. The
lower 6gure represents ~= vibrational quantum number, and rotational J=
the top section. Heating quantum number. These constants are further dehned
coils are inserted in the
round holes. The re- Removable P)ug or interrelated by the following equations":
movable plug allows
filling of the oven. B.= 0/ (Sx'I.), (6a)
Ig=pfg (6b)
oooooooooo )

D, = (4B,0)/pp, s, (6c)
5&e &e &e
—D 8&e~e (6d)
chose to employ the more versatile video-sweep 8, 248,'
spectrometer with an incompletely collimated molecular
beam. In this way we could obtain much new informa- Q )~5~8 g 68 e
tion in a reasonable time. Later, more precise measure- (6e)
2
CO@ (dg
ments can be made on selected molecules for which the
information gained would seem to merit the extra
Equation (6e) is valid only if the Morse potential" is
effort. . is useful for comparison with Dunham's
assumed, and
THEORY potential. Here, p, is the reduced mass, r, is the inter-
Ro/ational and Vibrational' Coestaets. Like Honig
nuclear distance, and I,
is the moment of inertia.
The V's in the Dunharn potential function are related
et al. , ' we have applied the theory of Dunham" to
approximately to these spectral constants by
obtain various molecular constants from the measured
rotational frequencies. For clarity, it is necessary to Vlo=&e) V pi —B„Vip=—o.,
reproduce here certain expressions from this theory. V20= e~ e7 Vo2= —D.,
Dunham assumed a potential function represented by
the power series, Ypp= JJ.) Yn= Pe. —
Usually these approximations are adequate, but in some
+B J(J+1)(1 2 +3 4 + ) (1

where $= (r r, )/r„J —
is the rotational quantum
Ypi = B.(1+B.'Ppi/~'), (8)
number, and the a's and 8,
are constants. He
employed the tA'entzel-Kramers-Brillouin method to
obtain a solution of the wave equation for the vibrating ppy —(Ylp Ys]/4Y01)+(16+1Y20/3Y01)
rotor. The energy eigenvalues which he found are —8~~ —«2+4~~', (9)
expressed by in which
—(Y11Y10/6Y01 ) 1
T„,g= Q Y;,, (v+ ')'J'(J+1)',
, -
(2)
G1 ~

For most of the alkah halides B.s/&v. s is usually less

than 10 0 and. pp& is small; therefore the above appro»-
where e is the vibrational quantum number, and where matjon with Vp» — 8, is adequate.
;
V, , are coe%cients which depend on the molecular Only six of the Dunham coefficients, Vpy, Vp2 Vp3,
properties. The first few of these coe%cients can be V», V», V», are necessary for expressing the observed
directly related to the spectral constants in the more millimeter wave spectra of the alkali halides to the
familiar expression for the eigenvalues": accuracy of our measurements. If the selection rules
Wg, /fs. = a&. (e+-', ) —co.x, (e+-,')'+ +
he= 0 and hJ= 1 are applied, the frequency equation
1s
+BJ(J+1)—DJ'(J+1)'+
+& J'(J+1)'+ (3) @=2Ypy(J+1)+2Yyy(5+0)(J+1)+2Ysy(e+s) (J+1)
"j.L. Dunham, Phys. Rev. 41, 721 (1932).
+4Yps(J+1)'+4Ygs(e+0) (J+1)'
'4
G. IIerzberg,
(D. Van Nostrand
Molecular Structure and Molecular Spectra + Y»(J+1)'t (J+2)' —J'3. (11)
Company, Inc. , Princeton, New Jersey, 1950),
Vol. 1, p. 109. "P.M. Morse, Phys. Rev. 34, 57 (1929).
820 J. R. RUSK AND W. GORDY
Actually, the term in V03 is too small to have a measur- we could in no instance obtain an improvement on the
able eRect on most of the transitions which we quadrupole coupling constants eQq already obtained
measured. In no case were we able to measure FX, by Honig et al. ' with low-J transitions in the centimeter
reliably. wave region. We have therefore used their values for
Nuclear Hyperfirte Structure. In the above theory eQtt with the above theory to calculate the expected
nuclear interactions are neglected which must now be splitting or broadening of the millimeter wave lines so
considered. Because the electronic ground states of all that corrections for these effects could be made in the
the alkali halides are nonmagnetic, 'Z states, there is no evaluation of other molecular constants.
resolvable nuclear magnetic interactions. However, all Limezvidths. In spectra observed at high temperatures
the common isotopes of the alkali metals and all the the chief contributions to linewidth are pressure or
halogens except Ruorine have nuclear spins of —, ', or collision broadening, collision-with-the-wall broadening,
greater, and hence have nonvanishing electric quadru- and broadening caused by the Doppler eRect. Other
pole moments which can give rise to observable splitting factors which might be significant are quadrupole
of the rotational levels. The stable isotopes of bromine broadening and the natural linewidth arising from the
and iodine have relatively large nuclear quadrupole uncertainty principle.
moments. Nevertheless, the nuclear quadrupole Pressure broadening results from collisions between
couplings in the alkali halides are not large, because molecules. In a well collimated beam of molecules the
the high degree of ionic character in these molecules number of collisions should be negligible, but when a
tends to give the halogen a negatively charged, closed- buffer gas is intentionally introduced there is some of
shell structure which has a too nearly spherically this broadening, which is proportional to the pressure.
symmetric electronic cloud to give much coupling. The Collision-with-the-wall broadening is also negligible,
alkali atom has only an s electron in its valence shell but not zero, since the molecule may not always lose
and thus tends to have a spherically symmetric cloud enough energy upon the first collision with the wall to
about the nuclei whether the bonding is ionic or condense. Also, there are always some collisions in a
covalent. In this study we did not observe any nuclear molecular beam because of the differing velocities of
quadrupole effects of the alkali atom, and only in a the molecules.
few instances did we observe the hyperfine structure Much of the line broadening which we encounter here
for the halogen. In these instances, where there was a can be attributed to the Doppler eRect because an
single nucleus with relatively small coupling in a exceptionally well collimated beam of molecules would
diatomic molecule, the splitting of the rotational lines be required to insure that all the molecules move in a
could be calculated with first-order perturbation theory. direction normal to the radiation path. Any component
For this simple case the splitting of the rotational levels of molecular velocity parallel to the incident path of
is given by" propagation of radiation gives the frequency shift
C(C+1) (— I I
I+1) (I+—1) r = ro(1+~/c).
&o = eQtI (12)
2I (2I —1) (2I —1) (2I+3)
where Because of symmetry in the molecular beam, this shift
C = F (F+1) I (I+1)—I (1+1), — results in broadening of the line rather than in a
(13) frequency displacement. If a Maxwellian distribution of
F=J, 5+1, — —I,
, 1 I, velocities is assumed, the half-width of the line at
and where I is the nuclea, r spin and eQtt is the nuclear half-power points caused by Doppler eRects can be
quadrupole coupling const. ant, in which Q is the nuclear shown to be"
quadrupole moment and q is the averaged value of d v= (7.15/2)10—'vo(T/M) l.
(ct'V)/(cts') for the electrons at the nucleus where s is
along the bond axis. The nuclear splitting of the levels Although at room temperature in ordinary absorption
decreases rapidly with increasing J. At the high J cells this broadening is usually smaller than collision
values observed in the shorter millimeter wave region, broadening, it is probably the most important broaden-
the quadrupole splitting is not resolvable for the heavier ing observed in these studies, if nuclear quadrupole
alkali halides such as CsI, which we studied, even with eRects are neglected. In all the molecules studied here,
the high resolution made possible by the molecular the Doppler half-widths at 100 kMc/sec would range
beam method. This quadrupole splitting was resolvable from 70 to 120 kc/sec, if reductions caused by colli-
for the lighter molecules such as I.iI, and broadening mation were not made. It is evident that these Doppler
of the lines by the unresolved quadrupole interaction widths have been reduced in our experiment to values
was noticeable in some of the heavier molecules for less than half those expected for molecules in a normal
which only high-J transitions were observed. Because hot cell.
we observed only high-J millimeter wave transitions, In estimating the amount of Doppler broadening
~~ W.
Gordy, W. V. Smith, and R. F. Trambarulo, 3Iicromave one must first subtract the broadening caused by
Spectroscopy (John Wiley tie Sons, Inc. , 1953). unresolved quadrupole structure (which is seldom
 MILLI METER WAVE MOLECULAR BEAM SPECTROSCOPY 82i

