JOURNAL OF THE OPTICAL SOCIETY OF AMERICA

VOLUME 43, NUMBER 10

OCTOBER,

19-53

Absorption Coefficient of Ozone in the Ultraviolet and Visible Regions

EDWARDC. Y. INN ANDYosHio TANAKA
Geophysics Research Directorate, Air Force CamnbridgeResearch Center, Air Research and Development Command,
Cambridge, Massachusetts
(Received June 16, 1953)
As an extension to previous measurements on absorption coefficients-of ozone in the vacuum ultraviolet,
similar measurements have been made in the near ultraviolet from the overlapping region at 2000A up to
about 3500A and in the visible from about 4000-7500A. The absorption coefficient measurements were made
using ozone gas obtained by direct vaporization from liquid ozone of high purity (at least 90-95 percent) and
the optical densities measured with a Cary recording spectrophotometer with a 10-cm quartz absorption dell.
The method is essentially different from those used by previous investigators, e.g., photoelectric detectors
instead of photographic, pure ozone instead of ozonized oxygen, and the absorption cell located at the exit
slit instead of between the light source and entrance slit. The absorption coefficients obtained are somewhat
consistently lower than those previously reported by Ny and Choong, hitherto considered the most reliable.

INTRODUCTION

jection with regard to impurities as compared to

CONSIDERABLE amount of measurements on ozonized oxygen. The presence of polymers of oxygen,2
absorption coefficients of ozone in the ultra- e.g., 04, which has been detected in ozonized oxygen,
violet'-6 and some in the visible7 -9 regions has been re- may effectively contribute to the ultraviolet absorption
ported by a number of investigators, particularly in of the gaseous mixture. We have therefore used the
connection with problems in atmospheric physics. gas vaporized from liquid ozone of high purity for all
However, there appears to be significant discrepancy our measurements. The method used for the preparaamong the reported results in the region 2000-3000A tion of pure liquid ozone is briefly as follows. Tank
oxygen was passed through a train of CaC12 and liquid
(for example see Fig. 2 in Vassy's 8 paper). The results
of Ny and Choong' appear to be the most reliable in N2 traps and thence into an ozonizer. The ozonized
this region and have been used by others8",Oas a basis oxygen was then allowed to pass into another liquid N 2
for measurement of ozone concentration. In all these trap, thus condensing ozone in this trap. Removal of
measurements ozonized oxygen was invariably used so any condensed oxygen and other impurities was
that the absorbing gas was a mixture of oxygen and effected by continuous pumping on the liquid ozone
ozone, the concentration of the latter being determined followed by several fractional distillations, and a small
chemically. Furthermore, the absorption intensity was amount of pure liquid ozone was thus prepared. The
measured by photographic photometry, with the ex- liquid ozone was then transferred to a gas filling
ception of Lauchli and Meyer who used photocell vacuum system which was in communication with the
quartz absorption cell. Repeated chemical analyses on
detectors.
Recently, we have reported absorption coefficient a number of different preparations (KI and Na 2S20 3),
measurements of ozone in the vacuum ultraviolet the samples for analysis being obtained by vaporization
region." We have since extended these measurements of the liquid ozone, indicated a purity of at least 90-95
toward the longer wavelength region from about 2000- percent, the average being 92 percent. Correction for
7500A, the results of which will be described below. this purity was applied to all subsequent measurements.
The method we have adopted is very similar to that It should be stated that impurities (02, CO, C0 2 ) in
used in the vacuum ultraviolet measurements, which the sample may well arise from decomposition at the
is somewhat different from those used by previous walls of the apparatus and also from the reaction of
ozone with stopcock grease. However, it is felt that
investigators mentioned above.
these impurities did not in any appreciable way limit
EXPERIMENTAL
the absorption measurements. Furthermore, any oxides
Although the preparation and use of pure ozone is of nitrogen, which may have been formed during the
extremely hazardous, it is certainly less open to ob- ozonization process, were apparently effectively removed, since such impurities were not detected when
1 E. Meyer, Ann. Physik 12, 849 (1903).
samples of the purified ozone were analyzed with a
2 F. Kruger and M. Moeller, Physik. Z. 13, 729 (1912).
mass spectrometer (limit of detection about 1 part in
3 C. Fabry and H. Buisson, J. phys. et radium 3, 196 (1913).

