Quality factor temperature dependence of a surface phonon polariton resonance

cavity
James S. Hammonds Jr.,1, a) Kimani A. Stancil,2 and Charlezetta E. Stokes2
1)

Mechanical Engineering Department

Howard University,
Washington, DC 20059
2)

Department of Physics and Astronomy

Howard University,
Washington, DC 20059
(Dated: 9 July 2014)

We consider the surface phonon polariton coupling in an SiO2 optical cavity with 260
nm metal (gold (Au)/chrome (Cr)) side walls, and find a temperature dependence of
the quality factor, Q = o /. By using optical cavities of varying widths between
parallel metal walls and FTIR-ATR measurements, we first observe that the quality
factor obeys an inverse power law dependence on the width. And by relating the
widths to the optical path length, and ultimately to the temperature using the general
thermo-optical coefficient, we show the quality factor temperature dependence. We
argue that the temperature dependence of the quality factor is a practical and almost
universal result that describes the energy dissipative behavior of both mechanically,
and optically responsive systems.

a)

Electronic mail: jhammonds@howard.edu

I.

INTRODUCTION
The ability to sense temperature changes in nanoscale systems is becoming increasingly

important in application areas such as microelectronics, microfluidics, and nanomedicine

which require temperature measurement with submicron (100s of nanometers) resolution1 .
Transmission electron microscopy (TEM) has been used to achieve temperature mapping
with submicron resolution by probing local phase changes in a heated sample2 . However,
these systems are currently impractical for many applications since they require specialized
conditions such as an evacuated chamber. Submicron temperature measurements are also
possible with near-field scanning optical microscopy (NSOM), which uses a nanoscale fiber
or metallic tip as a light source that illuminates a sample from a few nanometers above3 .
The changes in surface optical properties with temperature can then be monitored at the
nanoscale. High spatial resolution is achieved because the highly localized light field between the NSOM tip and sample surface is not diffraction limited. However, areas within
microfluidics and nanomedicine that deal with fluids at room temperatures are restricted
from NSOM analysis since vacuum or cryogenic temperatures are necessary1 .
The success of NSOM suggests that near-fields could be effective input signals for submicron temperature measurements in practice if the limitations of the experimental apparatus
can be overcome. NSOM alternatives for near-field light based temperature sensing include
the work of Davis and Deutsch (2010) who demonstrate the use of highly localized fields at
the liquid/metal film interface to probe changes in the liquid reflectance with temperature4 .
The near-field light source in this case is a thin metal film, and the physical mechanism
involves plasmonic resonance modes at the liquid/metal interface. To drive the plasmonic
resonance modes, however, the metal needs to be pumped with an external light source using
an optical system called the Kretschmann configuration5 . While the metallic film thickness
is submicron, the Kretschmann configuration itself is macroscale since it focuses far-field
light into the near-field. Its size, in effect, can limit its use in practical applications.
To overcome the size limitations of the Kretchmann optical configuration, Srivastava
(2013) proposed the use of a photonic crystal (PC) to couple an external light source into
plasmonic resonance modes6 . In this case, the PC is in thermal equilibrium with a sample,
and plasmonic resonance varies with sample temperature. Andersen et al (2011) show how
waveguides can be used to achieve light coupling, and to detect transmission dependence on
2