negligible) and the natural width caused by the finite parisons with the best previously known data. Also
time of interaction between the molecule and the given for each molecule is a listing of all measured line
radiation field. The latter can be estimated with the frequencies and all calculated ones. The latter were
—
uncertainty principle as 2 vent h. For a typical molecule based on a least-square solution of each set of measure-
in these experiments this type of broadening is of the ments. Reagent-grade salts were used in every instance.
order of 5 to 10 kc/sec. I.ithinm Bromide. Lithium bromide is the lightest
Artatysis oj Data. The data were treated by the molecule of the group. Since the frequency range of all
following, least-square method. After the line fre- observations is set by necessity, the lowest levels J
quencies have been corrected for the F03 term, the observed are for the lightest molecule. As was noted
remaining set of equations is in the theoretical discussion, low values show the J
greatest nuclear quadrupole effects. Bromine 79 of
f (Isla'02I11 I21I12) (16) Li'Brrs has a quadrupole coupling constant egt7 of
where i equals the number of equations or measurements +37.20&0.45 Mc/sec in. the ground vibrational state. '
made on a molecule. If F=P,
f,s, the i equations can Although the nuclear spin of Li' is ~, the coupling with
be reduced to the five normal equations, J is so small that it has not been observed. Bromine 81
has a coupling constant of +30.71&0.40 Mc/sec.
(itF/itV;, s) =0, (17) Since both bromine isotopes have a nuclear spin of —, ',
which minimizes the sum of the errors squared. When the energy-level structures are similar, though they
solved, these five equations will give the least-square differ somewhat in the magnitude of their splittings.
solutions to the five I"
s. Since, in the process of solving The nuclear quadrupole interaction splits each level J
for some of the constants, differences of very large into (2I+1) levels, or four levels for a sPin of ss. These
numbers are taken which give very small answers, it is levels are designated with the F quantum number which
desirable to keep ten significant figures through the varies in integral steps from (J+ss) to — ss). If pure J
calculations. By use of the Crout method, '7 which takes rotational transitions are assumed, the selection rules
advantage of the characteristics of the typical desk allow for the transitions QJ=+1 and AF=O, &1. Of
calculator, these calculations can be made quickly. the four strong lines corresponding to AF=+1, only
The accuracy quoted on these mean values is based a doublet was observed, because each of the lines is a
on a calculation of the "standard error of the mean. " degenerate pair.
With ten to twenty measurements on a molecule, over As shown in Table I, three transitions were studied
90% of the measured frequencies would fall within 2n
of the least-square value, where a is the standard error TABLE I. Lithium bromide (Li'Br").
of the line frequency, which is found by use of"
Present results Previous results
n'= $dd)/(i —rtt), (18) F01'= 16 650.318~0.06 16 650.570+0.05 Mc/sec
where i is the number of equations, nz is the number of 8,'= = 16 650.179~0.10 16 651.186&0.05 Mc/sec
Bp = = 16 565.9369+0.006 Mc/sec
unknowns, and Pdd] is the sum of the squares of the Dp = = 64.705&0.12 kc/sec
deviations of the measured line frequencies from the e, = —P11= 169.09+0.08 Mc/sec
656&40s kc/sec
calculated ones. The standard error of a particular I' D, '= —Yp2 = 64.71+0.25 kc/sec
can be found from" eqQ/h= +37 20&0 45b Mc/sec
'= La„jn'/6, Derived constants Previous constants

where cr
n
is the standard error of Ltt I', „j
is the 4X4
(19)
co,
co,x.
= F10 —563.5&2.2
= —I'20 = 3.88W0.02'
480" '
170
cm '
cm '

„„,
determinant obtained when perpendicular lines are
drawn through the coefficient a „corresponding to the
I", and where 6 is the 5)&5 determinant of all the
coefficients.
I, =
fe
30.3618~0.0011

Transitions
v=0
2.17042~0.00004

Measured lines
30.36012
2. 1704+0.0001
amu A
A

Calculated lines

EXPERIMENTAL RESULTS
Millimeter wave spectra have been obtained for al v=2-3i
'F=7/2-9/2
F = 5/2
v=3/2-5/2
7/2- 99 388.21+0.10

99 390.39a0.20
99 388.19 Mc/sec

99 390.49 Mc/sec

9/2-
the common isotopes of ten alkali metal bromides and ,Ii = 1/2 —3/2
iodides. The important points about each molecule

7/2-
'F=9/2-11/2 132 510.64+0.10 132 510.65 Mc/sec
are summarized in succeeding paragraphs, and tables F = 7/2
are given for each molecule showing all measured, J=3—4'
derived, and extrapolated constants, as well as com- F = 5/2 132 511.83&0.20 132 511.73 Mc/sec
P=3/2-5/2
"K. S. Kunz, Xznnerica/ Analysis (McGraw. -Hill Book Com- 1=5-6 not resolved 198 735.33+0.20 198 735.34 Mc/sec
pany, Inc. , New York, 1957), pp. 226 — 229.
's J. Topping, I&'rrors of Observation and their Treatment (Institute ts Extrapolated from other constants, o Theoretical values,
of Physics, London, 1955), pp. 99 112.— b Honig eI, al. , see referent:e 5,
822 J. R. RUSK AND W. GORDY
TABLE II. Lithium bromide (Li'Br"). by more than twice the sum of the estimated error
from the value obtained by Honig et ul. The reason for
Present results Previous results this deviation is not known.
Yp1 = 16 616.780+0.07 16 617.002&0.05 Mc/sec LAhilm IOChde. In lithium iodide the I"' nucleus gave
8,'= = 16 616.622~0. 13 16 617.617+0.05b Mc/sec rise to an observable hyperfine structure of its millimeter
Bp = = 16 532.653~0.02 Mc/sec
Dp = = 64.8&1.7 kc/sec wave rotational spectra, but again effects of the lithium
cv, = —YII = 168.58&0.08 Mc/sec nuclear interaction were not detectable. The value of
653 &40b kc/sec
eQq for I"' in LiI is
—198.15 Mc/sec. ' Because the
D, '= —Ypg = 64.8+1.7
egQ/h=
kc/sec
30.71+0.40b Mc/sec nuclear spin of iodine is ~, each J
level is split into six
Previous constants
components. With the selection rules for Ii, this gives

o), =
Derived constants
cm '
a possibility for 15 lines, although at the levels J
YIp= 561~15' studied only the AF =+1 lines are of significant
cv, x, = —Y2p = 3.86&0.1' cm '
I, = 30.4232+0.0011 amu A' strength.
"
&e 2.17042~0.00004 A Table III' gives the observed and theoretical
Transitions spectrum for 1=6 —
7, ELF=+1. These and the
&

