Beruflich Dokumente
Kultur Dokumente
Roux
B. E. Wood
Calspan Field Services, Inc., Arnold Air Force Station, Tennessee 37389
J.A. Roux*
Deportment of Mechanical Engineering, University of Mississippi, University, Mississippi38677
60
increments of 0.231 pm for interference maxima for 84 K and
3
0.238 Am for 20 K. The densities were 0.87 g/cm at 80.K and
3
0.76 g/cm at 20 K. Spectra obtained for NH 3 condensed at Z
a Ref. 8.
b Ref. 4.
Ref. 9.
d Ref. 5.
two bands are seen at 650 and 665 cm- 1 . These bands are .g
0
observed when the wave-number scale is expanded. These
bands are associated with the P2 13CO 2 and 12CO2 bands. The 20
strongest CO 2 band in the spectrum occurs at 2345 cm- 1 and
is the V3 12CO 2 band. The corresponding V3 band for the
carbon 13 isotope 13CO 2 occurs at 2285 cm-'. The combi- 0
nation V3 + lattice band is seen to occur at 2460 cm- 1 . The
I-
500 900 1300 1700 2100 2500 2900 3300 3700
Wave Numbers
final two bands of significance, but considerably weaker than
the V3 band, appear at 3600 and 3710 cm- 1 and are the 2v2 + Fig. 5. Transmittance of 3.88-tm-thick solid CO 2 on 20-K germa-
V3 and the P1 + P3 bands, respectively. The band at 3710 cm' nium.
is not shown in Fig. 4 but was observed in some sets of mea-
surements that were extended to 3950 cm-1. Although
transmittance data were taken for the 3700-3950-wave-
60
number range, the noise level past 3700 cm-' increased such
that accurate n's and k's could not be determined but spectral
features could still be easily seen. The transmittance spec-
trum of a CO2 film deposited on 20-K germanium is shown in ' 40
Fig. 5 for a film thickness of 3.88,um (the 15th laser-interfer- d
ence maximum for X = 0.6328 gin). The spectral-band lo-
cations agree with the values observed for the 80-K spectrum ' 20
in Fig. 4.
were all quite broad in contrast to those of the gases previously 1650 1595 1665 1665 1665 v2
- - 2245 2220 2220 P2 + PL
studied. These bands are summarized in Table 3. Observing
3300 3652 3255 3260 3300 i1
these bands in Fig. 6 makes it obvious that the band locations - 3756 - - - P3
are too broad to be designated by specific numbers, as in Table
3. Therefore, the numbers refer to the approximate centers a Ref. 4.
of the broadbands. The v1 band at 3260-3300 cm-' is ob- b Ref. 8.
724 J. Opt. Soc. Am./Vol. 72, No. 6/June 1982 B. E. Wood and J. A. Roux
, | ~ -
~ Ig
raees-Krsn
IlneariL.,
A.Sqares
ness of 0.25 ,m (first interference maximum) the transmission
205
varies for the PI band from 30% at 80 K to 32% at 50 K and fi-
nally to 34% at 20 K. Thus the depth of the absorption band
increases with increasing temperature, and the absorption
I L.5
s-~ I-
index also increases with temperature. For a film thickness I LO
cm-' for 20 K.) This trend agrees with warmup data for H20
films studied by Thompson et a.4 The libration band yL
observed in the solid at 825 cm-' at 80 K was observed at 800
cm-' for the 20-K films, a shift toward the opposite direction
from the vPband. The v2 band at 1665 cm-' was independent _1
of temperature, whereas the 2245-cm- 1 band at 80 K shifted I
ZI
OPTICAL CONSTANTS
In order to determine the complex refractive index (n = n - and NH 3). The n value appears to be defined primarily by
ik) of the thin solid film from the IR-transmittance mea- the period of the transmittance-versus-thickness curve at each
surements for wave numbers between 700 and 3500 cm-, an wave number. At small thicknesses, or high absorption, or
analytical model of film-plus-window transmission was used. low wave numbers, the transmittance-versus-thickness (for
The optical constants of the cryocontaminants were deter- each wave number) curve is not well defined; i.e., the period
mined by using this analytical transmission model' 2 -' 4 in of the interference as a function of thickness is not well de-
conjunction with a nonlinear least-squares convergence rou- fined. The k value, which is defined primarily by the mag-
tine. Also, the subtractive Kramers-Kronig relation between nitude of the transmittance, did not have this difficulty and
n and k was used in conjunction with the nonlinear least- was well defined over the whole spectral region (700 to 3700
squares determination of n. The subtractive Kramers- cm-'). Thus, to determine n, the k values were used with the
Kronig relation is given by subtractive Kramers-Kronig relationship; these new n values
were then used in the analytical model (along with the k val-
n(v) = nf(V) +2p vf2 k(v')v' - k(v)M ues) to see if good agreement occurred with the transmittance
data. For all wave numbers, the Kramers-Kronig n's along
with the least squares k's yielded good agreement when the
k (v') -k (vmn)vm dv', (3) analytical model and transmittance data were compared.
