Dielectric constant measurement of thin films by differential time-domain spectroscopy

Zhiping Jiang, Ming Li, and X.-C. Zhang

Citation: Applied Physics Letters 76, 3221 (2000); doi: 10.1063/1.126587
View online: http://dx.doi.org/10.1063/1.126587
View Table of Contents: http://scitation.aip.org/content/aip/journal/apl/76/22?ver=pdfcov
Published by the AIP Publishing
Articles you may be interested in
Determination of complex refractive index of thin metal films from terahertz time-domain spectroscopy
J. Appl. Phys. 104, 053110 (2008); 10.1063/1.2970161
Far-infrared optical and dielectric response of ZnS measured by terahertz time-domain spectroscopy
Appl. Phys. Lett. 86, 131111 (2005); 10.1063/1.1896451
Optical, structural, and electrical characteristics of high dielectric constant zirconium oxide thin films deposited by
spray pyrolysis
J. Vac. Sci. Technol. A 22, 1319 (2004); 10.1116/1.1701866
Time-domain dielectric constant measurement of thin film in GHzTHz frequency range near the Brewster angle
Appl. Phys. Lett. 74, 2113 (1999); 10.1063/1.123773
Far-infrared dielectric constant of porous silicon layers measured by terahertz time-domain spectroscopy
J. Appl. Phys. 83, 6007 (1998); 10.1063/1.367467

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
129.11.76.232 On: Mon, 27 Oct 2014 17:52:42

APPLIED PHYSICS LETTERS

VOLUME 76, NUMBER 22

29 MAY 2000

Dielectric constant measurement of thin films by differential time-domain

spectroscopy
Zhiping Jiang, Ming Li, and X.-C. Zhang,a)
Physics Department, Rensselaer Polytechnic Institute, Troy, New York 12180-3590

Received 28 February 2000; accepted for publication 6 April 2000

We present a theoretical model and preliminary experimental results on the dielectric constant
measurement of thin films by using differential time-domain spectroscopy. This technique greatly
reduces the minimum measurable thickness, and it promises the dielectric constant measurement of
m-thick thin films with the frequency range from GHz to THz. 2000 American Institute of
Physics. S0003-69510003922-X

Replacing silicon dioxide in the interlayer with low dielectric constant low- materials to reduce the RC time
delay is a method of boosting the speed of the semiconductor
device.1 It is crucial to know the dielectric constant or refractive index and its dispersion information of such film materials in the gigahertz GHz to terahertz THz range. Conventional metrology methods have frequency limits, e.g., the
low-frequency limit of a standard ellipsometry is about 20
THz and the mercury contact CV method cannot be operated
with a frequency over a few megahertz.2 Time-domain spectroscopy TDS, employing the phase-change measurement
of the electromagnetic wave traveling through the tested
sample, provides a promising tool for material characterization in this frequency range.35 However, due to the relatively long wavelength, the conventional TDS does not provide the sensitivity for the measurement of m-thick films.
Phase measurement near the Brewster angle shows
improvement,6 but it is very time consuming to complete
multiscans in the experiment. In this letter, we demonstrate a
differential time-domain spectroscopy DTDS to determine
the refractive index using the ratio of the differential amplitude to the field amplitude. The capability of measuring dielectric films with m thickness is experimentally demonstrated.
Figure 1 shows the notations used in this letter. We discuss the problem in the frequency domain. E 0 ( ) is the incident field, and E ref( ) and E diff( ) are the reference without the film and signal with the film field, respectively. d is
the thickness of the film. n 1 , n 2 , and n 3 are the refractive
indices. For a film covering on a substrate, n 1 1 air, and
n 3 is the refractive index of the substrate, while n 1 n 3 1
for the case of a free-standing film. The refractive index of
the film n 2 ( ) is the quantity under measurement and usually it is frequency dependent. Its relationship with the dielectric constant 2 ( ) is 2 ( )n 22 ( ), where is the
angular frequency. During the experiment, the beam is fixed
while the sample is moved in and out the THz beam. r i j and
t i j represent the reflection and transmission coefficients of
the i j interface i, j1, 2, and 3, and i j, respectively.
For example, t 23 is the transmission coefficient from the film
to the substrate.
a

Author to whom correspondence should be addressed; electronic mail:

zhangxc@rpi.edu

The reference field can be written as

E ref t 13E 0 exp i d/c .

The phase in E ref( ) is referenced to the 12 interface. By

considering the multireflection within the film, the field
E film( ) after the film has the form7
E film

t 12t 23 exp i d
E ,
1r 21r 23 exp 2i 2 d 0

where is the amplitude absorption coefficient, and

is the phase change of the field traveling through the film,

n 2 d/c.

c is the speed of light in vacuum.

