You are on page 1of 8

INFRAREDPHYSICS

&TECHNOLOGY
ELSEVIER

InfraredPhysics & Technology37 (1996) 539-546

Polarization-dependent angular reflectance of silicon


and germanium in the infrared
Z.M. Zhang, L.M. Hanssen, R.U. Datla
Radiometric Physics Division, National Institute of Standards and Technology, Gaithersburg, Maryland 20899, USA

Received 15 May 1995

Abstract

We have investigated the reflectance of crystalline silicon and germanium samples for infrared radiation polarized
parallel and perpendicular to the plane of incidence. A Fourier transform infrared (FT-IR) spectrometer was employed to
measure the specular reflectance at angles of incidence from 20 to 60 in the wavelength range between 2 Ixm and 25 la,m.
The measurements agree with the theoretical analysis to within the experimental uncertainty. At large angles of incidence,
the reflectance depends strongly on polarization. Since the beam in a spectrometer generally exhibits partial polarization, the
measured reflectance without a polarizer may differ from the average reflectance for the two polarization states. This work
demonstrates that silicon and germanium can be used as standard reference materials for specular reflectance in the
mid-infrared region.
Keywords: Infrared; Oblique incidence; Optical properties; Polarization; Reflectance

1. Introduction

Accurate knowledge of the infrared optical properties of materials is crucial l'or understanding the
physics of materials (e.g., the electronic and phonon
structures [1-4]) and for the control of materials
processing [5-7]. There exist a variety of spectrophotometers and accessories for measuring spectral optical properties of materials from the ultraviolet to the far-infrared [8-11]. Standard reference
materials for reflectance and transmittance have been
developed for the short wavelength regions (from
200 nm to 2.5 I~m) [10,11], but there has been a lack
of standards in both instrumentation and reference
materials in the mid- and far-infrared regions (beyond 2.5 Ixm). Even for the most common materials,
such as silicon, considerable disagreement exists be-

tween different theoretical models and between experimental data obtained for different samples [12].
Gupta and Varma recently published several papers on the development of specular reflectance standards in the wavelength region from 2 ixm to 25 Ixm
using a double beam grating spectrophotometer [ 1315]. Their measurement results agreed with theoretical predictions for Au and A1 samples for near
normal as well as for oblique incidence. Reasonable
results were also obtained for Si and Ge samples at
near normal incidence. However, at angles of incidence greater than 20; large differences were observed between the measured reflectance and that
calculated from the theory of reflectance using published optical constants of Si and Ge. The disagreement was found to be substantial, particularly at long
wavelengths and at large angles of incidence. For

1350-4495/96/$15.00 1996 Elsevier Science B.V. All rights reserved


SSDI 1350-4495(95)001 l 8-2

540

Z.M. Zhang et al./ lnfrared Physics & Technology 37 (1996) 539-546

example, for a both-surface-polished silicon wafer,


the measured reflectance at a wavelength of 20 Ixm
and an angle of incidence of 50 was -~ 0.05, in
contrast to the predicted value of = 0.46 [14,15].
This disagreement was attributed to crystallographic
disorder that affects the reflectance only at large
angles of incidence [14]. If this explanation were
correct, then important information about a material's
crystallinity could be obtained from infrared reflectance measurements. Furthermore, the existing
theory of the optical properties of materials would
have to be modified for oblique incidence, even for
isotropic materials such as Si and Ge.
For the sake of both fundamental understanding
and measurement methodology, we have investigated
the specular reflectance of Si and Ge wafers at
angles of incidence from 20 to 60 and for wavelengths from 2 p,m to 25 txm using a Fourier transform infrared (FT-IR) spectrometer. The reflectance
was measured for the electric field polarized both
parallel and perpendicular to the plane of incidence.
A brief review of the theoretical formulation is given
in Section 2. The samples and measurement techniques are described in Section 3. Comparisons between the experimental results and theoretical predictions are discussed in Section 4. Based on our experimental results, we present a possible alternative
explanation for the results of Gupta and Varma
[14,15].

