PHOTOCHEMICAL PROCESSES IN THE UPPER ATMOSPHERE 
1/noH, is 10-* em? . For maximum absorption at about 
the height of the ozone layer (h = 3H, for example), 
kyumoH = e® = 20.1. These formulas for ho and mmax or 
max are very useful in the study of atmospheric dissocia- 
tion and ionization. 
If the absorption of the light quanta produces dis- 
sociation or ionization, this will partly exhaust the ab- 
sorbing gas, which will not be distributed exponentially 
at great heights, where the separation products will be- 
come predominant. (See § 22.) 
If the absorbing gas is not distributed exponentially 
(and the ozone layer is such a case) the above formulas 
will not apply to it, but even so they may give some 
help towards estimating the nature of the distribution 
of absorption. If the light absorbed is not monochro- 
matic, but includes radiation of widely differmg atomic 
or molecular absorption coefficients, the resultant dis- 
tribution of the rate of absorption will be a superposition 
of layers of the type discussed above, with differing 
values of Imax aNd gmax for each small interval of fre- 
quency. In the case of the regions of absorption by 
molecular oxygen or ozone, for which the values of k, 
vary at least 10° fold, much of the radiation will be 
filtered out at levels far above those to which the less 
absorbable frequencies penetrate. 
The light considered here is unidirectional, but a 
parallel beam of sunlight will be scattered as well as 
absorbed, and the scattered light, relatively more in- 
tense for the higher frequencies v, will itself be scattered 
and absorbed. The rate of absorption at any level will 
thus depend on contributions from light of all directions 
—a, complication to which little attention has yet been 
given in the field of atmospheric photochemistry. 
If we knew k, and [,, for each frequency, and the dis- 
tribution of each absorbing constituent, the distribu- 
tion and amount of the absorption would be calculable. 
In many cases, however, some of these data are un- 
known, especially J,, at frequencies for which the radia- 
tion is completely absorbed at high levels. But this gap 
in our knowledge is now becoming partly closed by 
information concerning the spectra of sunlight ob- 
tained at different heights in the atmosphere by rocket- 
borne instruments. 
12. Reversibility. De-excitation with Radiation. 
Atomic and molecular processes are reversible. The 
converse of excitation by radiation is de-excitation (a 
fall from a higher to a lower energy level) with emission 
of radiation. The fall need not, however, be made in 
one stage only; there may be two or more falls through 
intermediate states. The emission occurs spontaneously, 
and not at one definite interval after attaining the 
higher energy level. For each type of transition there 
is a constant 7’, called the half-life (or more loosely the 
lifetime), such that the probability of the transition 
occurring during any given short interval from time ¢ 
to time ¢ + dt, after attaining the higher level, is 
e—/7dt. The transition probability is briefly expressed 
as 1/T. 
There are certain selection rules concerning the per- 
missible changes of the various quantum numbers, from 
the initial to the final state of the particle; these deter- 
mine whether or not any particular transition is allowed 
267 
or forbidden. In the case of allowed transitions, T is 
generally very small (e.g., 10-8 sec). 
Among the various states of a particle there may be 
some from which there is no allowed transition to a 
lower level. Such states are called metastable states. 
Even in such cases, however, there may be possible 
transitions (not of the normal type to which the selec- 
tion rules apply) with finite though small probability, 
and such states consequently have a finite lifetime, much 
longer than that of ordinary excited states. 
Neutral atomic oxygen is such a case, of great in- 
terest for upper-atmospheric chemistry. Its terms 1D, 
and 18, of its first electronic configuration (see § 4) are 
both metastable. Their energy levels (or excitation en- 
ergies) are approximately 1.96 and 4.17 ev above the 
ground term *P. There is a finite probability 2.0 sec™ 
for the transition 4S) to 1D., involving a change of 
energy by 2.21 ev and giving rise to the light of the 
famous “green auroral line” 5577 A. The transition 
probabilities for fall from 1D». to the ground levels *Pe, 
3P,, and *Py are 0, 2.5 X 10-%, and 7.5 X 107 sec™, 
giving rise (in the last two cases) to the two “red 
auroral” lines of atomic oxygen, 6364 A and 6300 A, of 
which the latter, because of its threefold greater prob- 
ability, is three times the more intense. Thus the life- 
times of the 4S» and ‘D2 states are of the order 14 sec 
and 100 see respectively. 
The first excited states of neutral atomic sodium are 
not metastable; transitions occur with probability 0.62 
< 108 sec from 2P? and 2Py, to the ground state 2S, 
giving rise to the sodium yellow line 5893 A (really a 
close doublet or pair of lines, with wave lengths 5890 A 
and 5896 A, of which the former is twice as intense as 
the latter, because the number of atoms in the *Py state 
is twice that in the 2P: state). The lifetime of the ?P 
state is 1.6 X 10- sec. 
13. Excitation and De-excitation by Collision. Atomic 
and molecular particles can be excited by impact as 
well as by absorption of radiation, and the process is 
reversible, that is, such particles can be de-excited by 
collision. In the first case the kinetic (and possibly 
also other) energy of the impinging particle (which 
may be atomic, molecular, or an electron) is reduced by 
the energy required for excitation; in the second, it is 
increased by this amount. Such collisions are called 
collisions of the second kind, or inelastic or swperelastic, 
in contrast to the elastic collisions considered in the 
kinetic theory of gases. In these, the speed of separation 
of the particles after collision equals that of their ap- 
proach, so that the translatory kinetic energy of their 
motion relative to their mass-centre is unaltered. A col- 
lision between two uncharged unexcited particles is 
elastic if this kinetic energy is too small to raise either 
to its first excited state. For such elastic collisions the 
particles, though without definite boundaries, have a 
certain joint collisional cross section (which in the case 
of rigid elastic spheres would be 7h*, where R denotes 
the sum of the radii of the two particles). Similarly there 
is a collisional cross section for any given type of excita- 
tion, depending on the nature of the particles and on 
their relative speed of approach; it is one mode of ex- 