temperature7 . The work of Srivastava and Andersen demonstrate that we are not limited
to Kretschmann optical configurations, but external optical pumps were used in both cases.
The work described above shows how near-field effects have been employed to sense temperature changes. A key example involves the use of metallic thin films as plasmonic resonance light sources in the Kretchmann configuration. These results suggest that analogous
systems may use a silicon dioxide or silicon carbide film as a phononic resonance light source.
The ability to monitor and couple to plasmonic or phononic resonance modes is related to the
mode lifetime, or its associated propagation distance. Plasmonic modes at metal/dielectric
interfaces can propagate 10s to 100s of microns, while optical phonon polariton propagation at silicon carbide thin film/dielectric interfaces, can be 1000s of microns8 . The longer
propagation distances associated with phonon resonances implies easier coupling, and so it
may be easier to design phononic optical systems with more compact optical sources. For
example, recent developments in our understanding of nanoscale thermal radiation, show
that optical phonon resonance at the surface of silicon carbide and silica are pumped by
thermal energy alone, i.e. heating, which has led to the suggestion that spectrally selective
thermal emitters can be designed by exploiting optical phonon resonance913 .
In this letter we experimentally show how phononic resonances can be temperature dependent and used for temperature sensing, paving the way for phononic thermosensors.
We estimate the temperature dependence of optical phonon resonance modes confined to a
metal-SiO2 -diamond-dielectric cavity, by studying how the cavity quality factor varies with
its optical width. We then relate the optical width to the dielectric thermo-optical properties
to obtain an estimate of Q-factor dependence on temperature.

II.

EXPERIMENTAL
Microfabrication at the Cornell NanoScale Science & Technology Facility (CNF) was used

to pattern and deposit 10 nm of Cr and 250 nm of gold onto a 500 m thick SiO2 wafer
to form a 1 mm long channel with 260 nm high metal walls, and an SiO2 floor of width,
w. This channel is flipped and pressed against the diamond crystal on a PerkinElmer Spectrum 100 FTIR-ATR spectrometer to form the cavity cross-section shown in Fig. 1. The
FTIR-ATR system provides the diamond crystal and the polarized light source that pumps
the optical phonon modes in the cavity. At = 45 the incident beam is totally reflected
3

SiO2
ni

Cr

Au
w
Di

FIG. 1. Longitudinal and transverse optical phonon cavity formed by a channel of width, w, with
gold/Cr walls, a silicon dioxide top and a diamond bottom. The channel is filled with a dielectric
of refractive index, ni . p-polarized infrared light, , impinges on the diamond/dielectric interface
at an angle . With ni < ndiamond and = 45 , evanescent waves at the diamond/dielectric
interface extend across the gap to support optical phonon modes at the SiO2 /diamond interface.
The resulting SiO2 resonance modes are confined within the cavity by gold walls.

as long as the refractive index of diamond is greater than the channel region refraction index, ni . The evanescent field resulting from the beam frustration at the diamond/dielectric
interface couples to optical phonon resonance modes localized at the dielectric/SiO2 interface. Furthermore, the metallic walls are highly reflective to the infrared optical phonon
modes resulting in light field confinement between the gold, SiO2 , and diamond interfaces.
We focus our analysis on transverse optical (TO) phonons that can be sustained in the
frequency range between a = 864cm1 and b = 1241cm1 . The transverse optical (TO)
modes within this region are the transverse Si-O-Si vibrations at 1064cm1 and 1160cm1 ,
and the Si-OH stretch at 960 cm114,15 . The coupled light-lattice vibrations at 1160cm1 ,
propagate long distances (1000s m) are called surface phonon polaritons, and have been
widely studied as possible sources for spectrally selective infrared emitters913 .
The FTIR-ATR detects the light coupling to transverse optical phonon modes as dips in
the reflection spectra. The results of the FTIR-ATR analysis is shown in Fig. 2. We estimate
the quality factor of the cavity by determining the full-width-half-max, , and the dip
minimum location, o , of the frequency pulse in the range defined by the vertical dashed lines
ab
in Fig. 2. is defined as follows: First note that 0 is the spectral location of Rmin
, which

is the minimum ATR reflectivity between the range a = 864cm1 and b = 1241cm1 , as
indicated by the dashed vertical lines in Fig. 2. So = +0.5min 0.5min , where +0.5min
4

Si-OH

100 m

80

200 m

100 m

15
10

400 m

60

600 m (solid)
no channel (dash)