(V')2 - V"
Errors in the refractive index, thickness, and density are all
where 11mis a reference frequency, PI = 700 cm-', v2 = 3700 estimated to be within 42%. The error in the absorption
cm-', and P indicates the Cauchy principle value of the in- index varies from about +2% at high k values (-0.01 or
tegral. Integration was performed using the simple trape- greater) to a value that is inaccurate for values of k below the
zoidal rule; the k values used in Eq. (3) were those determined slanted dashed line at the bottom of the absorption-index
by the nonlinear least-squares technique. The values of k graphs. The value of k can vary over several orders of mag-
outside 700 cm-' < S < 3700 cm-' were considered either as nitude, and therefore the smaller the k value the larger the
zero or as a constant at the end-point values; both approaches error; this is because of the error limits on the film-substrate
showed a negligible effect on the values of n that were ob- transmittance measurements.
tained from Eq. (3). Generally the nonlinear least-squares results (thin-film
Transmittance data recorded for all deposits discussed here model) did converge, and a comparison of the Kramers-
were digitized every 2 cm-'. The optical properties were Kronig and thin-film results is shown for NH 3 (Fig. 7). As
computed every 10 cm-', except in regions of strong absorp- stated, the optical properties were obtained through use of an
tion, in which computations were performed every 2 cmi1. analytical model and the nonlinear least-squares method. It
The optical properties were initially computed by the non- was assumed that the germanium substrate (window) acted
linear least-squares determination. However, in some in- as a thick film, and hence there was no phase coherence among
stances, the program did not appear to converge upon a unique multiple internal reflected rays (inside the germanium). The
value of n. This usually occurred in regions of strong ab- refractive index of the germanium was a modification of that
sorption or low wave number or for cases in which it was pos- given by Herzberger and Salzberg"4 to account for the re-
sible to form only a few thin deposit thicknesses (20-K H 20 fractive-index change that is due to the cryogenic tempera-
B. E. Wood and J. A. Roux Vol. 72, No. 6/June 1982/J. Opt. Soc. Am. 725
The exact expressions' for Eq. (4) are algebraically long and
tedious but straightforward; these expressions involve Fres-
nel's equations for the interface reflectances and transmit-
tances. The values of n and k were obtained by knowing v,
0, and ng and measuring the value of T at known values of d
(for CO 2 at 80-K, 25 values of r were measured at 25 known
values of d). Mathematically it was required to have at least
two values of r at two known values of d in order to have two
equations for the two unknowns, n and k. By using the
nonlinear least-squares convergence method the values of n
and k that produce the least error (in the least-squares sense)
among all the data (25 values for 80-K C0 2) and the analytical
700 1000 1500 20D0 2500 -3000
model [Eq. (4)] were determined. When an initial guess was Waee Numbere
3500
used for the solution for n and k, the nonlinear least squares Fig. 9. Comparison of NH3 results at 80 K with those of Ref. 15 at
generated improved values for n and k; these were then 190 K.
resubstituted to find an even better solution. This procedure
was continued until convergence was obtained. Usually
convergence was obtained at each wave number after about 1. For both temperatures the agreement between the
three iterations. Details of this standard method are available Kramers-Kronig method and nonlinear least-squares tech-
in Ref. 1. nique was good. The slanted dashed line on the k-versus-v
plots represents the minimum imaginary refractive-index
NH3 Optical Constants values that can be determined because of the error limits on
The optical constants for NH 3 at 80 and 20 K were determined the film-substrate transmission measurements. Values below
by using both the nonlinear least-squares technique and the the dashed line are shown but may not be accurate, as is noted
Kramers-Kronig method. The results are shown in Figs. 7 on the figures. The anomalous dispersion at the two strong
and 8, respectively. The tabulated results are given in Ref. NH3 bands is clearly seen in the n-versus-v plots of Figs. 7 and
8. The n values are higher at 80 K than at 20 K, as was the
2 5
density. Also, the k values at the two major absorption bands
. 1
- Prent
Work
- - - - Thoson,Arnold,
Sanderson.
Marts.(300Rd. 4
are greater at 80 K than at 20 K.