Unlike the conventional TDS experiment, where the reference E ref(t) and the signal E film(t) are measured in the
time domain, DTDS employs the reference and a differential
signal E diff(t) defined as
E diff t E film t E ref t .

In the frequency domain, the differential signal E diff(t) can

be written as
E diff E film E ref .

For a thin film, dc/ , hence ( )1. Under this approximation, we have

E diff
n 2 n 1 n 2 n 3
i d n 2 1
E ref
c
n 1 n 3

d 1

n 2 n 1 n 2 n 3
.
n 1 n 3

FIG. 1. n 1 , n 2 , and n 3 are the refractive indices. E 0 ( ) is the incident field,

and E ref( ) and E film( ) are the reference without the film and signal
with the film fields, respectively. d is the thickness of the film.

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
0003-6951/2000/76(22)/3221/3/$17.00
3221
2000 American Institute of Physics
129.11.76.232 On: Mon, 27 Oct 2014 17:52:42

3222

Appl. Phys. Lett., Vol. 76, No. 22, 29 May 2000

Jiang, Li, and Zhang

In principle, both n 2 ( ) and can be obtained by

Eq. 6. For a high-quality dielectric film with a small loss
tangent, the absorption can be neglected. Note that when the
film thickness is very small, E diff(t) is essentially proportional to the derivative of E ref(t). There is a /2 phase difference between E diff( ) and E ref( ), therefore n 2 ( ) is
given by E diff( )/E ref( ) .
Denoting
A

c E diff
,
d E ref

Eqs. 6 and 7 lead to

n 2 1A n 1 n 3 n 1 n 3 .

For a free-standing film (n 1 n 3 1), Eq. 8 is simplified to

n 2 2A 1 .

Equation 9 directly gives the refractive index n 2 ( ) from

the measured A( ).
To analyze the noise, the noise percentage of
E diff( )/E ref( ) left-side term of Eq. 6 is expressed as

E diff
E ref

diff ref .
E diff
E ref

10

diff( ) and ref( ) are the noise percentage for the measurement of E diff( ) and E ref( ), respectively.
In the conventional time-domain spectroscopy measurement, E diff( ) is obtained by subtracting E film( ) from
E ref( ). If the film is very thin, E diff( ) E ref( ) and
E diff( ) & E ref( ) , therefore, the relative noise
diff( ) ref( ). It is seen that
E diff E ref ,

11

with the film-induced phase ( ) n 2 ( )1 d/c. By

using Eq. 11, Eq. 10 can be rewritten as

ref b &/ 1 c .

12

When the phase change decreases due to the film thickness decrease, the measurement error increases, inversely
proportional to the phase change.
Because the measurement noise mainly comes from the
differential signal E diff( ), the remedy is to increase the
signal-to-noise ratio SNR of E diff( ). The differential measurement technique is ideal for this purpose.
Figure 2 shows the setup for the DTDS. This setup is
similar to the conventional THz system. The laser is a regenerative amplified Ti:sapphire laser Coherent Rega 9000
which provides laser pulses with an 800 nm central wave-

length, 120 fs laser pulse duration, and 250 kHz repetition

rate. A 2-mm-thick 110 ZnTe crystal is used as the THz
emitter via optical rectification. Part of the ultrashort laser
pulse is used to generate the THz pulse. The THz pulse is
collimated and refocused by parabolic mirrors, and is then
coherently detected. Unlike in the conventional setup that
uses a mechanical chopper to either modulate the optical
pump beam or the THz beam, here the film sample either
free standing or on the substrate is moved in and out of the
THz beam by a galvanometer. The sample is located in the
focal plane and is vibrated by the galvanometer at a frequency of 16 Hz. The shaking distance of about 5 mm is big
enough to move the sample in and out of the THz beam
completely. A lock-in amplifier synchronized to the galvanometer reference signal measures the differential signal.
Free-space electro-optic sampling in cross-balance geometry
is used to coherently measure the THz wave form.8
In the experiment, both the differential wave form
E diff(t) and the original reference E ref(t) are measured under
the same conditions, except when E ref(t) is measured, the
dielectric film is replaced by an aluminum foil that blocks
the THz beam completely. The differential signal reduces the
common noise, and the SNR is much greater than that in the
conventional TDS measurement.
Figure 3 plots the reference wave form and the differential wave form for a free-standing parylene-N Pa-N film
with a thickness of 1.8 m, the differential signal is much
smaller than the reference signal but is still measured with a
good SNR. Figure 4 shows the calculated refractive index
from the data in Fig. 3 by using Eqs. 7 and 9. Results
from four repeated measurements are shown in Fig. 4, the
height roughly represents the measurement-to-measurement
error. With the conventional TDS, it is nearly impossible to
measure such a thin film. For example, at 1 THz,
0.004 rad, according to Eq. 12, this gives 355 ref .
Even for a small ref1%, we still have 3.55. As a
comparison, the differential wave form measured by the
DTDS method, the reference and signal wave forms by the
conventional TDS method, as well as the difference between
the two repeated reference wave forms, are plotted in Fig. 5.
The noise is much bigger in the conventional TDS method
top and middle curves, information on n 2 ( ) cannot be
obtained.
With the conventional TDS, diff( ) cannot be reduced
by increasing the dynamic range of the reference signal because the increase of the dynamic range does not decrease
ref( ). The noise percentage in E diff( ) is bigger than in