40

900

1000

1100

1200

600 m

50

1000 m

20

30
10

800 m
0
800

SPhP

+ mode

ATR Reflectivity

100

TOPh

900

1000

900

1100

[cm1]

1200

1000

1100

1300

1200

1400

FIG. 2. p-polarized FTIR-ATR of geometry shown in Fig. 1. The full-width-half-max and maximum of pulse are calculated for the spectral range between the vertical dashed lines to determine
quality factor of the quartz optical phonon surface resonances confined within the cavity (c.f.
Fig. 3). The inset shows the 3-peak Gaussian fitting for spectra for sample widths, w = 100 m,
and 600 m. All spectral curves are labeled by channel widths including w = 100 m, 200 m,
400 m, 600 m, 800 m, 1000 m.

ab
ab
is the spectral location of A = 0.5 (1 Rmin
) to right of Rmin
, and 0.5min is the location
ab
of A = 0.5 (1 Rmin
) to its left.

III.

DISCUSSION

We now discuss the physical reason for this increase in Q, and show how Q variation with
w can be explored for temperature measurement. In Fig. 3 we plot the Q-factor variation
with channel width for both p-polarized and s-polarized IR light, over the same a to b
range. It is important to note that the dramatic increase in the Q-factor within this range
is specific to the p-polarization data. This observation suggests that long-range transverse
vibrations coupled to light fields called surface phonon polaritons (SPhP) are responsible
for the Q-factor increase, since SPhP are p-polarized phenomena.
To investigate the contribution of individual vibrational modes on Q-factor variations
with channel width, we used a summation of Gaussian functions to deconvolve the FTIR5

1.2

spol
ppol
fit

0.8
0.6
0.4
0.2
0

200

400

600

Channel Width

800

1000

FIG. 3. Q-factor variation with channel width, w (filled squares, dashed line included for reference).
Both s- and p- polarization are shown. P-polarization Q values exhibit a different behavior and
increase at smaller cavity widths. The solid curve is the equation Q = 450w1.3 .

80

70

SiOH
SPhP
TOPh

Ai [%]

60
50

60

40
30
20
10

Qi

40

200

400

600

800

1000

Channel Width

20

200

400

600

Channel Width

800

1000

FIG. 4. Examination of Gaussian fits for 3 peaks corresponding to Si-OH, TOPh, and SPhP
modes. Quality factors are determined by i /i where i indicates the characteristic peak, and
the Gaussian fit is used to determine i . We observe SPhP Q-behavior similar to Figure 3.

ATR spectra for every width. Representative examples of our Gaussian fits are shown in the
two insets to Fig. 2. In approaching our peak fitting, we are guided by the well-known optical
transverse vibrational modes associated with SiO2 : SiOH = 960cm1 , T OP h = 1064cm1 ,
SP hP = 1160cm1 . Gaussian fitting allows us to determine the spectral location of the
Gaussian curve peak, i most closely related to the TO modes i = Si-OH,TOPh, SPhP. We
then calculated i as full-width half max of each Gaussian associated with mode i. The
results of Q = i /i are plotted in Fig. 4 which shows that while the Si-OH and TOPh
vibrations lose energy as the channel width is decreased, the SPhP mode has a Q-factor
6

that grows as the channel width is shortened. We also used the Gaussian peak height to
estimate the change in absorptivity A = 1 R for each gaussian peak i for varying w. As
expected, the cavity absorptivity decreases for all three vibrational modes as the channel
width decreases. Since there is strong suggestion that the SPhP TO mode is responsible for
the increase in Q-factor, we estimate the TO lifetime by performing an exponential fit of
the Q-factor vs. width data as
Q = 450w1.3