v=0 Measured lines Calculated lines J=5~6 lines were measured. As an example of the
F=11/2-13/2 165 294. 10+0.20 same Mc/sec weaker transitions not measured, the strongest DF = 0
4 5 F=9/2-11/2
line in the J=6 — + 7 spectrum should be down by a
J =5-6 198 335.75+0.20 same Mc/sec
factor of 7 from the weakest ELF=+1 line, and the
' Theoretical strongest AF= — 1 line down by a factor of 300. The
a Extrapolated from other constants. estimates.
b Honig et at. , see reference 5. spectrum for AI'=1 was completely resolved, except
for the narrow separation of 0.25 Mc/sec, which was
barely visible.
in the 79 isotope; two more were studied in the 81 As for LiBr, only Bp and Dp were measured directly.
isotope (see Table II). Contrary to expectation, the The equilibrium constants 8, and D, were evaluated
quadrupole structure of the J=5 ~
6 transition was from them (see Table III) with the aid of the data by
not resolved but appeared as a broad line of medium
strength. This may be due to the presence of water TABLE III. Lithium iodide (Li7I"').
vapor in the beam which broadened the lines sufFiciently
to prevent resolution of the structure. The J=4~5 Present results Previous results
transitions were especially hard to observe because
Y„=13 286.262a0. 07 13 286.386~0.08 Mc/sec
both the resolvable hyperfine splitting and the decrease 8,'= = 13 286. 15~0.10 13 286.785&0.08 Mc/sec
in intensity with the cube of frequency make the lines Bp = = 13 225.066+0.011 Mc/sec
weak in comparison with the = 5 — J
+ 6 transition. Both Dp = = 43.41~0.13
o., = —YII =
kc/sec
122.62&0.10 Mc/sec
transitions were observed with fifth harmonic power 455+ 50b kc/sec
from the harmonic generator. D = —Yp2 = 43.41 ~0.25 kc/sec
eQII/h = —198.15+030b Mc/sec
No excited vibrational states were observed for
LiBr although the Boltzmann distribution indicates Derived constants Previous constants
that the population of v=1 states in LiBr should be o), = —490&14'
Y10 450d cm I
down only 56%%uo from the ground state at the tempera- ~,x, = —Y2p — 3.1&0.1' 1.5d cm '
ture required for vaporization. Possibly the effective
I, = 38.0494~0.0013 38.04799b amu A~
~e 2.39191+0.00004 2.3919&0.0001b A
temperature in the observed molecular beam was
Transitions Measured lines Calculated lines
considerably lower than the source temperature. This v=0 (Mc/sec) (Mc/sec)
effect was noticed mainly in the lithium salts. It may
be caused by collisions of the salt molecules with cooler
J= 5—6 F =13/2-15/2 158 664.68~0.20 158 664.66
F= 15/2-17/2 158 664. 12 158 664.06
water vapor molecules which evaporated from the salt F= 11/2-13/2 158 663.12 158 663. 15
and formed a cloud in the cell. There was evidence that x=5/2-7/2 158 662.00 158 661.91
P =9/2-11/ 2 158 661.28 158 661.31
the sample used contained water of crystallization. I' =7/2-9/2 158 660.44 158 660.57
Because no excited vibrational states were measured,
only 8 p and Dp could be obtained. The value of 8, was
J = 6—7 F = 15/2-17/2
Ii = 17//2-19/2
185 092.51&0.20 185 092.42
185 091.98 185 091.95
extrapolated from the values of Y~~ and Y2~, as measured P= 13/2-15/2 185 091.31 185 091.38
by Honig et al. , ' although there is some doubt about I' = 11/2-13/2, 7/2-9/2 185 090.17 185 090.21
their accuracy. The value of Dp was assumed to be F=9/2-11/2 185 089.47 185 089.50
approximately equal to that of D, . This assumption is
based on the fact that, in all the other molecules a Extrapolated from other constants. ' Theoretical estimates.
b Honig et al. , see reference 5. d Levi, see reference 19.
studied, the value of P, was never more than 1/500 as
large as that of D, . This assumption was also used in "H. Levi, dissertation, Berlin, 1934 (unpublished, quoted in
the derivation of ~, and co,x, . Our value of 8, deviates reference 5, p. 634).
 M ILL I M ETER BLAUE MOLECULAR BEAM SPECTROSCPOY

powder is very soluble in water. The spectra of LiI were

first observed at a temperature of 600'C.
Sodium Bromide. Sodium bromide is a good illus-
tration of a molecule that is best studied at either
V-0 rather high, or at very low, microwave frequencies. The
coupling of the nucleus of bromine with J, egq of +58
i~ Ir ~e

Pro. 3. The J=20~ 21 Mc/sec, produces just enough splitting at 24 kMc/sec

transition of NaBr ' at frequency for the lines with the hot cell to be very
187 kMc/sec (1.6-mm broad but not resolvable. ' At 100 kMc/sec the quad-
wavelength) in the ground
vibrational state and in rupole broadening is 180 kc/sec, enough to limit only
the first and second ex- slightly the accuracy of the measurements. At higher
cited vibrational states. frequencies this broadening becomes negligible.
The lines are about 100
kc/sec in width and are Examples of some lines measured are given in Fig. 3.
distorted by the amplifier There one sees a relative comparison of three vibra-
because of the rapid sweep tional transitions at approximately 1.6-mm wavelength
Used.
(187 kMc/sec). The lines are sharp, with no evidence of
the unresolved hyperfine structure.
I"
1 ~

~
8s
~
~
An evaluation of five s was made (see Tables IV
and V). All of them, even the vibration —rotation
~ ~
~ ~

interaction constants, were obtained much more

accurately than in the earlier work of Honig et al. The
H, term was found to be negligible in all of the measure-
ments. The stretching constant D„which had been
Honig et al. ' For LiI and in fact for all the molecules assumed to be about 7 kc/sec, ' was found to be 50% too
in this study except KbBr, the I'0& value is lower than high; this accounts for the value of Vo&, which is
that obtained by Honig et al. , but is within the limits larger than expected. The ratio of the mass isotopes of
of their standard errors. bromine is larger than expected in comparison with
From the line intensities it seems that LiBr, and to a the other determinations of that quantity, although
lesser extent LiI, contain a moderate percentage of within the error limits. The spectra were first observed
dimers in the beam. From work done here and other at a temperature of 740'C.
laboratories, " it appears that all the alkali halides Sodium Iodide. Sodium iodide is the only other
which are as light as, or lighter than, XaCl may contain
at least 50% dimers. TABLE IV. Sodium bromide (Na Br ).
The lithium halides were the most dificult of the
present group for experimental handling. Both of them Present results Previous results
are very deliquescent and release their water molecules = 4534.4603 &0.0032 4534.51+0.10' Mc/sec
"
at a temperature of about 50'C. Thus, these salts,
F'p1
8, = — = 4534.4658+0.0072 4534.52+0.10' Mc/sec
n, = Y11= 28.2091~0.0038 28.25+0.10' Mc/sec
which cannot be obtained in a purely anhydrous form, Y21
— 72.92+1.2 85~30' kc/sec
must be warmed very slowly so that all the water will D, = —F02 = 4.6574~0.0054 7+3' kc/sec
be pumped away. When the temperature suitable for P, = —Y12 = — 0.0149+0.0028 kc/sec
the beam has been reached, water may be released so Derived constants Previous constants
rapidly from some of the Lavite spacers and insulators = 298.49&0.17 315 cm '
that the vacuum deteriorates too quickly for obser- —Y1p
F2p = 1.16~0.03 1.15b cm ~
vation of transitions to be attempted. This condition 111.486+0.004 111.486~ amu A'
re 2.50201~0.00004 2.5020*0.0001' A
can be avoided with these two salts only if the Lavite
is washed free of all traces of salt after each run and is Measured lines Calculated lines
Transitions (Mc/sec) (Mc/sec)
baked in a very hot furnace for several hours so that
no water attracted by the salt and reabsorbed by the J== 11-12 108 456.85~0.10 108 456.84
J= 12-13 117 488.93 117 488.86
Lavite will remain. J= 15-16 144 575.72 144 575.78
These two salts, again, partly because of their J= 20-21
16-17 153 601.28 153 601.34
deliquescence, are the most dificult to remove from J 189 683.49 189 683.46
the interior walls of the 5-band wave guide. The drier J== 11-12 107 783.39+0.10 107 783.42
salts tend to Rake off when scraped, and the remaining J = 20-21
12-13 116 759.38 116 759.35
J 188 505.34 188 505.35
~ S. A. Ochs, R. E. Cote, and P. Kusch, J. Chem. Phys. 21, 459,
J == 20-21
2-13
'.
1. 116 033.62~0. 10 116 033.62
(1953).
"S.D. Hodgman, R. C. West, and S. M. Selby, Handbook of J 187 333.37 187 333.37
Chemistry arId Physics (Chemical Rubber Publishing Company,
Qevefand, Ohio, 1958), 39th ed. , p. 548, a Hqnig et at. , sye qefepgnce 5, b j,|;vi, gqe 1.e&epqnqq 19,
 J. R. RUSK AN D W. GORD Y