A. Lauchli, Helv. Phys. Acta 1, 208 (1928).

Ny Tsi-Ze and Choong Shin-Piaw, Chin. J. Phys. 1, 38 (1933).
6 E. Vigroux, Compt. rend. 234, 2529 (1952).
7G.
Colange, J. phys. et radium 8, 254 (1927).
8 A. Vassy, Ann. phys. 16, 145 (1941).
9E. Vigroux, Compt. rend. 227, 272 (1948).
'OA. Tournaire-Vassy, Compt. rend. 204, 1413 (1937).
"1Tanaka, Inn, and Watanabe, J. Chem. Phys. (to be published).
4

2000).

A Cary recording spectrophotometer (quartz optics)

was used for the absorption measurements, the spectrum being automatically scanned and optical density
recorded simultaneously. Wavelengths on the recorder

870

12I. Grundland, Compt. rend. 236, 476 (1953).

Octbberl953 OZONE ABSORPTION

COEFFICIENT

IN UV AND VISIBLE

871
8

creased to a maximum, followed by a gradual decrease

with time, the decrease probably being due to thermal
decomposition of ozone. Evidently these observations
Zo
a,
that in a thoroughly evacuated cell ozone is
indicated
T0
rapidly adsorbed on the walls of the cell windows and
E
possibly through some unknown surface reaction the
!JIlSc
-transparency decreased through the cell windows. It
was obvious that these observations necessitated a
thorough flushing of the absorption cell prior to the
measurements in order that consistent results be obtained.
a
\
4
To correct for thermal decomposition of ozone during
a run, which ordinarily required a few minutes, a
'0
separate decay curve was taken (at about the same
WAVELENGTH (A)
ozone pressure and for a fixed wavelength). The decay
FIG. 1. Absorption coefficient of ozone in the region
curve was then used for correcting the optical densities.
2000-3000A.
The consistency of the results over the entire absorption
region, but measured in sections with different decay
traces could be read to the nearest rh 1A in the ultra- times, indicates the applicability of this method of corviolet region, while in the visible, to the nearest ESA. rection for thermal decomposition. In all cases the
A Nestor-type H2 lamp and a tungsten source were decomposition did not exceed about 12 percent during
used for the ultraviolet and visible regions, respectively. the measurements (5-10 minutes), the percentage
The absorption cell (10-cm quartz) was located between decomposition being greater at lower pressures than at
the exit slit and detector, this arrangement minimizing higher pressures. All measurements were made at
any photochemical action on the ozone sample. Other about 270 C.
investigators placed the absorption cell between the
light source and entrance slit, thus greatly increasing
RESULTS AND DISCUSSION
the probability of photochemical action on the ozoFrom the measured optical densities the absorption
nized oxygen. Although the slit width automatically
was calculated using the equation
coefficient
changes while the spectrum is scanned, it was adjusted
so that in the ultraviolet region it ranged from 0.01 to
I= Io 10-x,
several hundredths of a mm, and in the visible region,
a few hundredths to a few thousandths. Type P28 where lo and I are the incident and transmitted intensities, respectively, x the layer thickness reduced
photomultiplier tubes were used as detectors.
The absorption cell was filled with ozone gas at any to NTP and a the absorption coefficient in cm-l. The
desired pressure by vaporizing the required amount a values were calculated for over 200 different wavefrom the liquid ozone sample. The pressure was meas- lengths over the spectral region 2000-7500A, the a
ured with a U-type H 2SO4 manometer and the optical values for each wavelength being the average of
density measured simultaneously. It was necessary to several values from measurements at different ozone
make these measurements as rapidly as possible in pressures. The maximum deviation from the average a
order that no appreciable thermal decomposition take value amounted to about i45 percent.
place prior to the optical density measurements. The
pressures used ranged from about 0.3-650 mm Hg in
the ultraviolet region and for the weakly absorbing
visible region only the highest pressures were used.
Pressure measurements for the lowest pressures used
were made by a previously calibrated expansion method
and for the highest pressures a mercury manometer.
It was observed that although pressure equilibrium
-

.1

<a
0

IL
0

A
B-

20001

is obtained

XA. VASSY
n-TANAKAINN

as2,5,//

2100

INN JTANAKA
CHOONG
NY

220

WATANABE

2301

in a matter

of several seconds after ozone

is vaporized into the thoroughly evacuated gas filling

system and absorption cell, the optical density at a
fixed wavelength, after a rapid initial rise, slowly increased to a maximum value in a few minutes and then
decreased with time. However, with rapid evacuation
and refilling of the cell with ozone, this effect became

much less pronounced until with a few repetitions of

this procedure the optical density very rapidly in-

32W
WAVE

E 330
LENGTH

FIG. 2. Absorption coefficient of ozone in the region

3000-3500A.

