Complex refractive-index measurement based on

Fresnel’s equations and the uses of

heterodyne interferometry

Ming-Horng Chiu, Ju-Yi Lee, and Der-Chin Su

The phase difference between s and p polarization of the light reflected from a material is used for
measuring the material’s complex refractive index. First, two phase differences that correspond to two
different incidence angles are measured by heterodyne interferometry. Then these two phase differ-
ences are substituted into Fresnel’s equations, and a set of simultaneous equations is obtained. Finally,
the equations are solved by use of a personal computer by a numerical analysis technique, and the
complex refractive index of the material can be estimated. © 1999 Optical Society of America
OCIS codes: 120.3180, 120.4530, 120.5050.

1. Introduction equations, and the complex refractive index is ob-

A complex refractive index N共n, k兲 is an important tained.
characteristic constant of thin-film materials, In this paper a simple method for measuring a
where n is the refractive index and k is the extinc- complex refractive index is proposed. A light beam
tion coefficient. There are several methods for of either s or p polarization reflected from a mate-
measuring the complex refractive index of a mate- rial with a complex refractive index has a phase
rial, e.g., R-versus-␪ 共reflectance versus incidence shift. The phase difference between these two
angle兲 methods1– 6 and ellipsometry.7,8 In those kinds of polarization is a function of n, k, and inci-
methods, the reflectances of s and p polarization at dence angle ␪. The phase differences that corre-
several different incidence angles or polarization spond to two different incidence angles are
conditions must be measured. Consequently, most measured with the heterodyne interferometry pro-
of them are related to the measurement of light- posed by Chiu et al.10 These data are substituted
intensity variations. However, the stability of the into Fresnel’s equations,11 and a set of simulta-
light source, the scattering light, the internal re- neous equations is obtained. Then these equations
flection, and other factors influence the accuracy of are solved by use of a personal computer by a nu-
the measurement results. Recently, Feke et al.9
merical analysis technique, and the complex refrac-
proposed a novel method to measure the complex
tive index of the material can be estimated. The
refractive index. The phases that correspond to
two orthogonal polarizations at different incidence technique has several merits, including a simple
angles are measured with a phase-shifting inter- optical setup, easy operation, high stability, high
ferometric technique. These data are substituted measurement accuracy, and rapid measurement.
into the special equations derived from Fresnel’s We demonstrate its feasibility.

M.-H. Chiu is with the Department of Electric Engineering, 2. Principles

Chien Hsin College of Technology & Commerce, 229, Chien Hsin
Road, Chung-Li, 320, Taiwan, China. J.-Y. Lee and D.-C. Su
共t7503@cc.nctu.edu.tw兲 are with the Institute of Electro-Optical A. Phase Difference Resulting from Reflection from an
Engineering, National Chiao Tung University, 1001, Ta-Hsueh Absorbing Material
Road, Hsin-Chu, 300, Taiwan, China.
Received 12 March 1999. A ray of light in air is incident at ␪ onto an absorb-
0003-6935兾99兾04047-06$15.00兾0 ing material with a complex refractive index N共n,
© 1999 Optical Society of America k兲, as shown in Fig. 1. According to Fresnel’s equa-

1 July 1999 兾 Vol. 38, No. 19 兾 APPLIED OPTICS 4047

 Fig. 1. Reflection at the surface of an absorbing material.

tions,11 the amplitude reflection coefficients of s and

p polarization can be expressed as
cos ␪ ⫺ 共u ⫹ iv兲
rs ⫽ ⫽ 兩rs兩exp共i␦s兲, (1)
cos ␪ ⫹ 共u ⫹ iv兲
N 2 cos ␪ ⫺ 共u ⫹ iv兲
rp ⫽ ⫽ 兩rp兩exp共i␦p兲, (2)
N 2 cos ␪ ⫹ 共u ⫹ iv兲
respectively, where
u2 ⫽ 1⁄2兵共n2 ⫺ k2 ⫺ sin2 ␪兲
⫹ 关共n2 ⫺ k2 ⫺ sin2 ␪兲2 ⫹ 4n2k2兴1兾2其, (3)
v2 ⫽ 1⁄2兵⫺共n2 ⫺ k2 ⫺ sin2 ␪兲
Fig. 2. Schematic diagram of measurement of the phase differ-
⫹ 关共n2 ⫺ k2 ⫺ sin2 ␪兲2 ⫹ 4n2k2兴1兾2其, (4) ences owing to reflection at 共a兲 an absorbing material and 共b兲 a
beam splitter. Other abbreviations are defined in text.
and ␦s and ␦p are the phase shifts of s and p polar-
izations and can be expressed as