The data given in Figs. 8 and 9 are a comparison of the re-
sults of this work with those of others. The results of Rob-
'5 ertson et al.1 5 (Fig. 9) were obtained by solidifying liquid
ammonia. The Lambert absorption coefficient was deter-
0.0
mined by using thicknesses in the range from 0.50 to 20 Aum.
as5 Interference effects were not mentioned for these small
thicknesses. The n values were then determined by using the
0 700 1000 15W0 2000 25W 5
3000 W
3500
substractive Kramers-Kronig method. Data of Ref. 15 were
WaneNumbers obtained at 190 K, whereas the present data were obtained at
80 K. However, the agreement for the n values is very good,
and the agreement for k is generally good. (Results were
limited to 950 cm-1 in Ref. 15 because of the use of CaF 2
windows.)
Shown in Fig. 8 is a comparison of the present data at 20 K
with the data of Thompson et al.4 at 30K. The data of Ref.
4 were obtained by employing basically the same approach as
that used in the present tests, except that thin-film reflection
data were recorded. In general the agreement is good, but the
results of the present work yielded higher values for both n
and k; the results of Ref. 4 were shown only to 2400 cm-'.
Results shown in Fig. 10 illustrate the comparison with the
WaneNumbers work of Pipes et al.5 Similar comparisons were made with
Fig. 8. Comparison of NH3 results at 20 K with those of Ref. 4 at 30 the results of Sill et al. 16 (not shown) for NH3 at 80 K. The
K. agreement among all three is excellent. The same techniques
726 J. Opt. Soc. Am./Vol. 72, No. 6/June 1982 B. E. Wood and J. A. Roux
2.
-
Present
----- -Pips,.
WIrk
Rour,
Smith,ScootM
i8K, Ret.5
nonlinear least-squares technique and the Kramers-Kronig
2.0
method were in good agreement. Figure 11 shows the effect
of temperature on the CO2 optical properties. As was true for
I
I 1.5 NH3 , the higher temperatures yielded higher values for n and
II
k as well as for the density.
L l
I Determination of n and k for CO 2 at 80 K was made by
J. r I using transmittance measurements at 25 thicknesses; at 20
K, 14 thicknesses were used. The agreement between theory
0 700 loc 1520 2000 2020 3to 3500
and data is shown in Fig. 12 for 80 K. After the n and k values
WaveNumbers
were determined, they were used in the analytical model to
generate the theoretical curves.
No n or k values in the wave-number range of interest were
found in existing studies with which the present work could
be compared. Possibly these n and k values are unique in the
literature.
H2 0 Optical Constants
The optical constants for H2 0 at 80, 50, and 20 K are shown
in Fig. 13. In some regions (i3300 cm-') the nonlinear least
W-e. Numbers
.2
' L!
I .t
Is
6.0
Thickness, lnn
Fig. 12. Comparison of theory and data for 80-K solid CO 2 for three
WaneNumbers
different wave numbers.
--- 80K
--- 0K
2.0 -20K
1.5
L5
c L
ez'1!NaIL thlkoeventIy,.
blmYss
10 0 Inacou
b7e1 nte 25. 3aI 35
Numbers
.Wave
were used both in the present work and by Pipes et al. the
same research cell was also used, except that the present work
employs a much better technique for determination of film
thickness.
and are shown in Fig. 11. Tabulated results for these con- Fig. 13. Effect of substrate temperature on H2 0 optical properties
stants for CO 2 at 80 and 20 K are available.' Results of the (20, 50, and 80 K).
B. E. Wood and J. A. Roux Vol. 72, No. 6/June 1982/J. Opt. Soc. Am. 727
SUMMARY
Experimentally determined transmittance measurements of
condensed gases on a cryogenically cooled germanium window
were made. Thin films of NH3 , C0 2 , and H2 0 were deposited
at 20 and 80 K. The infrared spectral transmittance was
studied over the 500-3700-cm-1 range (the 2.70-20.0-,um
wavelength range). The film thickness was accurately de-
termined by using a two-angle laser-interference technique.
WaveNumbers The densities of the films were determined by using quartz-
Fig. 14. Comparison of H 2 0 results at 20 K with those of Ref. 4 at crystal micdbalances (QCM's) and by calculating the re-
30 K. fractive index at 0.6328 Am to permit accurate thickness
measurement. Some fracture and shattering effects were
PreentWork
observed for the films condensed at 20 K. From the trans-
------- Bertie,
LabOW,
andWhalley(IWKt,
(Ref.10 mittance data, the optical properties n and k were determined
for the 700-3700-cm-1 range. These results are available in
I.l L.
graphical and tabular' form for a wave-number interval of 2
L A cm-1 . Additional n and k properties have been determined
D)LLI for CO, NO, CH 4 , HCd, N 2 0, 02, N2 , and Ar films.17
The work reported herein was conducted at the Arnold
Engineering Development Center (AEDC), U.S. Air Force
us
A
700 15WD
.
