Chapter 1

Overview

See how various the forms, and how unvarying the principles.

Owen Jones
The grammar of ornament, p. 157
Portland House, New York, NY, USA, 1986

Downloaded From: http://ebooks.spiedigitallibrary.org/ on 09/25/2013 Terms of Use: http://spiedl.org/terms

 2 Chapter 1

A bird’s eye view of research on sculptured thin films is presented in this chapter.
Its scope is wider than the combined scopes of the remaining chapters, but the
discussion is, of course, shallower. We recommend that this chapter be read before
any other in this book.
Sculptured thin films (STFs) are nanostructured inorganic materials with an-
isotropic and unidirectionally varying properties that can be designed and realized
in a controllable manner using physical vapor deposition (PVD). This overview
chapter traces the development of STFs from their precursors, with emphasis on
two seminal events that occurred in 1959 and 1966, as well as on the sculptured
columnar morphology of these materials. The defining continuum electromagnetic
constitutive relations of STFs are presented, along with a control model to link
the nanostructure of STFs to their continuum properties via a local homogeniza-
tion formalism, for optical purposes. Finally, a tour of accomplished and emerging
applications is taken.

1.1 Introduction

In its decadal survey entitled Physics in a New Era conducted during the 1990s, the
U.S. National Research Council (NRC) explored research trends and requirements
in the materials sciences [1]. A dominant theme that emerged is of nanosciences
and nanotechnologies. The nanoscale is Janusian: matter at the 10- to 100-nm
length scale exhibits continuum characteristics, but molecules and their clusters
of small size can still display their individuality. For that reason, the U.S. National
Science Foundation (NSF) has begun to focus on material morphologies and archi-
tectures with at least one dimension smaller than 100 nm in its research initiatives.
Among the nanoengineered materials identified by the NRC are STFs [1,
p. 123]. These nanostructured inorganic materials with anisotropic and unidirec-
tionally varying properties can be designed and fabricated in a controllable manner
[2–5] using PVD, a century-old technique.1 The ability to virtually instantaneously
change the growth direction of their columnar morphology, through simple varia-
tions in the direction of the incident vapor flux, leads to a wide spectrum of colum-
nar forms. These forms can be

(i) two-dimensional, ranging from the simple slanted columns and chevrons
to the more complex C- and S-shaped morphologies [6]; or
(ii) three-dimensional, including simple helixes and superhelixes [7].

A few examples of STFs are presented in Figs. 1.1 and 1.2, and a representative
list of materials deposited as STFs is provided in Table 1.1.
For most optical applications envisioned, the column diameter and the column
separation normal to the thickness direction of any STF should be constant. The
1 An excellent history of vacuum coating technologies by D.M. Mattox was serialized in the Febru-
ary 2002, March 2002, May 2002, August 2002, October 2002, November 2002, and February 2003
issues of the magazine Vacuum Technology & Coating. The complete work recently became available
as a book [8].

Downloaded From: http://ebooks.spiedigitallibrary.org/ on 09/25/2013 Terms of Use: http://spiedl.org/terms

 Overview 3

Figure 1.1 SEM micrographs of STFs made of magnesium fluoride (MgF2 ) with
two-dimensional morphologies: (a) 7-section zigzag; (b) C-shaped; (c) S-shaped.

Figure 1.2 SEM micrographs of STFs with three-dimensional morphologies: (a) helical,
made of silicon oxide (SiO); (b) superhelical, made of MgF2 .

Table 1.1 Materials deposited as STFs.

oxides metals semiconductors
aluminum oxide aluminum carbon
silicon oxide bismuth germanium
silicon dioxide chromium silicon
tantalum oxide copper
titanium oxide iron polymers
zirconium oxide platinum parylene
titanium teflon
fluorides tungsten
calcium fluoride
magnesium fluoride

column diameter can range from ∼ 10 to 300 nm, while the density may lie be-
tween its theoretical maximum value and less than 20% thereof. The crystallinity
must be at a scale smaller than the column diameter. The chemical composition
is essentially unlimited, ranging from insulators to semiconductors to metals. De-

Downloaded From: http://ebooks.spiedigitallibrary.org/ on 09/25/2013 Terms of Use: http://spiedl.org/terms