2. Theory
When a plane wave propagates from the air
(medium 1) to a semi-infinite crystal (medium 2), the
ratio of the reflected electric field to the incident
electric field at the interface (i.e., the complex Fresnel's reflection coefficient) is given by [16]
N l cos 0 2 - N 2

cos

In this equation, p-polarization and s-polarization


refer to the electric field being parallel and perpendicular to the plane of incidence, respectively; 01 is
the angle of incidence; N l = 1 is the refractive index
of air; N 2 = n + i K is the complex refractive index
of the crystal; and 02 is a complex number determined by Snell's law, i.e.,
cos 0 2 = / / 1 -

( N I sin 01//N2) 2 .

(2)

The power reflectance at the interface is


(3)

Ri2 = r12r?2,

where * denotes the complex conjugate.


For a wafer with parallel surfaces, multiple reflections occur inside the material. When the thickness
of the wafer is much greater than the coherence
length of the incident radiation, the interference effect can be neglected. Then, the reflectance and
transmittance of the wafer are [ 16,17]
Ti2T2jRl2e -4~

R=RI2 +

1 - R22 e -

46

(4)

and
TI2T21e-26

T=

1 -- Ri22e- 4 8 '

(5)

where Tl2 = (1 + rl2)(l + rj*2), T21 = (1 -- rl2)(1 -rl*2), and 3 = 217" I m ( N 2 cos 02)d2/A, where A is
the wavelength in vacuum. In general, the refractive
index and Fresnel's coefficients are wavelength dependent. In the spectral regions where absorption is
negligibly small, i.e., K = 0, Eq. (4) reduces to
2RI2
R = 1 +Rl2

(6)

which was used in Ref. [14]. Eq. (4) takes into


consideration the effect of absorption in the wafer
and is used in this work.

01

N l cos 0 ~ + N 2 cos 0 l '


r12

for p-polarization
N I cos 0 I - N 2 cos 0 2
Nt cos 01 + N 2 c o s
for s-polarization.

02 '

(1)

3. Sample description and measurement technique


Single-crystal silicon and germanium wafers were
used in this study. The samples are 25 mm in

Z.M. Zhang et al. / lnfrared Physics & Technology 37 (1996) 539-546

diameter and several tenths of a millimeter thick.


Both surfaces of the samples were polished to eliminate surface scattering losses. The surface roughness
is of the order of 1 nm, which is three to four orders
of magnitude smaller than the wavelengths of interest. Therefore, the effect of surface roughness on the
transmittance and reflectance is negligibly small [18].
The measurement beam will be deviated if the two
surfaces of a wafer are not parallel to each other. For
Si and Ge, at a 60 angle of incidence, the angular
deviation between the beam reflected by the front
surface and that reflected by the back surface is
nearly 15 times the wedge angle. For the samples
used in this study, the thickness variation is less than
5 txm, corresponding to a wedge angle of less than
0.011 .
For sample thicknesses greater than 1 mm, the
absorption would be too strong in the absorption
bands and the beam would shift position after several
internal reflections. For sample thicknesses less than
100 ~ m to 1 ~Lm, distinct interference fringes would
appear in the spectral region of interest. Ultrathin
samples of less than 0.1 ~Lm would be ideal because
the interference effect could be eliminated [19], but it
is difficult to fabricate and handle such thin samples.
Hence, we have selected samples with thickness of
several tenths of a millimeter in this study. The
interference fringes were averaged out by using a
spectral resolution that is greater than the free spectral range (i.e., the spacing between interference
maxima).
A Biorad Digilab FTS-60A FT-IR spectrometer
was used for the reflectance and transmittance measurements 1. A SiC ceramic source and a pyroelectric detector were used for the wavelength range
from 2 ~Lm to 25 ~m. For the reflectance measurements, a modified commercial accessory was used.
The optical setup in the sample compartment (positioned between the interferometer and the detector)
is illustrated in Fig. 1. The sample holder and mirror

J Trade names and company products mentioned in the text or


identified in an illustration are used only to adequately specify the
experimental procedure and equipment used. in no case does such
identification imply recommendation or endorsement by the National Institute of Standards and Technology, nor does it imply
that the products are necessarily the best available for the purpose.