TAHI, E V. Sodium bromide (Na 33r ). molecule studied which has a significant amount of
quadrupole coupling at the J
values observed. It has
Present results Previous results the largest quadrupole coupling of any of the alkali
Fol =4509 1958~0 0055 4509.34+0.10' Mc/sec halides, with an I"'
eQq of —
259.87 Mc/sec. ' The
8, = = 4509.2009&0.012
Ae= F ll — 27.9688+0.0066
4509.35&0.10' Mc/sec
28.06+0.10' Mc/sec
J= 13 — + 14 transition, which is the lowest frequency

F2l = 71.37&1.6 observed, has a quadrupole broadening of 900 kc/sec

Pe
D, = F o2 = 4.5950~0.0070
Pe= F l2 = — 0.0087 ~ 0.0062
kc/sec
kc/sec
kc/sec
while at the J= ~
19 20 transition this broadening is
only 400 kc/sec.
Derived constants Previous constants Evaluation of five V's was made. The H, correction
Flp=298. 00&0.22 cm ' was unnecessary. Results are more accurate than
MeXe = —F2p = 1.20+0.05 cm ' previously available ones; D, and P. were measured for
I, = 112.111~0.004 amu A' the 6rst time. A summary of the results is given in
2.50201&0.00004 A
Table VI. Lines were first observed at a source tern-
Measured lines Calculated lines perature of about 660'C. This salt was deliquescent,
Transitions (Mc/sec) (Mc/sec)
and some of the precautions mentioned earlier had to
J == 11-12 107 853.84~0. 10 107 853.77 be taken.
J 12-13 116 835.58 116 835.62
Potassium Bromide. Fabricand et al. , ' using molecular
JJ== 16-17
—
152 747.53 152 747.58
19 20 179 662.30 179 662.27 beam electric resonance (MBER), have already
J= 20—21 188 629.57 188 629.58
made very accurate evaluations of Vp1, F11, and F21
J == 11-12 107 186.03 107 186.01 for KBr. Several millimeter wave transitions were
JJ= 12-13
—
116 112.12 116 112.22 nevertheless observed in both isotopic species so that
16 17 151 801.77 151 801.67
J =20—
= 19-20 178 549.47 178 549.51 Vpg and V» might be measured very carefully. The
J 21 187 461.19 187 461.20 centrifugal stretching constant D, was evaluated
J= 16-17 150 860.60 150 860.60 separately for each isotopic species, but an average
solution for P, was made for both isotopic species, with
a Honig et al. , see reference 5. the isotopic correction taken into account. For this
molecule and for CsI, evaluation of P, is more accurate
than it is for the other molecules. Tables VII and VIII
TABLE VI. Sodium iodide (Na ~I 7). give a summary of the results obtained. The derivation
of &u, x, from P, shows that a more accurate value of Per
Present results Previous results is possible now than before. Previously, the error of
Fol=3531 7232~0 0043 3531.759+0.035' Mc/sec
8, = = 3531.7187&0.0072 3531.778~0.040 Mc/sec TABLE VII. Potassium bromide (K3'Brr').
19.4198&0.0052 19.439~0.030' Mc/sec
Y2l = 42.9+ 1.6 46.9 &6.0' kc/sec
D = —Yp2= 2.9183~0.0036 kc/sec Present results Previous results
Pe = —Ylg = — 0.0014+0.0022 kc/sec Fpl= 2434.947&0.001' Mc/sec
Derived constants Previous constants Be= = 2434.945+ 0.002 2434.953~0.001' Mc/sec
Ae= F ll 12.136~0.001' Mc/sec
F'lp = 259.20+0.16 286b cm l Pe= F'2l = 23+1' kc/sec
= —F2() = 0.964~0.024 0.75b cm ' D, = —F02 = 1.33763~0.00034 kc/sec
Pe= —F l2 = —
COeSe
I, = 143.140~0,005 143.138~ amu A~ 0.00006&0.00021 kc/sec
2.71143&0.00004 2.7115&0.0001' A II, = Fp3 ——170b 10 ' cps
Measured lines Calculated lines Derived constants Previous constants
Transitions (Mc/sec) (Mc/sec) = F lp = 219.170&0.029 '
toe 230c,d cm
v=0 J== 13-14 98 584.60%0.20 98 584.65 ~,x, = —F 2p —— 0.758+ 0.005 0 7c,d Cm l
J= 15-16
14-15 105 621.23 105 621.33 I, = 203.4372 ' amu A2
J= 112 656.83 112 656.97 2.82075 &0.00005 2.8207 &0.0001' A
J 19-20 140 787.59%0.10 140 787.60
J== 24-25 175 918.87 175 918.87 Measured lines Calculated lines
J 26-27 189 959.62 189 959.59 Transitions (Mc/sec) (Mc/sec)