872

E. C. Y. INN AND Y. TANAKA

For convenience, the absorption coefficient curve

has been divided into three regions: (1) 2000-3000A
(Hartley band); (2) 3000-3500A (Huggins band);
(3) 4000-7500A (Chappius band). The results for
these regions are shown in Figs. 1, 2, and 3, respectively.
(1) Region 2000-3000A
This is the region in which ozone is strongly absorbing. For comparison Ny and Choong's (B) results are
also plotted in Fig. 1. It can be seen that almost
throughout this region our results (A) are lower than
Ny and Choong's, the difference ranging from a few
percent to about 20 percent and, in general, somewhat
greater than can be explained by experimental error.
This discrepancy could be expained on the basis of
difference in technique, both in the preparation of the
ozone and method of measurement of optical density.
Thus, it may be due to an inherent difference between
photographic and photomultiplier detector methods.
It could also be due to the nature of the ozone sample
0.09-

C
cnO.05
f

I-

z
a:0.0!

z02

Vol. 43

It is interesting to note that the shape of the absorption continuum underlying these bands is more
symmetric in our results as compared with Ny and
Choong's. The oscillator strength or f value of this
continuum was calculated to be 0.088.
It should be mentioned that for the range in pressures used (change of about 25-fold or less) no pressure
effect' 3 was observed on the absorption

coefficient.

(2) Region 3000-3500A

In this region appear the well-known bands of
Huggins, as shown in Fig. 2. It is seen that the long
wavelength tail of the strong absorption continuum of
Region 1 underlies the Huggins bands and appears to
extend beyond 3500A. The curves in Fig. 2 were plotted
by evaluating the maxima and minima of all bands
with a few points chosen between bands.
In general our results agree in a qualitative way with
those reported by other investigators.5", 4 However, as
in the case of Region 1, Ny and Choong observed many
more bands. Furthermore, their a values for the maxima
and minima of the bands are again consistently higher
than ours, the difference being about the same amount
as in Region 1. The reasons for this difference are
probably the same as discussed above (Region 1).
Although our measurements were made at a higher
temperature (about 270C) than those under which
other investigators have made their measurements
(18C), the effect of temperature 4 on the absorption
coefficient cannot totally account for the discrepancy.
(3) Region 4000-7000A

swu

owu

am

WAVELENGTH4A)

b.V

low

7150

Fie. 3. Absorption coefficient of ozone in the region

4000-7500A.

used, i.e., ozonized oxygen as compared

to ozone of

Figure 3 is a plot of our results in the region of the

Chappius bands, and the curve represents the average
for two different measurements at about the same
ozone pressure (639.5 mm and 655.5 mm Hg). It was
not possible to measure the optical density at lower

high purity, in which ozonized oxygen may contain pressures using 10-cm cells since the absorption is exsmall amounts of polymers of oxygen, such as 04, the tremely weak in this region (max-=0.055 cm-l). Howlatter conceivably being strongly absorbing in this ever, the a values obtained from these two sets of
spectral region.'2
measurements agreed to 5 percent or better. Since the
In the region 2000-2200A our results agree very well bands are very broad the wavelengths of the maxima
with those reported previously from measurements in may have an uncertainty of somewhat more than t5A.
the vacuum ultraviolet." However, in this overlapping
Comparing our results with those previously reported
region Vassy's results8 (see Fig. 1), which appear to indicates a general agreement. However, quantitaagree with the extrapolation of Ny and Choong's tively our values are somewhat greater (about 10 percurve, are higher than ours. Since Vassy's measure- cent at the maximum) than Colange's7 and somewhat
ments depend on those of Ny and Choong, i.e., since smaller (about 20 percent) than those reported by
Vassy8 used the absorption coefficients obtained by the Vassy8 and Vigroux.9 Assuming a temperature effect
latter to determine the reduced pressure for calculating as reported by Vassy and Vassy,'5 it is much too small
the a values in the overlap region, it is not surprising to account for the observed difference. Since the bands
that Vassy's results are consistently higher than ours.
are very broad, difference in resolution could not exAlthough we have not observed as many bands as plain the discrepancy. It is possible that a systematic
Ny and Choong, due probably to lower resolution of error introduced by the method of indirect measurement
our spectrophotometer, the shapes and position of the
13B. Mahmoudian and E. Vassy, Compt. rend. 234, 2350 (1952).
observed bands correspond very well with those re14 D. Barbier
and D. Chalonge, Ann. phys. 17, 272 (1942).
ported by them.
15A. Vassy and E. Vassy, J. Chem. Phys. 16, 1163 (1948).