BE

541

!R

Fig. 1. Schematic of the optical setup for the reflectance measurement, where M1, M2, and M4 are plane mirrors and M3 is a
concave mirror.

M2 are mounted on a turntable. All the mirrors are


aligned individually for each angle. The beam bypasses mirror M2 in the case of 20 incidence. The
beam is focused at the sample by a parabolic mirror
in the spectrometer (not shown in Fig. 1) and refocused at the regular focal point by a concave mirror
M3. The cross-section of the beam at these two focal
points is approximately 3 mm in diameter when
using the smallest aperture (--. 1.5 mm diameter) in
the spectrometer. The beam spot on the sample is
elliptical. The actual angles of incidence are within a
range determined by the beam divergence. The range
of variation is ___9.6 for the standard beam geometry
( f / 3 optics), which can lead to large uncertainties at
large angles of incidence. The reflectance is a strong
function of the angle of incidence. To reduce the
uncertainties, an additional aperture (25 mm diameter) was placed between the sample compartment
and the interferometer, which decreased the beam
divergence to 6.2 from the optical axis. The angle of
incidence indicated by the dial on the commercial
accessory was found to have an error of as large as
3 for some settings. Therefore, a protractor was
used to measure the angle of incidence with an
uncertainty of 1 (expanded uncertainty with a coverage factor of two is used in this paper). This corresponds to an uncertainty of 0.01 in the reflectance
for large angles of incidence.
A 150 nm-thick gold film was deposited by thermal evaporation onto a silicon substrate. It was used

542

Z.M. Zhang et aL / Infrared Physics & Technology 37 (1996) 539-546

as the reference for the reflectance measurements.


The reflectance and absorptance of gold have been
investigated by many researchers [15,20-22] The
measured results generally agree with calculations
based on the Drude model, i.e., the reflectance is
greater than 0.98 for 2.5 ixm < h < 25 ixm and at
angles of incidence from 10 to 60 [15,20]. At 10.6
Izm and 60 incidence, the reflectance calculated
from the measured absorptance by Kimura and Ford
[22] is approximately 0.98 and 0.99 for the electric
field polarized parallel and perpendicular to the plane
of incidence, respectively. We assume the reflectance of the gold reference to be unity. The
maximum relative uncertainty caused by this assumption is + 2%.
The alignment repeatability was tested by interchanging the mirror and the sample many times. The
measurement was found to be repeatable to within
+_2%. The resolution used for reflectance measurements was 8 cm - l . The signal was reduced by the
apertures, the mirrors, and the polarizer by almost
two orders of magnitude. Nevertheless, the noise-tosignal ratio was less than 0.05% for 512 scans. The
total measurement time was six minutes, including
the computation time.
Considering the reflectance of the mirror, the
alignment repeatability, and the uncertainty in the
determination of the angle of incidence, we have
estimated the expanded uncertainty (with a coverage
factor of two) to be 0.02 for the reflectance measurements of Si and Ge for the largest angle of incidence.
This uncertainty could ultimately be decreased by
reducing the uncertainty in the determination of the
reflectance of the mirror and by improving the alignment accuracy.

4. Results and discussion

The transmittance of the Si and Ge samples was


measured to check the agreement with reported data
of the optical properties of these materials. The
measured tra_gsmittance (with a spectral resolution of
8 cm - t ) is shown in Figs. 2a and 2b. The transmittance values calculated from Eq. (5.) using the optical
constants of Si and Ge given in Ref. [20] and the
extracted extinction coefficient of the Si sample (discussed below) are also shown. The measured trans-

(a)
lU
0
Z

0.65

"

i[

0.55

............