␦s ⫽ tan⫺1 冉 2v cos ␪
u ⫹ v2 ⫺ cos2 ␪
2 , 冊 (5)
dence angles, ␪1 and ␪2. Hence a set of simultaneous
equations

冋 册
␾1 ⫽ ␾1共n, k, ␪1兲, (9)
2v cos ␪共n2 ⫺ k2 ⫺ 2u2兲
␦p ⫽ tan⫺1 , (6)
u2 ⫹ v2 ⫺ 共n2 ⫹ k2兲2 cos2 ␪ ␾2 ⫽ ␾2共n, k, ␪2兲 (10)
respectively. Hence the phase difference of s polar-
ization relative to p polarization is is obtained. If these simultaneous equations are
solved, the complex refractive index of the material

␾ ⫽ ␦s ⫺ ␦p ⫽ tan⫺1 冉ad ⫺ bc
ac ⫹ bd
, 冊 (7)
can be estimated.
B. Phase-Difference Measurements with Heterodyne
where Interferometry
Chiu et al.10 proposed a method for measuring the
a ⫽ 2v cos ␪, refractive index of a transparent material by using
b ⫽ u2 ⫹ v2 ⫺ cos2 ␪, total-internal-reflection heterodyne interferometry.
A schematic diagram of the optical arrangement of
c ⫽ 2v cos ␪共n2 ⫺ k2 ⫺ 2u2兲, our method, which is based on similar considerations,
d ⫽ u2 ⫹ v2 ⫺ 共n2 ⫹ k2兲2 cos2 ␪. (8) was designed and is shown in Fig. 2共a兲. Linearly
polarized light passing through an electro-optic mod-
From Eqs. 共7兲 and 共8兲 it is obvious that phase dif- ulator 共EO兲 is incident upon a beam splitter 共BS兲 and
ference ␾ is a function of n, k, and ␪, and ␾ can be divided into two parts, a reference beam and a test
experimentally measured for each given ␪. To eval- beam. The reference beam passes through analyzer
uate the values of n and k we require two phase ANr, and enters photodetector Dr. If the amplitude
differences ␾1 and ␾2 that correspond to two inci- of light detected by Dr is Er, then the intensity mea-

4048 APPLIED OPTICS 兾 Vol. 38, No. 19 兾 1 July 1999

 Table 1. Experimental Conditions and Measurement Resultsa

Material ␪1 共°兲 ␪2 共°兲 ␾1 共°兲 ␾2 共°兲 n k ⌬n ⌬k

60 80 144.96 74.15 2.007 3.781 0.01 0.006

Ni 70 80 121.32 74.15 2.006 3.782 0.01 0.006
60 80 134.22 65.22 0.27 3.430 0.05 0.004
Cu 70 80 108.25 65.22 0.31 3.428 0.07 0.005
a
Reference values from Ref. 14: N共Ni兲 ⫽ 1.98 ⫹ i3.74 and N共Cu兲 ⫽ 0.272 ⫹ i3.42.

sured by Dr is the Ir ⫽ 兩Er兩2. Here Ir is the reference we used a personal computer to record and analyze
signal. The test beam is incident at ␪ upon a test the data. The experimental conditions and results
material. Finally, the reflected light passes through are summarized in Table 1. The average experi-
analyzer ANt and is detected by another photodetec- mental values of the complex refractive indices of
tor, Dt. If the amplitude of the test beam detected nickel and copper are N共Ni兲 ⫽ 2.01 ⫹ i3.78 and N共Cu兲
by Dt is Et, then Dt measures the intensity It ⫽ 兩Et兩2, ⫽ 0.29 ⫹ i3.43, respectively.
and It is the test signal.
For convenience the ⫹z axis is in the propagation 4. Discussion
direction and the y axis is in the vertical direction. From Eqs. 共9兲 and 共10兲 we get
Let the incident light be linearly polarized at 45° with
respect to the x axis, the fast axis of EO under an
applied electric field be in the horizontal direction, 冏 冏
⳵␾2
⳵k
兩⌬␾1兩 ⫹
⳵␾1
⳵k
冏 冏
兩⌬␾2兩