15W0
.
2DO s
..
3000 3500 Systems Command (AFSC) while the authors were with ARO,
WaveNumbers
Inc., a Sverdrup Corporation company, an operating con-
1'S
tractor for the AEDC.
"II
I"A * Formerly with ARO, Inc.
10-2 REFERENCES
1. J. A. Roux, B. E. Wood, and A. M. Smith, "IR optical properties
of thin H2 0, NH 3 , and CO2 cryofilms," AEDC-TR-79-57 (AD-
A074913, Septembei 1979).
10o3 NOTE:
k values
belnw
thisline
2. K. E. Tempelmeyer and D. W. Mills, Jr., "Refractive index of
mayte inaccurate. carbon dioxide cryodeposit," Appl. Opt. 39, 2968-2969 (1968).
3. A. M. Smith, K. E. Tempelmeyer, P. R. Muller, and B. E. Wood,
"Angular distribution of visible and near IR radiation reflected
70 15W 15W 25W 25X 35W 35W from CO2 cryodeposits," AIAA J. 7, 2274-2280 (1969).
WaveNumbers 4. S. B. Thompson; F. Arnold, R. B. Sanderson, and A. W. Monty,
Fig. 15. Comparison of H2 0 results at 80 K with those of Ref. 10 at "Optical properties of cryodeposits on low scatter mirrors," in
100 X. Thermophysics and Spacecraft Thermal Control, Vol. 35 in
Progress in Aeronautics and Astronautics (MIT Press-, Cam-
bridge, Mass., 1974), pp. 229-248.
squares did not converge on b unique value of n. Therefore 5. J. G. Pipes, J. A. Roux, A. M. Smith, and H. E. Scott, "Infrared
the k values from the least-squares technique were used with transmission of contaminated cryocooled optical windows," AIAA
J. 16, 984-990 (1978).
the Kramers-Kronig method to generate n values. Then the 6. H. Wolff, H. G. Rollar, and E. Wolff, "Infrared spectra and vapor
least-squares k values and Kramers-Kronig n values were pressure isotope effect of crystallized ammonia and its deuterium
used in the analytical model and compared with the data to derivatives," J. Chem. Phys. 55, 1373-1378 (1971).
ensure good agreement. All the k values in Fig. 13 were above 7. F. P. Reding and D. F. Hornig, "The vibrational spectra of mol-
ecules and codnplex ions in crystals. V. Ammonia and deutero-
the error line and thus should be considered accurate values. ammonia," J. Chem. Phys. 19, 594-601 (1951).
The n and k values were determined by using absolute 8. G. Herzberg, Infrared and Raman Spectra of Polyatomic Mol-
transmittance measurements at 25 thicknesses for 80 K; for ecules (Van Nostrand, Princeton, N.J., 1954).
728 J. Opt. Soc. Am./Vol. 72, No. 6/June 1982 B. E. Wood and J. A. Roux
9. W. E. Osberg and D. F. Hornig, "The vibrational spectra of mol- 14. M. Herzberger and C. D. Salzberg, "Refractive indices of infrared
ecules and complex ions and crystal. VI. Carbon dioxide," J. optical materials and color correction of infrared lenses," J. Opt.
Chem. Phys. 20, 1345-1357 (1952). Soc. Am. 52, 420-426 (1962).
10. J. E. Bertie, H. J. Labbe, and E. Whalley, "Absorptivity of ice I 15. C. W. Robertson, H. D. Downing, B. Curnutte, and D. Williams,
in the range 4000-30 cm-'," J. Chem. Phys. 50, 4501-4520 "Optical constants of solid ammonia in the infrared," J. Opt. Soc.
(1969). Am. 65, 432-435 (1975).
11. N. Ockman, "The infrared and raman spectra of ice," Adv. Phys. 16. G. Sill, U. Fink, and J. R. Ferraro, "Absorption coefficients of solid
7, 199-220 (1957). NH3 from 50 to 7000 cm-1 ," J. Opt. Soc. Am. 70, 724-739
12. A. Vasicek, Optics of Thin Films (Trans. H. Watney-Kaczer) (1980).
(Interscience, New York, 1960). 17. J. A. Roux, B. E. Wood, A. M. Smith, and R. R. Plyler, "Infrared
13. 0. S. Heavens, Optical Properties of Thin Films (Dover, New optical properties of thin CO, NO, CH 4 , HCl, N2 0, 02, N 2, Ar, and
York, 1965). air cryofilms," AEDC-TR-79-81 (AD-A088269, August 1980).