J == 13-14 J== 20-21 101 963.70&0.10 101 963.79

J 14-15
98 043.38+0.20 98 043.31 J 21-22 106 813.95 106 813.94
J== 24-25
105 041.57 105 041.33 J== 31-32 155 273.22 155 273.18
JJ=26—
15-16 112 038.42
174 952.22+0. 10
112 038.30
174 952.23
J 39-40 193 968.20 193 968.16
27 188 915.62 188 915.66 J== 20-21 101 455.71+0.10 101 455.60
J == 13-14 J= 31-32 154 499.55 154 499.59
J = 14-15
97 504.24~0. 20
104 463.92
97 504.37
104 463.90
J 39-40 193 001.15 193 001.25
J = 24-25 :1.73 989.84~0. 10 1. 73 989.91
J = 39-40 192 037.50&0.10 192 037.45
J 26-27 187 876.42 187 876.36
a Fabricand et al. , see reference 6. o Levi, see reference 19.
a Honig et al. , see reference 5. b Levi, see reference 19. b Theoretical estimate. d Barrow and Caunt, see reference 25.
the B, term was seven times the error of Vo~, mainly et al. and with y, and P, obtained from mass ratios, 8,
because of the uncertainty in co,x, . Potassium bromide and D, were found. An accurate evaluation of the
is the lightest molecule for which we detected effects Br"/Br" mass ratio was obtained with this molecule.
of the H, term. At 1.5 mm, effects of this term were The spectra were first detected at the temperature of
of the order of the estimated error of the line frequency 690'C. Tables X, XI, and XII give a summary of the
measurement. results.
Potassium IOChde. Potassium iodide has been studied Rubidium Iodhde. The spectra of both Rb"I and
in the first three vibrational states. Again five V's have Rb"I were examined. Although 8,
and D. were found
been found. The values for Vo~ and B were learned with & much more accurately than previously known, the n.
30 times the accuracy of the previous values. Our and y, values of Honig et al. ' could not be improved.
values for the vibrational constants o, , and y, are less This molecule would be a good one to use for a very
accurate than the previous ones by Honig et al. ' but accurate determination of the mass ratio of the rubidium
fall within the sum of the two estimated errors. Our isotopes if many more transitions were observed with
more accurate evaluation of D, shows that the value great care. The spectra were first observed at a tempera-
given by Honig et af is 30. % too large, but the correct ture of 690'C. Results are summarized in Tables XIII
value falls within their estimated error limits. They and XIV.
were able to measure n, and y, better than we because Cesium Bromide. The spectra of cesium bromide were
they observed rotational transitions in vibrational fully investigated. It was the only molecule for which
levels up to v = 7. Line broadening by nuclear quadrupole five vibrational states were observed. As a result, the
interaction in KBr is only 56 kc/sec at 3 mm wave-
length. The spectra of this molecule were first observed
at a temperature of about 740'C. Table IX gives a
CsBr" analysis gives the most accurate results of the
set of vibration-rotation interaction constants, o.
and P, . Hence, more information is obtained about the
„y„
summary of the results we obtained. potential function. The mass ratio of the bromine
Rubidium Bromide. Rubidium bromide with its isotopes, comparable in accuracy to that obtained from
combination of four common isotopic species has the RbBr, is in very close agreement with the RbBr value.
most complicated spectra of all the molecules studied. All five V's were obtained accurately for both isotopes.
By chance, the stretching constant obtained by Honig
et al. s and estimated to only 50% accuracy is within &%
TABLE IX. Potassium iodide (K89F~r).
of the correct value. Again, the five V's were found after
the H, correction was made. All the V's were obtained
Present results Previous results
more accurately for Rb85Br~' than were previously
known. The isotopic species Rb "Br"was solved for the F'pj = 1824.9786+0.0013 1825.006+0.030' Mc/sec
first time. Since only the ground vibrational state was
8, = — = 1824.9778+0.0014 1825.012+0.030' Mc/sec
,
ot.= F gj. = 8.0272+0.0015 8.0337+0.0018' Mc/sec
examined for Rb "Br ', only a 8 0 and a Do were measured F2g = 11.62%0.39 12.21~0.25' kc/sec
directly. By extrapolation, with o. , obtained from Honig D, = —F 02 — 0.77749+0.00031 1.03+0.3' kc/sec
P, = —F — 0.00013+0.00018
gg kc//sec
H, = Fpe= — 80b 10 6
cps

TABLE VIII. Potassium bromide (K3 Br '). Derived constants Previous constants
F gp —186.53&0.04 200'' ~ cm ~
—F2p = 0.574+0.006 0.5 cm ~
Present results Previous results 277.006%0.010 277.00945' amu A'
Fpi= 2415.075 &0.001' Mc/sec 3.04781&0.00005 3.0478+0.0001' A
= 2415.073&0.002 2415.089+0.007' Mc/sec Measured lines Calculated lines
11.987 +0.001' Mc/sec Transitions (Mc/sec) (Mc/sec)
Pe F2g= 22+1' kc/sec
D =—F'p2 = 1.31.593~0.00012 kc/sec J =26-27 98 271.00&0.10 98 271.06
p =—F = —0.00006~0.00021
g2 kc/sec J== 29-30
28-29 105 540.31 105 540.30
JI, = F03 = —170b 10 ' cps J =39-40 109 174.23 109 174.12
J 145 478.34 145 478.38
Derived constants Previous constants J =49-50 181 707.95 181 707.96
= 218.271~0.010
F']p cm
—F20 — 0.752~0.005 cm ' J ==28-29
26-27 97 838.77&0.10 97 838.84
209.323&0.007 amu A' J= 29-30 105 076.04 105 076.06
2.82075+0.00005 A J=39-40 108 693.88 108 693.87
J =49-50 144 838.07 144 838.02
Measured lines Calculated lines J 180 907.50 180 907.50
Transitions (Mc/sec) (Mc/sec)
J== 20-21 J == 26-27 97 407.94&0.10 97 407.87
J= 31-32
101 132.87&0.10 101 132.84 J=29-30
28-29 104 613.14 104 613.16
21-22
J= 39-40
105 943.73 105 943.68 J=49-50 108 214.97 108 215.01
J
154 009.07
192 389.96
154 009.08
192 389.97
J 180 109.35 180 109.63

a Honig et at. , see reference 5. 'Levi, see reference 19.

a Fabricand et al. , see reference 6. b Theoretical estimates. b Theoretical estimates. d Barrow and Caunt, see reference 25.
 R. RTJSK AN B K. GORDY
TABLE X. Rubidium bromide {Rb 5Br7 ). Cesium Iodide. Cesium iodide is the heaviest diatomic
molecule included in the present study, and for this
Present results Previous results reason its spectra have some interesting features. The
Yp1 = 1424.8523&0.0012 1424.8342+0.02' Mc/sec spectra were the most difficult for us to interpret. The
= 1424.8522~0.0016 1424.840%0.02' Mc/sec estimated value of D, was between 0.13 and 0.18
F11= 5.5760+0.0012 5.5782+0.0056' Mc/sec
Ve 6.83+0.32 7.9&1.1' kc/sec kc/sec, whereas the value was found to be 0.11 kc/sec,
D. = Yo2 = 0.44833&0.00016 0.45&0.20' kc/sec well outside the estimated error calculated by Honig
Pe= Y12 = 0.00000&0.00015 kc/sec et a/. ' Among the observed CsI lines, a line of slightly
H, = Yo2= —46b 10 'cps
greater intensity than most of the others was con-
Derived constants Previous constants sistently observed. This anomalous line was finally
F1p = 169.46~0.03 181~ 0 cm ' identified as an HsO transition at 183 310 Mc/sec,
co, x,= —F2p = 0.463~0.007 0 35' cm ' previously observed by King and Gordy. 7 Apparently
I, = 354.795~0.012 354.8006~ amu A'
~e= 2.94471+0.00005 2.9448+0.0001' A H20, tightly bound to the CsI molecules, is not released
until the CsI is vaporized. In most of the other mole-
Measured lines Calculated lines
Transitions (Mc/sec) (Mc/sec) cules, the water that sometimes becomes attached to the
J=49-50 141 982.40+0.10 141 982.35
salt, as in LiBr and LiI, may be released at a lower
J== 50-51 144 812.72 144 812.75 temperature than that required for vaporization of the
J 54-55 156 128.74 156 128.76 salt. Otherwise, it is not known why the water line
J=62—63 178 731.66 178 731.63
was not detected during the course of study of the
J= 68-69 195 655.53 195 655.56
JJ==50-51 144 245.34+0.10 144 245.39 TABLE XII. Rubidium bromide (Rb"Br»).
63—64 180 842. 11 180 842.07
J= 68-69 194 887.95 194 887.96
Present results Previous results
J =63 —64 180 131.87+0.10 180 131.83
J= 68-69 194 122.20 194 122.24
Bp=
Fog@= 1409.080&0.015
= 1406.3406~0.01
1409.0573 +0.02s Mc/sec
Mc/sec
8, = = 1409.080a0.015 Mc/sec
a Honig et al. , see reference 5. e Barrow and Caunt, see reference 25. 5.4744+0.0085s Mc/sec
Theoretical estimate.
D, '= Dp =
b
0.43877+0.0004 kc/sec
Derived constants
The spectra were first observed at a temperature of
about 650'C. Tables XV and XVI give the results for F1p = 168.46+0.07 cm
G)eXe= —F2p' — 0.448+0.007 cm '
cesium bromide. I, = 358.766+0.012 amu A2
2.94471&0.00005 A
TABLE XI. Rubidium bromide (Rb 5Br '). Measured lines Calculated lines
Transitions (Mc/sec) (Mc/sec)
Present results Previous results J =63—64 179 551.22+0. 10 179 551.23
Yo1= 1406 6192~0 003 1406.5944&0.02' Mc/sec
J=64—65 182 342.01 182 341.99
J3, = = 1406.6192&0.0043 Mc/sec
J = 68-69 193 498.03 193 498.03
n =— Y11 = 5.4667+0.0043 5.461+0.011' Mc/sec
F12= 5.71&1.2 kc/sec + Extrapolated from other constants. b Honig et at. , see reference 5.
D = —Yp2 = 0.43722&0.00044 kc/sec
P, = —F, = 0.00001+0.00012 kc/sec
H, = Yp, 44b 10 ' cps J=6~7 transition of LiI which occurs near this
Derived constants Previous constants frequency.
YIp= 168.32+0.06 cm '
Because very high J
transitions could be observed,
F2p = 0.459+0.006 cm j H. was measured most accurately for CsI. Although H,
359.394~0.012 amu A' is less than 10 ' cps for J=139 —
1.40, the H, term
&