冏 冏
and the transmission axes of both ANr and ANt be at ⌬n ⬵ , (14)
45° with respect to the x axis. If a sawtoothed signal ⳵␾1 ⳵␾2 ⳵␾2 ⳵␾1
⫺
of frequency f and amplitude V␭兾2, the half-wave volt- ⳵n ⳵k ⳵n ⳵k
age of the EO, is applied to the EO, then by using
Jones calculus we get12,13

Ir ⫽ 1⁄2关1 ⫹ cos共2␲ft ⫺ ␾BS兲兴, ⳵n

冏 冏
⳵␾1
兩⌬␾2兩 ⫹
⳵␾2
⳵n
冏 冏
兩⌬␾1兩

冏 冏
(11) ⌬k ⬵ ,
⳵␾1 ⳵␾2 ⳵␾2 ⳵␾1

It ⫽ 1⁄2 冋兩rs兩2 兩rp兩2

2
⫹
2
冏 冏冏 冏 册
⫹ rs rp cos共2␲f ⫺ ␾兲 , (12)
⳵n ⳵k
⫺
⳵n ⳵k
(15)

where ␾BS is the phase difference between s and p where ⌬n and ⌬k are the errors in n and k and ⌬␾1
polarization that is due to the reflection at the beam and ⌬␾2 are the errors in the phase differences at two
splitter. These two sinusoidal signals are sent to the different incidence angles ␪1 and ␪2, respectively.
phasemeter, and their phase difference The errors in the phase differences in this technique
may be influenced by the following factors:
␾⬘ ⫽ ␾ ⫺ ␾BS (13)
共1兲 Angular resolution of the phasemeter
can be obtained. Next, let the test beam enter pho- The angular resolution ⌬␾p of the phasemeter can
todetector Dt directly without the reflection from the be expressed as15
test material, as shown in Fig. 2共b兲. The test signal
still has the form of Eq. 共12兲 but this time with ␾ ⫽ 0. f
Therefore the phasemeter in Fig. 2共b兲 represents ⌬␾p ⫽ ⫻ 360°, (16)
⫺␾BS. Substituting ⫺␾BS into Eq. 共13兲, we obtain fc
the phase difference ␾.
where f and fc are the frequencies of the input waves
3. Experiments and Results and the reference clock in the phasemeter, respec-
tively. In our experiments we used f ⫽ 800 Hz and
To show the feasibility of this technique, we mea-
fc ⫽ 32 MHz, so the angular resolution of the phase
sured the complex refractive indices of nickel 共Ni兲 and
meter is better than 0.01°.
cupper 共Cu兲. A He–Ne laser with 632.8-nm wave-
length and an EO 共Model PC200兾2, England Electro-
共2兲 Second-harmonic error
Optics Developments, Ltd.兲 with a half-wave voltage
The second-harmonic error comes from the devia-
of 170 V were used in this test. The frequency of the
tion angle ␪r between the polarization directions of p
sawtoothed signal that was applied to the EO was
polarization of the incident beam and the incidence
800 Hz. We used a high-precision rotation stage
plane. It introduces an error in the phase differ-
共PS–␪–90兲 with an angular resolution of 0.005° 共Ja-
ence16:
pan Chuo Precision Industrial Company, Ltd.兲 to
mount and rotate the test material and a high-
resolution phasemeter with an angular resolution of tan ␾共sec 2␪r ⫺ 1兲
⌬␾r ⫽ (17)
0.01° to measure the phase difference. In addition, 1 ⫹ sec 2␪r tan2 ␾

1 July 1999 兾 Vol. 38, No. 19 兾 APPLIED OPTICS 4049

 Fig. 3. ⌬␾r versus ␾ for several values of ␪r. Fig. 4. ⌬␾m versus ␾⬘ for a polarizer with the extinction ratio
shown.