(1)

which is also plotted in Fig. 3. The cavity resonator establishes a standing wave condition
that may support traveling waves along the interface which are long lived if the cavity walls
are narrow enough. When the width is large, measured effects should not differ from when
no cavity is present. As the cavity narrows, the cavity walls support multiple evanescent
waves associated with the beam and an expected enhancement should occur. The beam
properties assumed to be Gaussian do not allow for a prediction of when this enhancement
should occur but the observation of enhancement, should occur at lengthscales greater than
the wavelength. Sub-wavelength behavior was not considered by this, our initial study.
Typically, resonant cavity quality factors are expressed by ratios of polynomials for fitting
purposes, however, the expected increase at diminished cavity width was qualitatively described by an exponential fit to simply estimate how long-lived the phonon-polariton or
other excitation influences would be, i.e. using the interpretation that the Q-factor roughly
estimates the number of cycles needed for the system to substantially lose most of its energy.
To probe the physics more deeply, we expanded the spectral range to include the characteristic TO phonon peak14 at 795 cm1 and used Igor Pro software to do an iterative 8
gaussian fit for spectra across all widths, w. In Table I, we list the peaks and their standard
deviations. The referential peaks seen in the literature are listed on top for each column.
The 960 cm1 SI-OH peak is not strongly evident but we expect its influence. Our experimental or material preparation does not enable a clear estimate of its significance to our
system.
Ref. Peak [cm1 ] 795
1064
1111
1160
795
960
960
1111
1
Peak Fit [cm ] 766.1586 1069.671 1119.729 1169.271 801.9957 916.1329 1005.884 1097.143
STD
31.96713 10.25292 14.92534 15.01141 3.429372 18.49387 10.62418 6.284865
TABLE I. Detailed gaussian curve fit.

To determine Q-factor dependence on temperature we observe that the optical path length
between the cavity walls is also related to the temperature. The optical path is defined for
waves traveling along the SiO2 interface. This path is fixed with respect to refraction index
and should change with allowable standing waves. Thus, the propagation vector, k, changes
inversely with the width. Each spectrum is width dependent. For temperature variation,
one should explore our result using fixed channel width and varied temperature as well as
considering different widths to include variations in the refraction index with temperature.
This is ultimately realized by the optical path length, wopl = ni w where both ni and w
may depend on temperature. In our study, we compare different widths at the same fixed
temperature only. We do note that for liquids, additional variations may occur for large vs.
small widths when considering non-equilibrium or quasi-equilibrium conditions where the
diffusive freedom of particles may play a role.

For liquids, the differential path optical path length, dwopl , is related to the thermo-optical
coefficient, dn/dT , and the physical length, wo , by
dwopl,liq = wo

dni
dT
dT

(2)

while for solids




dni
= wo
+ + q dT
dT

dwopl,sol

(3)

where accounts for the optical effects of thermal expansion, and the q term characterizes the thermal stresses16 . The physical length is viewed as based on the surface structure
of the polar substrate. For a lattice type structure the physical length, wo , corresponds to
the characteristic lattice spacing. It is clear from Eqs. 2 and 3 that the form of the differential optical path is dwopl = CdT when dni /dT , , and q do not vary with temperature.
The dwopl = CdT approximation is valid over small temperature ranges around room temperature. Integrating to a temperature T , which is a short distance from the equilibrium
temperature, T0 , or
Z

wopl

dw = wo
wi,opl

CdT

(4)

T0

wopl wi,opl = wo C[T0 T ]

8

(5)

and since wi and T0 are constants that can be arbitrarily chosen, we can write the optical
path length dependence on temperature as
wopl (T ) = wo CT

(6)

which is applicable to liquids and solids for small changes in T . Because of the empirical
influence of both the transverse optical phonon modes and the apparent phonon-polariton
influence, we estimate that the cavity width serves as an effective optical path length, wopl =
w, such that with the discussion above on temperature-dependence, then Q can be rewritten
to depend on temperature. Now using Eq. (6) with Eq. (1), the Q-factor variation with
temperature may be written as
Q = 450(wo CT )1.3 .