0.50

..............

I I

Measured

;-! """" Calculated(2'

0.60 .............

!1

" Ca ,at

t-"-~

r .,s W o . . ) / t

i . . . . . . . . . .

T.......................

= ....................

T .........

'o:

-'-'--I

I-

Z
<
n,

0.45

............

. iiii
. . . . . i_
..... L .....

0.40
0.35

. . . .

!__ ...........

i ........

266 l~m thick~slllcon i '!

..

J ....

i , , , ,

l0

. . . . . . . .

15

20

25

WAVELENGTH, X ~m)
(b) 0.55
Measured

o,., 0.0
!-

-t o

0.451

ii
.

Z
J
0.40
<

.
....

P" 0.35

0.30
0

L ,,,t

i..., i _ . .
5

10

15

20

25

WAVELENGTH, X (p.m)
Fig. 2. Transmittance of Si and Ge wafers at near-normal incidence, where the calculated values are based on the optical
constants from Ref. [20] and those obtained in this work.

mittance of the Ge sample agrees with the calculated


values to within 0.01. The measured transmittance of
the Si sample is lower at long wavelengths than the
transmittance calculated using the optical constants
given in Ref. [20], suggesting the existence of a
broad electronic absorption. Transport measurements
indicated that the electrical resistivity of the Si sample is between 1 l-I cm and 2 f~ cm, whereas the
electrical resistivity of the samples used to obtain the
optical constants was greater than 10 fl cm [20].
We have investigated the optical constants of the
Si sample used in this study. The refractive index, n,
can be determined from the measured free spectral
range, F, and the thickness, d, using the formula:
n = 1/(2Fd).
To discern interference fringes, we
measured the transmittance of the Si sample with a

Z.M. Zhang et al./ Infrared Physics & Technology 37 (1996) 539-546

0.5 cm-1 resolution. The thickness measured with a


micrometer was 266 _+ 2.5 tzm. The refractive index
values obtained with this method agree well with the
handbook values (listed in Table 1). For example, at
h = 10 ixm and 20 Ixm, the calculated n is 3.416
(0.15% difference) and 3.404 (0.47% difference),
respectively. Thus, the handbook values of the refractive index were used in the specular reflectance
calculations. The extinction coefficient, K, was extracted from the measured transmittance at an 8
c m - 1 resolution using Eq. (5) for 7.5 Izm < A < 25
Ixm at 2.5 ixm intervals. The transmittance calculated using the extracted n and r is also shown in
Fig. 2a. The optical constants of silicon at 7.5 ixm <
h < 25 ixm are listed in Table 1. The extracted K
values of the Si sample were used in the reflectance
calculations.
The measured wavelength-dependent reflectance
of the Si and Ge samples is shown in Figs. 3 and 4,
respectively, for both p-polarization and s-polarization. The angle of incidence varies from 20 to 60 .
The calculated values are based on the published
optical constants [20] of Ge and the extracted optical
constants of Si, as discussed above. For wavelengths
between 4 Ixm and 25 I~m, the measured reflectance
agrees with the calculated values to within the measurement uncertainty. The discrepancy at short wavelengths is attributed to the reduction in the degree of
polarization of the polarizer. The polarizer is made
of aluminum lines with submicrometer spacing, on a
KRS-5 substrate. The polarization contrast ratio (i.e.,
the ratio of the maximum transmittance to the minimum transmittance for a perfectly polarized beam) is
greater than 300 at 10 ixm, but reduces to -- 9 at 2
Ixm. This indicates that approximately 10% of the

Table 1
Optical constants o f silicon
Wavelength
(p,m)

n (Ref. [20])

K (Ref. [20])

K (this w o r k )

7.5
10
12.5
15
17.5
20
22.5
25

3.423
3.421
3.421
3.421
3.421
3.420
3.420
3.420

2.49)< 10 -5
6 . 7 6 ) < 1 0 -5
1.77)< 10 - 4
1.29>( 10 - 4
4.08)< 10 - 4
2.86)< 10 - 4
1.34)< 10 - 4
9.15)< lO - 5