into ␾. The relation curves of ⌬␾r versus ␾ for ␪r ⫽

0.5°, 1.0°, 1.5°, 2.0° are depicted in Fig. 3, and their ␾Ar ⫽ ␾␣r, ␾Br ⫽ ␾␤r, and 兩Ar兩兾兩Br兩 ⫽ 兩␣r兩兾兩␤r兩. Con-
maximum errors are 0.004°, 0.02°, 0.04°, and 0.07°, sequently Eq. 共19兲 can be rewritten as
respectively. It is obvious that ⌬␾r equals zero as
Ir ⫽ 共兩Ar兩 ⫹ 兩␣r兩兲2 ⫹ 共兩Br兩 ⫹ 兩␤r兩兲2 ⫹ 2共兩Ar兩兩Br兩 ⫹ 兩Ar兩兩␤r兩
␾ ⫽ ⫺180°, ⫺90°, 90°, 180°. This error can be made
nearly zero by accurate modification of the azimuthal ⫹ 兩␣r兩兩Br兩 ⫹ 兩␣r円円␤r兩兲cos共2␲ft ⫹ ␾BS兲. (20)
angle of every polarization component, as was done
by Chiu et al.13 Similarly, the intensity of the test signal is given as
It ⫽ 兩At兩2 ⫹ 兩Bt兩2 ⫹ 兩␣t兩2 ⫹ 兩␤t兩2
共3兲 Polarization-mixing error
Owing to the extinction ratio effect of a polarizer, ⫹ 2共兩At兩兩␣t兩 ⫹ 兩Bt兩兩␤t兩兲cos ␾
mixing of light polarization occurs. After a ray light ⫹ 2兵关共兩At兩兩Bt兩 ⫹ 兩␣t兩兩␤t兩兲cos ␾ ⫹ 兩At兩兩␤t兩 ⫹ 兩Bt兩兩␣t兩兴2
passes through a polarizer, its Jones vector should
have the following form: ⫹ 共兩At兩兩Bt兩 ⫺ 兩␣t兩兩␤t兩兲2 sin2 ␾其1兾2 cos共2␲ft ⫺ ␾⬘兲,
(21)

冋册 冋
Ex
Ey
⫽
A exp共i2␲ft兾2兲 ⫹ ␤ exp共⫺i2␲ft兾2兲
B exp共⫺i2␲ft兾2兲 ⫹ ␣ exp共i2␲ft兾2兲
⫽ 册 冋
兩A兩exp共i2␲ft兾2 ⫹ i␾A兲 ⫹ 兩␤兩exp共⫺i2␲ft兾2 ⫹ i␾␤兲
兩B兩exp共⫺i2␲ft兾2 ⫹ i␾B兲 ⫹ 兩␣兩exp共i2␲ft兾2 ⫹ i␾␣兲
, 册 (18)

where ␣ and ␤ are the complex amplitudes of com- where

冋 册
ponents with vibration directions along the x and y
axes, and they are mixed into Ey and Ex, respectively; 共兩At兩兩Bt兩 ⫺ 兩␣t兩兩␤t兩兲sin ␾
␾⬘ ⫽ tan⫺1 ,
f is the frequency difference produced by the EO. 兩At兩兩␤t兩 ⫹ 兩Bt兩兩␣t兩 ⫹ 共兩At兩兩Bt兩 ⫹ 兩␣t兩兩␤t兩兲cos ␾
For clarity the coefficients related to the reference
signal and the test signal are expressed by the sub- (22)
scripts r and t, respectively. Based on the deriva- and we obtain it by comparing the test signal 关Eq.
tions of Jones calculus, the intensity of the reference 共21兲兴 with the reference signal 关Eq. 共22兲兴. Hence the
signal can be written as17,18 polarization mixing error is

Ir ⫽ 兩Ar兩2 ⫹ 兩Br兩2 ⫹ 兩␣r兩2 ⫹ 兩␤r兩2 ⫹ 2兩Ar兩兩␣r兩cos共␾Ar ⫺ ␾␣r兲 ⌬␾m ⫽ ␾⬘ ⫺ ␾. (23)