(7)

It is interesting to note that Q T a dependence is also observed in other mechanical and

acoustic resonance systems as well. For example, to gain understanding into the energy
loss mechanisms in microelectromechnical (MEMS) resonators, Kim et al (2008) analyzed
the temperature dependence of the quality factor in a MEMS resonator17 . They found the
same Q T a and concluded that the dependence could allow for thermoetery. Galliou
et al (2008) also found Q T 4 dependence in bulk acoustic wave resonators at cryogenic
temperatures18 .

IV.

CONCLUSION

We demonstrate that it may be possible to observe temperature changes with nanoscale

resolution via the q-factor of optical surface phonon modes confined to a rectangular cavity.
We show that temperature dependence can be approximated as following Q T a , a power
rule that has been observed in other mechanical and acoustic resonance systems. This
work may contribute to the development surface phonon polaritons based thermometry with
broader applicability than current temperature sensing systems with submicron resolution.
9

V.

ACKNOWLEDGMENT
Author K.A.S. acknowledges the 2010 NNIN Training grant with the Cornell NanoScale

Science & Technology Facility (CNF) and the Howard University Program for the Expansion
of Research and Education in Nanotechnology (HUPEREN) grant from Howard University.
Author C.E.S. acknowledges the Howard University Graduate School for an assistantship
award. This work was performed in part at the Cornell NanoScale Facility, a member of
the National Nanotechnology Infrastructure Network, which is supported by the National
Science Foundation (Grant ECCS-0335765).

REFERENCES
1

C. D. Brites, P. P. Lima, N. J. Silva, A. Millan, V. S. Amaral, F. Palacio, and L. D.

Carlos, Nanoscale 4, 4799 (2012).

T. Brintlinger, Y. Qi, K. H. Baloch, D. Goldhaber-Gordon, and J. Cumings, Nano letters

8, 582 (2008).

E. Hutter and J. H. Fendler, Advanced Materials 16, 1685 (2004).

L. Davis III and M. Deutsch, Review of Scientific Instruments 81, 114905 (2010).

H. Raether, Surface plasmons on smooth surfaces (Springer, 1988).

T. Srivastava, R. Das, and R. Jha, Plasmonics 8, 515 (2013).

T. B. Andersen, Z. Han, and S. I. Bozhevolnyi, Sensors 11, 1992 (2011).

K. A. Willets and R. P. Van Duyne, Annu. Rev. Phys. Chem. 58, 267 (2007).

J. Le Gall, M. Olivier, and J.-J. Greffet, Physical Review B 55, 10105 (1997).

10

J.-J. Greffet, R. Carminati, K. Joulain, J.-P. Mulet, S. Mainguy, and Y. Chen, Nature
416, 61 (2002).

11

N. Dahan, A. Niv, G. Biener, V. Kleiner, and E. Hasman, Applied Physics Letters 86,
191102 (2005).

12

S. Shen, A. Narayanaswamy, and G. Chen, Nano Letters 9, 2909 (2009).

13

E. Vinogradov, G. Zhizhin, and V. Yudson, Surface Polaritons. Electromagnetic Waves at

Surfaces and Interfaces, eds VM Agranovich and DL. Mills (North-Holland, Amsterdam,
1982) ch 4, 145 (2012).

14

W.-T. Liu and Y. Shen, Physical Review B 78, 024302 (2008).

10

15

W.-T. Liu and Y. Shen, Physical review letters 101, 016101 (2008).

16

A. Andrade, T. Catunda, I. Bodnar, J. Mura, and M. Baesso, Review of scientific instruments 74, 877 (2003).

17

B. Kim, M. A. Hopcroft, R. N. Candler, C. M. Jha, M. Agarwal, R. Melamud, S. A.

Chandorkar, G. Yama, and T. W. Kenny, Microelectromechanical Systems, Journal of
17, 755 (2008).

18

S. Galliou, J. Imbaud, R. Bourquin, N. Bazin, P. Abbe, et al., in Proceedings of European

frequency and time forum (2008).

11