8.4)< 10 - 5
2.1)<10 -4
4.0)< I 0 - 4
5.0)< 1 0 - 4
9.0)< 1 0 - 4
1.0)< 10 - 3
1.1X 1 0 - 3
1.3>( 10 - 3

543

(a)
0.g

" '-pol~lrizati'o 'n

....

i so
Itl
O
Z
<
I0
uJ
,,J
nL
tU

il "
Jl

Calcu~edl

0.7

0.6

i SO

I0

15

20

n"

0.5

0.4
0

25

WAVELENGTH, ~, 0.m)
(b)

0.5
[
,.

I ....
! ....
p-pDlarizltioh

+
i

0.4 r~_

0
ZI_

o,.

....

Measured I

__=.o.

i
!

:
i

_
o--

L
-0

03

,,,,I

+.+

0.2

I .... + . + o + o.o2..-.

:oo+ o__

f
0.1

. . . . . . . .
0

_:.___....-2-

~ ....
10
WAVELENGTH,

' ....
15

" " " , ,


20
25

k (pm)

Fig. 3. Spectral reflectance o f the Si w a f e r for different polarization states, w h e r e the error b a r indicates m e a s u r e m e n t uncertainty.

radiation is in the orthogonal polarization state at


A = 2 ixm. At 60 incidence and 2 I~m wavelength,
the reflectance calculated from 90% s-polarization
and 10% p-polarization is 0.647, which compares
well with the measured value of 0.66 (see Figs. 3a
and 3b). The calculated values in Figs. 3 and 4
assume perfect polarization because the degree of
polarization of both the polarizer and the instrument
has yet to be calibrated accurately.
For 50 incidence, the measured reflectance values beyond 3 ixm, as shown in Figs. 3 and 4, are
higher than those calculated for s-polarization and
lower than those calculated for p-polarization. This is
attributed to the error in the determination of the
angle of incidence, indicating that the actual angle of
incidence is smaller than 50 . For the case of 60

544

Z.M. Zhang et al. / lnfrared Physics & Technology 37 (1996) 539-546

incidence, the actual angle of incidence could be


greater than 60 by the same argument. The reflectance versus angle of incidence at 5 i~m is plotted in Fig. 5 for s- and p-polarization as well as the
average. The reflectance for s-polarization increases
monotonically, while the reflectance for p-polarization decreases to zero at the Brewster angle and then
increases rapidly. It is difficult to accurately measure
reflectance at angles of incidence greater than 60
without a collimated beam.
The polarization state of the beam in the FT-IR
spectrometer was investigated by rotating the polarizer. The signal varied by ___14% depending on the
wavelength. The reflectance measurements made
without a polarizer were compared with the average
reflectance of the two polarization states. The maxi-

(a) if: 266' IX~thlck'sliicon. . . .


"

0.8

..... ~ = s ~

........

l...L._l/l

40

60

80

~ 0.6
~

0.4
0.2
0

20

ANGLE OF INCIDENCE, 0 (degree)


(b)

Z~

0.S

I~rr

'

,r

l.tm~ - - - _ge#5ma~nlumt+h l c ki

(a)
+ [ " .~O'.+,ZOilOn+

+ ....

I " r " ~SU,

"I :

"I-............. - ............. ~......................I


tu
0
Z
<
I-

0
Ill
.ILL
J

<,<,/~.

"

'- .......

+ +amulatml+
- :

,,o-~.._

t ....

_+_*:++ ............

0.6 l ' - - , ~ - - ~ - + g - , - - - ' ~ O + ~" L ~ -~- - - -

+,o0+

10

15

20

25

(b)
0.5
aJ
0
Z

~ ....

- - ~

,o +

0.4

m
0.3
-J
w
=

! ....