2.94471&0.00005 A makes a contribution of 2.32 Mc/sec to the line fre-

Measured lines Calculated lines quency. For this heavy molecule, co, is only 119 cm '.
Transitions (Mc/sec) (Mc/sec) Thus, if the pure vibrational transition ~=0 — 1 is to
J= 50-51 142 964.40+0.10 142 964.42 be observed with microwave techniques, wavelengths of
J == 55-56 156 928.04 156 928.10 approximately 0.1 mm must be achieved. In the Duke
J =69-70
63-64 179 238.80 179 238.83
J 195 943.97 195 943.91 Microwave Laboratory a wavelength of 0.51 has
already been reached with harmonics from an 8-mm
J== 51-52
J= 69-70
63-64
145 190.94~0.10
178 540.49
145 190.92
178 540.54 klystron. "
With 3-mm klystrons, now available, the
J 195 180.19 195 180.16 pure vibrational spectra of this diatomic molecule may
eventually be reached.
v=2 J=51-52 144 624.81&0.10 144 624.75
J= 69-70 194 417.98 194 418.00
~
M. J. Cowan and W. Gordy, Bull. Am. Phys. Soc. 5, 241
a Honig et al. , see reference 5. b Theoretical estimate, (1960).
 MILLI METER WAVE MOLECULAR BEAM SPECTROSCOPY 827
TABI,E XIII. Rubidium iodide (Rb 'I"'). TABLE XV. Cesium bromide (Csu~Br79).

Present results Previous results Present results Previous results

Foi = 984.3066+0.0021 984.3137~0.012~ Mc/sec Ypy = 1081.33314+0.0020 1081.3392+0.02' Mc/sec
B.= = 984.3062~0.0024 9843166+0.012' Mc/sec B,= = 1081.3333~0.0022 1081.3429+0.02' Mc/sec
Ae= —Fyy = 3.28156+0.0017 3.2806&0.0012' Mc/sec ne = —Ygg = 3.72052+0.00040 3.7175+0.0029' Mc/sec
Pe Y = 3.53&0.25 2.98~0.2' kc//sec Y2g = 3.234+0.044 3.1+0.4' kc/sec
D, = —Fpm = 0.221239&0.0001 1 0.234+0.035' kc/sec D, = —Y — 0.251775w0. 000022 0.27&0.08' kc/sec
Pe= —Fy2 = 0.000158&0.00008
p2
kc/sec Pe = —Ygg = 0.000006~0.000011 kc/sec
H, = Fps= —17b 10 'cps H, = Fp, — 22b 10 6 cps
Present constants Previous constants Derived constants Previous constants

areSe = —
FIo= 138 511~0 035 147 ' cm '
cm '
= 149.503&0.007
Yyo 171 cm '
Y2p = 0.335 +0.006 0
230 COeSe— — = 0.3602 +0.0008
Y2p 0.30 cm '
I, = 513.590~0.018 513.5883~ amu A' I, = 467.506m 0.016 467.5039. amu A'
3.17684+0.00005 3.1769+0.0001' A 're= 3.07221+0.00005 3.0720a0.0001' A
Measured lines Calculated lines Measured lines Calculated lines
Transitions (Mc/sec) (Mc/sec) Transitions (Mc/sec) (Mc/sec)
J=79-80 156 773.00+0.10 156 773,07 J==71-72
53-54 116 424. 56~0.10 116 424. 57
J =98-99 193 708.04 193 708.05 J= 155 068.30 155 068.31
J =99-100 195 647.00 195 647.05 J 72-73 157 211.47 157 211.38
J =79-80 156 248.91&0.10 156 248.83
J=90-91 195 705.22 195 705.28
J =98-99 193 059.15 193 059.08 J == 53-54 116 023.35+0.10 116 023.45
J=99-100 194 991.56 194 991.52 J=91-92
72-73 156 669.12 156 669.11
JJ=99-100
= 79-80 155 725.77+0.10 1.55 725.71
J 197 155.57 197 155.53
194 337.34 194 337.38 J == 72-73 156 127.90+0.10 156 127.80
J = 79-80 155 203.66&0.10 155 203.73
J 83-84 179 507.76 179 507.80
J=83-84
= 73-74 157 707.97+0.10 157.707.80
&
b
Honig et al. , see reference 5. c Barrow and Caunt, see reference 25. J 178 885.78 178 885.81
Theoretical estimate.
J=72-73 155 048.02 155 048.00
The values obtained for I'o~ and for 8, in the present
& Honig et al. , see reference 5. e Barrow and Caunt, see reference 25.
study are outside the limit of the estimated error of b Theoretical estimate.
the data from Honig et al. ' This might be expected,
since an error in F02 of that data would also cause an
error in the value of Vo~. These spectra were erst
observed at a temperature of about 630'C. A summary
TABLE XVI. Cesium bromide (Cs'~Br+).

of results for cesium iodide appears in Table XVII.

Present results Previous results
TABLE XIV. Rubidium iodide (Rb'7I'"). Ypg = 1064.58192&0.0032 1064.5853+0.02' Mc/sec
B.= — — = 1064.5819+0.0042 Mc/sec
o.e = Yj I. 3.63152&0.0036 3.6313+0.0024'. Mc/sec
Present results Previous results F2g = 2.708+0.41 3.1&0.5'b kc/sec
D, = —Yp2= 0.244159+0.00076
Foi = 970.7523+0 0035 970.7601+0.012' Mc/sec kc/sec
Pe = — 0.000123+0.00046
B.= =970.7519+0.0040 Mc/sec
Yj.2
H, = Ypg= — 22b
kc/sec
ne = — Fy] = 3.2133+0.0037 3.2135&0.0030' Mc/sec 10 P cps
F2g = 3.24+1.2 kc/sec Derived constants Previous constants
D = —Fp2= 0.2151.6&0.00015
P e = —Yy2 = 0.00015+0.00008b
kc/sec
kc/sec
G)eXe = —
-
F~p—148.301+0.023 cm '
H, = Fp, 16 4b 10 'cps F2p = 0.363&0.035 cm ~
I, = 474.862+0.023 amu A'
Derived constants 3.07221&0.00005 A
O)e= Yyp = 137.56+0.05 cm Measured lines Calculated lines
COeÃe = 2o= 0.330~0.006b cm ' Transitions (Mc/sec) (Mc/sec)
Ie 520.761+0.018 amu A2
~c 3.17684~0.00005 A J =72-73 154 783.88a0. 10 154 783.94
J=73-74 156 893.66 156 893 64
Measured lines Calculated lines J==91-92
83-84 177 965.90 177 965.83
Transitions (Mc/sec) (Mc/sec) J 194 788.37 194 788.41
v=0 J= 80-81
J= 100-i01
156 543.86a0. 10 same
J=73 —74 156 356.70~0.10 156 356.77
194 879.52 same
J==92-93
84-85 179 453.83 179 453.70
v=1 J==80-81
80-81 156 024.03&0.10 same J 196 213.95 196 214.02
v=2 J 155 505.26+0.10 same
J=73-74 155 285.45~0. 10 155 285.45
& Honig et al. , see reference 5. & Extrapolated from other constants.
b Theoretical estimates. & Honig et al. , see reference 5. b Theoretical estimate.
 J. R. RUSK AND W. GORDY
TABLE XVII. Cesium iodide (Cs'"I"rl. TABLE XVIII. Bromine isotopes mass ratio.