⫹ 2兩Br兩兩␤r兩cos共␾Br ⫺ ␾␤r兲 ⫹ 2兩Ar兩兩Br兩cos共2␲ft In our experiments, the extinction ratio of the polar-
izer 共Japan Sigma Koki, Ltd.兲 is 1 ⫻ 10⫺5, that is,
⫹ ␾Ar ⫺ ␾Br兲 ⫹ 2兩Ar兩兩␤r兩cos共2␲ft ⫹ ␾Ar ⫺ ␾␤r兲 兩␣兩兾兩A兩 ⫽ 兩␤兩兾兩B兩 ⫽ 0.0032. Introducing these data
⫹ 2兩␣r兩兩Br兩cos共2␲ft ⫹ ␾␣r ⫺ ␾Br兲 into Eqs. 共22兲 and 共23兲 results in the relation curve of
⌬␾m versus ␾⬘ depicted in Fig. 4. We obtain the
⫹ 2兩␣r兩兩␤r兩cos共2␲ft ⫹ ␾␣r ⫺ ␾␤r兲. (19) values of ␾1 and ␾2 in Table 1 by modifying the cor-
responding values of ␾⬘ with the relation curves
Because both interfering beams 共s and p polarization兲 shown in Fig. 4. With this modification, the total
propagate along the same optical path and they are error of the phase difference can be decreased to
influenced by the reflection from the beam splitter, 0.03°.
we have 兩Ar兩 ⫽ 兩Br兩, ␾Ar ⫺ ␾Br ⫽ ␾␣r ⫺ ␾␤r ⫽ ␾BS, Consequently we can obtain the curves of measure-

4050 APPLIED OPTICS 兾 Vol. 38, No. 19 兾 1 July 1999

 Fig. 7. Constant n and k as functions of ⌬n and ⌬k at ␪1 ⫽ 60° and
␪2 ⫽ 80°, where the values of n and k are 0.5 to 4 in steps of 0.5.

close to 60°. ⌬n and ⌬k corresponding to our exper-

imental conditions were calculated and are included
in Table 1.
To investigate the effects of experimental condi-
tions on the measurements we depict in Fig. 6 the
curves of constant n and k as functions of phase dif-
ference ␾ for ␪1 ⫽ 60° and ␪2 ⫽ 80°. In the figure the
values of n and k are shown from 0.2 to 4.0 in steps of
0.2. According to Humphreys-Owen,4 the sensitiv-
ity to the experimental conditions is indicated by the
spacing between contours: If the spacing is large,
Fig. 5. Measurement resolution versus ␪1 for ⌬n and ⌬k of 共a兲 the sensitivity is good and vice versa. It is obvious
nickel and 共b兲 copper at ␪2 ⫽ 80°.
that our experimental conditions are useful if n and k
are small. For good sensitivity, it is better to choose
optimum incidence angles with the method proposed
ment resolution versus ␪1 for ⌬n and ⌬k of nickel and by Logofatu et al.19 Moreover, curves of constant n
copper at ␪2 ⫽ 80° by substituting 兩⌬␾1兩 ⫽ 兩⌬␾2兩 ⫽ and k as functions of ⌬n and ⌬k are shown in Fig. 7.
0.03° into relations 共14兲 and 共15兲. The results are We obtained them by substituting the experimental
shown in Figs. 5共a兲 and 5共b兲, respectively. Obvi- conditions ␪1 ⫽ 60°, ␪2 ⫽ 80°, and 兩⌬␾1兩 ⫽ 兩⌬␾2兩 ⫽
ously, the best resolution can be obtained when ␪1 is 0.03° into relations 共14兲 and 共15兲. In Fig. 7 the val-
ues of n and k are from 0.5 to 4 in steps of 0.5. It can
be seen that both ⌬n and ⌬k are smaller than 1 ⫻
10⫺2 for the test materials with n ⬎ 1 and k ⬍ 2.
Furthermore, this method is highly stable against air
turbulence because of its common path configuration.

5. Conclusion
Based on Fresnel’s equations and the use of hetero-
dyne interferometry, we have developed a new
method for measuring a complex refractive index.
The phase difference between s and p polarization of
the reflected light from an absorbing material is mea-
sured with a heterodyne interferometer. Two phase
differences, corresponding to two different incidence
angles, are measured. These two phase differences
are substituted into Fresnel’s equations to yield a set
of simultaneous equations. Then the equations are
Fig. 6. Constant n and k as functions of ␾ at ␪1 ⫽ 60° and ␪2 ⫽ 80°, solved by use of a personal computer by a numerical
where the values of n and k are 0.2 to 4 in steps of 0.2. analysis technique, and the complex refractive index