October1953OZONE ABSORPTION

IN UV AND VISIBLE

COEFFICIENT

of reduced pressure of ozone used by these investigators

could very well explain the discrepancy.
Underlying the broad bands in this region we note a
weak continuum starting from about 4000A and extending beyond 7500A. The f value calculated for this absorption continuum was about 2X 10-5, indicating a
highly forbidden transition.
CONCLUSIONS

It is particularly important in the study of atmospheric ozone, as well as atmospheric optics, that a
consistent set of absolute absorption coefficients of
ozone be available. It is therefore interesting to note
that, despite the difference in technique and method

JOURNAL OF THE OPTICAL SOCIETY OF AMERICA

873

used in our measurements as compared to the careful

measurements of Ny and Choong, there is generally
good agreement between our results and theirs. There
is, however, a consistent though not very large difference in results, and this can be attributed basically to
the different methods used. Since the difference in results is not very large, it is felt that the true absorption
coefficients lie somewhere between our values and those
of Ny and Choong.
ACKNOWLEDGMENT

We wish to acknowledge the help of Dr. L. E. Miller

in making the chemical analysis of the purified ozone
samples.

VOLUME 43, NUMBER 10

OCTOBER, 1953

An Intensity-Compensating System for Infrared Spectrophotometry

R. G.
H. M. RICHARDSON,*

ANDM. L. COFFMANt
FOWLER,
University of Oklahoma, Norman, Oklahoma
(Received April 6, 1953)

A novel compensating device for use in infrared spectrophotometry has been designed, constructed, and
its properties investigated. The device consists of two sector disks rotated synchronously, with phase control
on one disk permitting an absolute variation of the transmission factor of the compensator. No operational
tests on an actual spectrophotometer have been made.
AN ABSOLUTE COMPENSATING

INTRODUCTION

IN

present infrared spectrophotometers the compensating device used is usually a mechanical shutter
which cuts across the beam to be attenuated. One
shutter is a guillotine-like blade which is moved across
the beam by a gear and rack device. Another type is a
disk which has a radius that varies monotonically as a
function of the polar angle for some angular interval
equal to or less than 3600. This disk is mounted perpendicular to an axis parallel to the beam of radiation.
With this axis sufficiently near the beam, the beam may
be gradually eclipsed by rotating the disk about this
axis. These shutters are nonlinear in operation unless
the beam is of uniform cross section, i.e., the attenuation of the beam is not a linear function of the position
of the shutter. Perhaps the most successful compensator
is a comb-like shutter used in Wright and Herscherl as
well as the Baird, O'Bryan, Ogden, and Lee instruments.2 The comb shutter is slid across the beam with
the teeth of the comb attenuating the radiation. By
adjusting the number and shape of the teeth of the
shutter, the position vs beam attenuation relationship
may be made linear to within one percent.
* Now at Sandia Corporation, Albuquerque, New Mexico.
t Now at Texas Agricultural and Mechanical College, College
Station, Texas.
IN. Wright and L. W. Herscher, J. Opt. Soc. Am. 37, 211
(1947).
O'Bryan, Ogden, and Lee, J. Opt. Soc. Am. 37,754 (1947).
2Baird,

DEVICE

A different compensator device has been investigated

at the University of Oklahoma. Basically it is similar
in principle to the Keuffel and Esser photometer, described many years ago.3 Two sector diaphragms are cut
so that the two opposite 900 sectors are removed. The
two open sectors on each will transmit radiation which
is beamed in a direction parallel to the axis and at a
distance less than the radius of the solid sectors from
the axis. The solid sectors will block radiation in such
a beam. Thus, the rotating diaphragms will transmit
the radiation at regular intervals as the open sectors
sweep across the beam.
The average intensity of the transmitted beam is
determined by the intensity of the incident beam and
the relative position of the diaphragms. The two
diaphragms rotating with the same angular velocity
may be positioned, one with respect to the other, so
that the open sectors of the two diaphragms are adjacent. With this relative position the radiation in a
beam incident on the diaphragms will be transmitted
one-half of the time, or the average intensity of the
transmitted beam is one-half of the incident beam. It
is also possible to adjust the relative positions of the
diaphragms so that open sectors on one diaphragm are
adjacent to closed sectors on the other diaphragm.
I C. W. Keuffel, J. Opt. Soc. Am. and Rev. Sci. Instr. 11, 403
(1925).