Present results Previous results Ratio

Molecule Method Br"/Br" Error
Fp1 = 708.32920& 0.00089 708.3568+0.02~ Mc/sec
= 708.32904~0.00090 708.3579+0.02' Mc/sec NaBr Present method 0.9753019 87
—F11= 2.04638+0.00057 2.0441+0.002' Mc/sec RbBr Present method 0.9753104 66
'Ye = F2g = 1.482+0.082 1.45+0.4' i:c/sec CsBr Present method 0.9753132 74
D. = —Fp2 = 0.11 1330~0.000025 0. 152+0.025' kc/sec KBr MBER~ 0.9753088 20
p, = —F12 = 0.000074~0.000015 kc/sec RbBr Microwave absorption 0.9752999 65
H, = Fpg = — 7.51b 10 ' cps CsBr Microwave absorption 0.9753068 45
CsBr Mass spectroscopy' 0.9753075 13
Derived constants Previous constants
co, = FIp —119.195+0.013 120' ' cm ' a Molecular beam electric resonance; see reference 6.
coax& = —F2p = 0.2542&0.0021 0.2c cm ' See reference 5.
b

I, = 713.694%0.025 713.6686a amu A'

' See reference 24.
r8 3.31515+0.00006 3.3150+0.0001' A
Measured lines Calculated lines Because As=0 in our experiments, we do not obtain
Transitions (Mc/sec) (Mc/sec)
the vibrational energies directly but rather by an
J= 103-104 146 618.15&0.10 146 618.09 evaluation of co, and co,x, from the rotational constants
J= 110-111 156 412,03 156 412.02
J= 127—128 180 134.69 180 134.67 with the use of Eqs. (6) and (7). Since the centrifugal
J= 134-135
—
189 874.72
195 432.52
189 874.69
195 432.49
stretching constant D, is evaluated with good accuracy
J=138 139 from the high-J millimeter wave transitions (where the
J= 110-111 155 957.84+0.10 155 957.98 stretching effects become large), one can obtain the
J= 127—128 179 610.97 179 610.95 fundamental vibrational frequency ~, with good
J= 128-129 180 999.32 180 999.33
J= 138—139 194 863.62 194 863.63 accuracy from Eq. (6c). The much smaller, an-
harmonic constant co,x, is also evaluated with better
J= 110-111 155 504.71+0.10 155 504.61
accuracy than previously known. It is of interest that
J= 127—128 179 088.04 179 087.99
J= 138—139 194 295.66 194 295.60 the vibrational structure of these halides is now
evaluated indirectly in this way from millimeter wave
J= 127—
—128 178 565.72+0.10 178 565.78
rotational spectra much more accurately than it could
J=139 140 195 104.71 195 104.71
be obtained directly from optical spectroscopy. The "
a Honig et al. , see reference 5. & Barrow and Caunt, see reference 25. values obtained for ~, and ~,x, for the molecules studied
b Theoretical estimate. are listed in Tables I through XVII along with previous
values by others for comparison.
DERIVED RESULTS Muss Ratios. For the molecules with atoms having
Interrtuclear Distance. The equilibrium value of the more than one isotope, the mass ratios of any two
fundamental rotational constant B. can be evaluated isotopes (1) and (2) of a given atom can be obtained
with high accuracy from these measurements, From it, from the relative 8, values with the relation
the moment of inertia I,
and the equilibrium value r,
(B,(si/B (i)) —(tr (ii/ti(si)
of the internuclear distances can be obtained with the
relations of Eq. (6). The accuracy of the absolute values
= P(M+Ms)/MMs7/P(M+Mi)/MM, 7, (21)
of I,
and r, is limited, however, by Planck's constant h from which the mass ratios are
in which the possible error occurs in the fifth significant
figure. We have used the value h=6. 62517&0.00023 Mi/Ms —(M/Ms) (B,&"/B, i'&)/
erg sec given by DuMond and Cohen. The masses " (1+M/Ms —B i"/B &'~) (22)
employed are those tabulated by Johnson, Quisenberry,
and Xier. '4 It should be noted that is not measured 8, where M=Mi+Ms. The ratio M/Ms need not be
directly but is calculated from I'0& and the other known precisely for Mi/M& to be derived with accuracy.
constants which are evaluated from the observed Townes' shows that a fractional error 5' in M/Ms
frequencies with Eq. (11).The values for
are given in Tables I through XVII.
r, B„I„and gives a fractional error e' in Mi/Ms of
5'= (Ms —Mi) e'/(M+Ms). (23)
Vibrational States. In Eq. (3) the first two terms on
the right represent the vibrational energy Taking account of both the error in B,&"/B, ~" and in
w„/h= .(.+-', ) — .*.(.+-,')'. (2p)
M/Ms, Townes estimates the over-all fractional error
in obtaining masses from the 8, ratios of diatomic
~ J. W. M. DuMond and E. R. Cohen, in Handbook of Physics,
edited by E. U. Condon and H. Odishaw, (McGraw-Hill Book
Company, Inc. , New York, 1958), p. 7.169. "R. F. Barrow and A. D. Caunt, Proc. Roy. Soc. (London)
2 W. H.
Johnson, K. S. Quisenberry, and A. O. Nier, in Hand- A219, 120 (1953).
26 C. H. Townes and A. L.
book of Physics, edited by E. U. Condon and H. Odishaw (Mc- Schawlow, Microzane Spectroscopy
Graw-Hill Book Company, Inc. , New York, 1958), p. 9.55. (McGraw-Hill Book Company, Inc. , New York, 1955), pp. 15—18.
 M ILL I METER WAVE MOLECULAR BEAM SPECTROSCOP Y
TABLE XIX. Rubidium isotopes mass ratio. TABLE XX. Potential coe%cients.

Ratio Previous' Present Previous' Present Rittner'sb

Molecule Method Rbss/Rbsr Error Molecule ao Qp

RbI Present method 0.9770162 93 LiBr 1.02 1.42 —2.45 —2.71 —1.71
Rbcl MBER' 0.9770163 45 LiI 1.14 1.35 —2.56 —2.70 —1.75
RbF MBERa 0.9770148 52 NaBr 1.64 1.48 —3.16 —3.05 —3.09
RbI Microwave absorption 0.9770177 45 NaI 1.74 1.42 —3.23 —3.02 —3.32
RbBr Microwave absorption 0.9770146 55 KBl 1.63 1.48 —3.35 —3.24 —3.35
RbBr Mass spectroscopy' 0.9770191 22 KI 1.64 1.43 —3.41 —3.25 —3.50
RbBr 1.73 1.53 —3.49 3 33 —3.53
—3.49 —3.34 —3.55
4

RbI 1.64 1.44

a See reference 6. e See reference 24. CsBr 2.03 1.55 —3.72 —3.38 —4.03
» See reference 5,
CsI 1.53 1.50 —3.45 —3.43 —3.53
molecules as a See reference S. b See reference 27.