1 July 1999 兾 Vol. 38, No. 19 兾 APPLIED OPTICS 4051

of the material can be estimated. The method has 8. J. Lekner, “Determination of complex refractive index and
several merits, including a simple optical setup, easy thickness of a homogeneous layer by combined reflection and
operation, high stability, high measurement accu- transmission ellipsometry,” J. Opt. Soc. Am. A 11, 2156 –2158
共1994兲.
racy, and rapid measurement. Its feasibility has
9. G. D. Feke, D. P. Snow, R. D. Grober, P. J. De Groot, and L.
been demonstrated. Deck, “Interferometric back focal plane microellipsometry,”
This study was supposed in part by National Sci- Appl. Opt. 37, 1796 –1802 共1998兲.
ence Council Taiwan under contract NSC-87-2215-E- 10. M.-H. Chiu, J.-Y. Lee, and D.-C. Su, “Refractive-index mea-
231-002. surement based on the effects of total internal reflection and
the uses of heterodyne interferometry,” Appl. Opt. 36, 2936 –
References 2939 共1997兲.
1. E. D. Palik, Handbook of Optical Constants of Solids 共Academ- 11. M. Born and E. Wolf, Principles of Optics, 6th ed. 共Pergamon,
ic, New York, 1985兲, pp. 69 – 87. Oxford, 1980兲, p. 40.
2. I. Simon, “Spectroscopy in infrared by reflection and its use for 12. D. C. Su, M. H. Chiu, and C. D. Chen, “Simple two frequency
highly absorbing substances,” J. Opt. Soc. Am. 41, 336 –345 laser,” Precis. Eng. 18, 161–163 共1996兲.
共1951兲. 13. M.-H. Chiu, C.-D. Chen, and D.-C. Su, “Method for determin-
3. D. G. Avery, “An improved method for measurements of optical ing the fast axis and phase retardation of a wave plate,” J. Opt.
constants by reflection,” Proc. Phys. Soc. Landon Sect. B 65, Soc. Am. A 13, 1924 –1929 共1996兲.
425– 428 共1952兲. 14. Ref. 1, pp. 285 and 323.
4. S. P. F. Humphreys-Owen, “Comparison of reflection methods 15. N. M. Oldham, J. A. Kramar, P. S. Hetrick, and E. C. Teague,
for measuring optical constants without polarimetric analysis, “Electronic limitations in phase meter for heterodyne inter-
and proposal for new methods based the Brewster angle,” Proc. ferometry,” Precis. Eng. 15, 173–179 共1993兲.
Phys. Soc. 5, 949 –957 共1961兲. 16. J. M. De Freitas and M. A. Player, “Importance of rotational
5. W. R. Hunter, “Error in using the reflectance vs angle of inci- beam alignment in the generation of second harmonic errors in
dence method for measuring optical constants,” J. Opt. Soc. laser heterodyne interferometry,” Meas. Sci. Technol. 4, 1173–
Am. 55, 1197–1204 共1965兲. 1176 共1993兲.
6. W. R. Hunter, “Optical constants of metals in the extreme 17. W. Hou and G. Wilkening, “Investigation and compensation of
ultraviolet. I. A modified critical-angle technique for mea- the nonlinearity of heterodyne interferometers,” Precis. Eng.
suring the index of refraction of metals in the extreme ultra- 14, 91–98 共1992兲.
violet,” J. Appl. Phys. 34, 15–19 共1964兲. 18. A. E. Rosenbluth and N. Bobroff, “Optical sources of nonlin-
7. R. M. A. Azzam, “Simple and direct determination of complex earity in heterodyne interferometers,” Precis. Eng. 12, 7–11
refractive index and thickness of unsupported or embedded 共1990兲.
thin films by combined reflection and transmission ellipsom- 19. P. C. Logofatu, D. Apostol, V. Damian, and R. Tumbar, “Op-
etry at 45° angle of incidence,” J. Opt. Soc. Am. 73, 1080 –1082 timum angles for determining the optical constants from re-
共1983兲. flectivity measurements,” Meas. Sci. Technol. 7, 52–57 共1996兲.

4052 APPLIED OPTICS 兾 Vol. 38, No. 19 兾 1 July 1999