I . . . . . . . .

i -"

2 0*i

40+ +

s o~---ii

-.

i,

i ....
10

15

0.2 .....~---~ ..............+'*-.-?-.-?.....................


0.|

. . . . . . . . . . .

, , . . i ....
5

'

20

40

ANGLE

WAVELENGTH, ~, (~m)

....

20

~ 17.4

.........
0

0.2

+ %/

ae
0.5

,....

.......

i ++_.+_+ 4o.
i . ~ :

~ 0.6 ~
~

25

WAVELENGTH, ~. (gin)
Fig. 4. Spectral r e f l e c t a n c e o f the G e w a f e r for d i f f e r e n t polarization states, where the error bar indicates measurementuncertainty.

60
80
OF INCIDENCE, 0 (degree)

Fig. 5. Angular-dependent reflectance of Si and Ge at 5 p.m


wavelength.

mum deviation was approximately _+0.02 at 60


angle of incidence, which is consistent with the
apparent polarization state of the beam.
The reflectance measurement results are in good
agreement with the calculated values, in contrast to
the results obtained by Gupta and Varma [14,15]. As
discussed by Breidne and Maystre [23], the grating
efficiency of a monochromator depends on wavelength and differs greatly for the two polarization
states. Thus, polarization in the spectrophotometer
could have affected the measurements of Gupta and
Manna [14,15]. Spectroscopic techniques have been
widely used in the study of crystallographic disorders and crystalline defects in the high-photon-energy regions, such as UV spectroscopy, electron
microscopy, or X-ray diffraction [24-26]. Infrared

Z.M. Zhang et al. / lnfrared Physics & Technology 37 (1996)539-546

spectroscopy is not sensitive to crystalline defects


whose size is much less than the wavelength. This
does not exclude the possibility that the samples
used by Gupta and Varma [14,15] may have a large
wedge angle or may be inhomogeneous due to
large-scale defects.
Although FF-IR spectrometers are widely used to
measure the optical properties of materials, their
radiometric accuracy is difficult to characterize. One
method is to calibrate the FF-IR measurements with
laser measurements at specific wavelengths. For
both-surface-polished wafers, the FT-IR measurements cannot be compared with laser measurements
because of the interference effect [17,19]. This problem can be overcome by eliminating multiple reflections. For samples with only one polished surface,
the radiation scattered by the second surface can
reach the detector and cause a measurement error.
This is particularly true for FF-IR spectrometers
where the collection angle is large. One alternative is
to use wedged samples. For a Ge sample with a 2
wedge, the beam reflected off the back surface will
deviate from the beam reflected off the front surface
by 16 at normal incidence and 3t at 60 incidence.
Another advantage of using a wedged wafer is that
the absorption due to the wafer becomes negligible.
Since the K value is much less than the n value even
for the strongest absorption band of Si and Ge in the
mid-infrared region, the effect of K on the reflectance is negligibly small.

5. Conclusions
We have investigated the specular reflectance of
both-surface-polished Si and Ge samples for wavelengths from 2 Ixm to 25 txm and at angles of
incidence between 20 and 60 . The experimental
results agree with the theoretical calculations to
within the expanded uncertainty of _+0.02 in reflectance for both polarization states. It can be concluded that Si and Ge are suitable candidates for use
as reflectance standard reference materials in the
infrared. However, the samples must be carefully
selected and well characterized, and the effect of
instrument polarization must be taken into consideration for oblique incidence. Further improvement of
the experimental setup will facilitate the develop-

545

ment of specular reflectance standards in the wavelength region from 2 ~ m to 25 p~m. Wedged samples should be investigated as possible specular reflectance standards.

Acknowledgements
The authors thank Dr. D. Gupta of the National
Physical Laboratory, India, for discussing his experiments with us.