(M/Ms-1-1) (g&'-&/&s&")
) calculate these constants from the more directly
(M/M, +1)—go&/&s&'& l derived ones with the Dunham expressions,

ap p&, s/48, „— ar ——(n, p&, /68, ) 1.

— —(2'/)
Also, a~ and a3 can be derived from theoretical expres-
where 6 is the fractional error in mass M and 5 is the sions involving o. , and y„but
our values of ag and a3
fractional error in &s "&/p "&. are not thought to be more significant than the earlier
Tables XVIII and XIX give the mass ratios Br "/Br" ones.
and Rb"/Rb" as obtained from the present results. Table XX gives a comparisonof our values of ap and
They are believed to be slightly less accurate than the a~ with those of Honig et al. and with the Rittner
values presently known from other methods. With constant 3&. The present data show that the ap s ale
refinements in the measurements with the present much more nearly the same for the different molecules
molecular beam spectrometer, these ratios could prob- than was indicated by the earlier work. Even the ap's
ably be ascertained to a greater accuracy than the for the lithium halides are essentially the same as those
presently known ratios if that were the primary object for other halides. This uniformity in the ap values
of these experiments. indicates that the harmonic component in the potential
I'otee6a/ Fled, tioes. It is interesting to compare the function does not vary much from one alkali hahde to
potential constants-derived from our data with those another. The peculiarly high value of ap for CsBr
from the earlier results of Townes' group' and with found in the earlier work is not substantiated.
those calculated theoretically on the assumption of a
completely ionic model by Rittner. "
If Rittner's
Our values for a~ are not the same, within the experi-
mental error, for the different halides but show a smail
expression for the bond energy is expanded about the consistent decrease in value from the Cs halides to the
equilibrium point Lwith (= (r — r, )/r, j, it reduces to I.i halides. However, in agreement with the trend
the form, earlier found, this decrease is not as much as that of the
U=C+A&P(1+Ar&+AsP+ (25) Rittner 3& for the completely ionic model. The devia-
)
tions of our a~ values from the Rittner A~ is even wider
which, except for the constant C, has the form of the than that found earlier. This indicates that the com-
dominant term,
U= apP(1+a&&r+asts+ ), (26) TABLE XXI. Comparison of measured —F» with
theoretical Morse o,
in Dunham's potential function. Thus, the potential
constants ap, a~, etc. which are derivable from our
data are directly comparable to the Rittner constants, Molecule (Mc/sec) (Mc/sec)
2 p, c4 y etc. Since the latter are calculated with the LiBr" 169.09 162
assumption of an ionic model, these comparisons give LiI 122.62 119
NaBr'~ 28.209 24.4
a test of the validity of this model. Such comparisons NaI 19.420
were previously made by Honig et al. , ' but our values
17.9
KBr7' 12.136 11 1 ~

for the constants ap and ay of the most significant terms KI 8.027 7.40
Rb85Br79 5.576 5.09
in the Dunham potential should be more reliable than Rb"I 3.2816 3.07
those obtained earlier with less complete data. We CsBr" 3.7205 3.38
CsI 2.0464 1.92
"E. S. Rittner, J. Chem. Phys. 19, 1030 (1951).
830 J. R. RUSK AN D W. GORD Y

pletely ionic model is not valid for the lithium halides, is applicable. Table XXI compares the — I' ll and n,
i.e., their bonds have significant covalent character. values. The comparisons indicate that this compact
As earlier indicated, the Dunham — Yil reduces to the expression for the potential holds to a good approxi-
o, , value when the Morse" potential,
mation for all the molecules but that it does not provide
V(r) = VD/1 —e (" ")]' (28) an exact description for any of them.

PHYSI CAI REVIEW VOLUM E 127, QUM HER 3 AUGUST 1, 1962

Perturbation Energy CoefBcients and Ionization Potentials of the GroIInd State

of Three- to Ten-Electron Isoelectronic Atomic Series*
CHARLEs W. ScHERR, JEREMIAH N. SILVERMANp AND F. A. MATsEN
Departments of Physics and Chemistry, The University of Texas, Austin, Texas
(Received January 26, 1962)

The well-known perturbation expansion,

E„,(Ã)(g) =Z2 ~-0

g s (Ã)Z '

of the eigenvalues of the nonrelativistic Schrodinger equation for lV electrons about a nucleus of charge Z,
has been widely used in the past for the extrapolation and interpolation of atomic energies. The presence of
many small effects not explicitly taken into account by the perturbation expansion analysis reduce such
calculations to a process of empirical curve fitting of limited range and reliability. These small effects include
re]ativistic effects, the mass polarization, and the Lamb terms; to a good approximation, these effect can
also be expanded in a descending power series, but with a leading term containing Z . On the basis of three
plausible assumptions, theoretical approximations make it possible, in a semiempirical fashion, to remove a
major portion of these small effects from the experimental data. In this way accurate values for e2(~& and
good estimates for e3(~) have been obtained for 3 ~&E ~& 10. These coefficients have been used to disclose
inaccuracies and to fill gaps in the existing atomic energy data and to estimate electron affinities.

I. INTRODUCTION coefficients obtained by the analysis are, in themselves

LECTRONIC energies of high accuracy for atoms of theoretical interest.
~ can in theory be obtained by applying the varia- For light atoms, the total or experimental energy4
tion principle to a wave function constructed from a E(N)(z) is given to a good approximation by
large set of suitably chosen basis functions. High ac-
curacy has been obtained only for the He isoelectronic
g(N)(Z) g (N)(Z)+g ()v)(Z) (] )
series'; more recently, such calculations have been made where Z is the nuclear charge and X is the number of
at a somewhat lower level of accuracy for the Li- and electrons. The E„„(N)(Z)and E„(N)(Z) are the non-
Be-isoelectronic series. ' Calculations at an equivalent relativistic and the relativistic energies, respectively,
level for a larger number of electrons are not yet avail- which are defined and discussed separately below.
able. Further, experimental electronic energies of high
accuracy are also not available because, in general, only IL NONRELATIVISTIC ENERGY E,(N)(&)
the first few ionization potentials of a given isoelectronic The nonrelativistic energy E„„(N)(z) is the eigenvalue
series have been accurately determined. ' In this paper of the appropriate nonrelativistic Schrodinger equation.
a semiempirical scheme is developed, based on conven- A well-known result of conventional perturbation
tional perturbation theory, for the accurate extrapola- theory' is
tion of total electronic energies or of ionization po-
tentials as a function of the nuclear charge. The scheme
is used to disclose inaccuracies, to fill gaps in existing
(N)
(z) —zs P e, ()v)z—
1',

(2a)
2'=0
experimental data, and to obtain estimates of the
electron afFinities. The higher-order perturbation energy where e;&~& is the ith-order electronic energy perturba-
* This research was supported in part by the Robert A. Welch 4 Unless otherwise
indicated, the data throughout are in atomic
Foundation of Houston, Texas. units of energy, 2hcE~=pe'A ', where E~ is the Rydberg wave
t Present address: National Bureau of Standards, Washington number of the atom or ion in question and p is the appropriate
25, D. C. reduced electronic mass.
' C. L. Pekeris, Phys. Rev. 112, 1649 (1958); 115, 1216 (1959). ' See for example, H. A. Bethe and E. E. Salpeter, Handbuch
' A. W. Weiss, Phys. Rev. 122, 1826 (1961). der Physik, edited by S. Flugge (Springer-Verlag, Berlin, 1957),
' B. Edlen, J. Chem. Phys. 83, 98 (1960). Vol. 35, Part 1, p. 214 and p. 237 ff.