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
FIG. 2. Experimental setup.
FIG. 3. Measured THz wave forms of the differential and reference signals.
129.11.76.232 On: Mon, 27 Oct 2014 17:52:42

Appl. Phys. Lett., Vol. 76, No. 22, 29 May 2000

FIG. 4. Measured refractive index of the 1.8 m Pa-N film.

E ref( ) by a factor of &/, which is always much larger

than 1 for thin films. However, with DTDS, diff( ) can be
as small as ref( ), provided that there is a strong THz signal. In this sense, DTDS converts the large dynamic range
into a better SNR.
The higher the dynamic range of the system, the stronger
the difference signal E diff(t). Thus, the minimum measurable
thickness of a free-standing film depends only on the system
dynamic range. For a film on the substrate, the minimum
measurable thickness depends on the uniformity of the substrate, too. If the substrate is not uniform, it also contributes
to the differential signal. For a THz system with an amplified
laser, the dynamic range can be greater than 1 million,9
hence, the uniformity of the substrate is the ultimate factor
that limits the minimum measurable thickness.
With conventional TDS, it is even harder for the lowfrequency measurement because the phase change is
very small. For instance, for 100 GHz, the wavelength is 3
mm. If the film thickness is 10 m, the phase change is only
a 1/600 wavelength with a refractive index of 1.5. Even if
there is no noise at all, to obtain such a small phase change at
such a low frequency, a fine step size much less than 10
m and a long scanning range much larger than 3 mm are
required, leading to a very long measuring time. With
DTDS, the phase change is obtained by the relative amplitude ratio E diff( )/E ref( ), the step size is determined only
by the Nyquist sampling theorem.
In conclusion, we demonstrated the dielectric constant
measurement of thin films by using differential time-domain

Jiang, Li, and Zhang

3223

FIG. 5. Wave form comparison. From bottom to top: differential signal

E diff(t) using the DTDS technique and the difference between two repeated
(t), and E film(t)E ref(t) using the conreference wave forms E ref(t)E ref
ventional TDS. No clean signal can be obtained from the top curve conventional TDS.

spectroscopy. We present the theoretical model and preliminary experimental result. This technique greatly reduces the
minimum measurable thickness, and promises dielectric constant measurement of submicron film at a frequency below
10 GHz.
The work is supported in part by the Army Research
Office, the National Science Foundation, and the Semiconductor Research Corporation Grant 448.045. The authors
are grateful for helpful discussions with F. G. Sun and the
sample preparation by Jeff Fortin.
W. W. Lee and P. S. Ho, MRS Bull. 19 1997.
D. K. Schroder, Semiconductor Material and Device Characterization,
2nd ed. Wiley, New York, 1998, p. 79.
3
L. Duvilaret, F. Garet, and J.-L. Coutaz, IEEE J. Sel. Top. Quantum Electron. 2, 739 1996.
4
D. Grischkowsky, S. Keiding, M. van Exter, and Ch. Fattinger, J. Opt.
Soc. Am. B 7, 2006 1990.
5
M. C. Nuss and J. Orenstein, in Millimeter and Submillimeter Wave Spectroscopy of Solids, edited by G. Gruner Springer, Berlin, 1998, Chap. 2,
pp. 743, and references cited therein.
6
M. Li, G. C. Cho, T.-M. Lu, X.-C. Zhang, S.-Q. Wang, and J. T. Kennedy,
Appl. Phys. Lett. 74, 2113 1999.
7
M. Born and E. Wolf, Principles of Optics, 6th ed. Cambridge University, Cambridge, U.K., 1980, Chap. 7.6.
8
Z. Jiang and X.-C. Zhang, IEEE Trans. Microwave Theory Tech. 47, 2644
1999.
9
Z. G. Lu, P. Campbell, and X.-C. Zhang, Appl. Phys. Lett. 71, 593 1997.
1
2

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
129.11.76.232 On: Mon, 27 Oct 2014 17:52:42