References
[I] F. Wooten, Optical Properties of Solids (Academic Press,
New York, 1972) pp. 1-172.
[2] Z.M. Zhang, B.I. Choi, T.A. Le, M.I. Flik, M.P. Siegal and
J.M. Phillips, J. Heat Transfer 114 (1992) 644.
[3] M.I. Flik, Z.M. Zhang, K.E. Goodson, M.P. Siegal and J.M.
Phillips, Phys. Rev. B 46 (1992) 5606.
[4] Z.M. Zhang, B.I. Choi, M.1. Flik and A.C. Anderson,J. Opt.
Soc. Am. B 11 (1994) 2252.
[5] C.H. Chen, R.C. Phillips and P.W. Morrison, Thin Solid
Films 218 (1992) 291.
[6] B.I. Choi, A.C. Anderson,A.C. Westerheim and M.I. Flik, J.
Vac. Sci. Technol. A 11 (1993) 3020.
[7] Z.H. Zhou, B.I. Choi, M.I. Flik, S. Fan and R. Reif, J. Appl.
Phys. 76 (1994) 2448.
[8] R.M.A. Azzam and N.M. Bashara, Ellipsometry and Polarized Light (North-Holland Publ., Amsterdam, 1977) pp.
364-416.
[9] P.R. Griffithsand J.A. de Haseth, Fourier Transform Infrared
Spectrometry (John Wiley & Sons, New York, 1986) pp.
1-55, 166-219.
[10] K.L. Eckerle, J.J. Hsia, K.D. Mielenz and V.R. Weidner,
NBS MeasurementServices: Regular Spectral Transmittance,
NBS SP250-6 (US Department of Commerce, National Bureau of Standards, July 1987).
[11] V.R. Weidner and JJ. Hsia, NBS Measurement Services:
Spectral Reflectance, NBS SP250-8 (US Department of
Commerce, NationalBureau of Standards, July 1987).
[12] J.A.A. Engelbrecht,Infrared Phys. Technol. 35 (1994) 701.
[13] D. Gupta and S.P. Varma, Appl. Opt. 29 (t990) 1872.
[14] D. Gupta and S.P. Varma, lnfrared Phys. 34 (1993) 55.
[15] D. Gupta and S.P. Varma, Metrologia 30 (1993) 433.
[16] M. Born and E. Wolf, Principles of Optics, 6th edn. (Pergamon Press, Oxford, UK, 1980) pp. 38-47, 627-632.
[17] Z.M. Zhang, in: Heat Transfer 1994--Proceedings of the
Tenth InternationalHeat Transfer Conference, ed. G.F. Hewitt, Vol. 2 (Institutionof Chemical Engineers, Rugby, UK,
1994) pp. 177-182.
[18] B. Hapke, Theory of Reflectance and Emittance Spec-

546

Z.M. Zhang et al. / lnfrared Physics & Technology 37 (1996) 539-546

troscopy (Cambridge University Press, Cambridge, UK, 1993)


pp. 61-62.
[19] A. Frenkel and Z.M. Zhang, Opt. Lett. 19 (1994) 1495.
[20] E.D. Palik, ed., Handbook of Optical Constants of Solids
(Academic Press, Orlando, USA, 1985), pp. 286-295, 465478, 547-569.
[21] M.A. Ordal, L.L. Long, R.J. Bell, S.E. Bell, R.R. Bell, R.W.
Alexander, Jr. and C.A. Ward, Appl. Opt. 22 (1983) 1099.

[22]
[23]
[24]
[25]

W.D. Kimura and D.H. Ford, Appl. Opt. 25 (1986) 3740.


M. Breidne and D. Maystre, Appl. Opt. 19 (1980) 1812.
P.J. Zanzucchi and M.T. Duffy, Appl. Opt. 17 (1978) 3477.
G.D. Robertson, Jr., R. Baron, P.K. Vasudev and O.J. Marsh,
.I. Crystal Growth 68 (1984) 691.
[26] E. Mizera edited, Defects in Crystals--Proceedings of the
7th International School (World Scientific Publ., Singapore,
1987) pp. 161-270, 514-517.

You might also like