 4 Chapter 1

spite the fact that precursors of STFs have been made for over a century [9–16],
systematic exploration of the science and technology of STFs began only during
the mid-1990s. What is emerging from recent research [17–23] is that STF tech-
nology has a bright future.
At visible and infrared wavelengths, a single-section STF is a unidirectionally
nonhomogeneous continuum with anisotropic constitutive properties. Several sec-
tions can be grown consecutively into a multisection STF, which can be conceived
of as an optical circuit that can be integrated with electronic circuitry on a chip.
Being porous, an STF can act as a sensor of fluids and can be impregnated with
liquid crystals for switching applications too. Their application as low-permittivity
barrier layers in electronic chips as well as for solar cells has also been suggested.
Optical applications of STFs began to be systematically reported in 1999, although
earlier instances can be found in the literature [12, 16, 24].
Looking back at the histories of various technologies, one finds two distinct
patterns of early evolution. More often than not, a set of key experimental observa-
tions spurs theoretical developments. Less often, experimental research is engen-
dered by theory—as exemplified by STFs. Right from the beginning, STF research
was theoretical [2, 25]. STFs were conceived for optical purposes to be nonhomo-
geneous in the thickness direction but transversely homogeneous, in effect. In ad-
dition, anisotropy was incorporated as a key attribute of STFs. These features gave
rise to a mathematical framework for enabling precise and predictable engineering
of the polarization state of light, in addition to the widely practiced spectrum en-
gineering and bandwidth control. Much experimental effort has been subsequently
directed toward the realization of that goal. Therefore, nanoengineering of thin-
film morphology, if not consonant with the theoretical conceptualization of STFs,
lies outside the scope of this book.

1.2 From columnar to sculptured thin films

1.2.1 Columnar thin films

Chronologically as well as morphologically, it is sensible to begin with the so-

called columnar thin films (CTFs). The growth of oblique-angle CTFs by PVD
is usually credited to Kundt in 1886 [9]. At low temperature and pressure, ma-
terial in a source boat evaporates toward a substrate held at an angle to the in-
coming directional vapor flux, as shown in Fig. 1.3, and the arriving atoms set-
tle on it to form a thin film. Nucleation clusters 1 to 3 nm in diameter initially
form on the substrate. The clusters evolve into expanding and competing cone-
shaped columns as the film thickness increases, provided the film temperature is
maintained below about a third of its melting point [26]. This results in a fractal-
like array of surface features and an associated void network defining the fea-
tures.
The cross-section morphologies of a thin film of hydrogenated amorphous sil-
icon, a thick film of pyrolytic graphite, and an agate are shown in Fig. 1.4. All of

Downloaded From: http://ebooks.spiedigitallibrary.org/ on 09/25/2013 Terms of Use: http://spiedl.org/terms

 Overview 5

Figure 1.3 Schematic of the basic system for PVD of CTFs on planar substrates. Al-
though an electron-beam (e-beam) evaporation point source is shown, distributed directional
sources—such as those used in sputter deposition—can be used to similar effect.

Figure 1.4 (a) SEM micrograph of the cross section of a 25-µm-thick film of hydrogenated
amorphous silicon prepared by plasma-assisted, low-pressure, chemical vapor deposi-
tion. This device-quality film required decomposition of silane gas. (Reprinted from Ross
et al. [27] with permission of Elsevier.) (b) Optical micrograph of the cross section of a
45-mm-thick film of pyrolytic graphite (a commercially available material) prepared by va-
por-phase pyrolysis and deposition. (c) Cross section of a 50-mm-thick agate (Katanga, De-
mocratic Republic of Congo). The geological deposition mechanism for this mineral, though
similar to that for vapor-deposited thin films, contains the same basic physics of directional
deposition onto a surface with essentially no movement upon condensation. (The mineral
sample was provided by A. López Galindo, Universidad de Granada.)

Downloaded From: http://ebooks.spiedigitallibrary.org/ on 09/25/2013 Terms of Use: http://spiedl.org/terms

 6 Chapter 1

these exhibit the common cone-shaped, competitive growth evolution—which is

characteristic of the simply physical situation of directional serial growth onto a
surface in which the depositing unit (atoms for the hydrogenated amorphous sili-
con and pyrolytic graphite, and 300-nm spherical clusters for the agate) has limited
mobility upon condensation.
The addition of ion bombardment during evaporation can eliminate columns,
thereby yielding dense, smooth, and stable thin films that meet the stringent re-
quirements for laser-based applications of optical coatings.
Significantly, an intermediate state occurs between columnar expansion and
the elimination of the columns. In that state, competition between neighboring
columns is frustrated [28] in the early growth stages (
1 µm), and stable columns
grow thereafter with no signs of their initial conical shape. This CTF morphology
is achieved either through intermediate levels of ion bombardment or simply by de-
positing the films at oblique angles [7]. SEM micrographs of a series of amorphous
silicon CTFs are shown in Fig. 1.5. The steady-state, straight-sided columns thus
grow at a controllable angle χ ≥ 25 deg to the substrate, as shown schematically
in Fig. 1.6, while the average direction of the incident vapor flux is delineated by
the angle χv ≤ χ . These columns have to be regarded as nanowires, because their
linear cross-section dimensions are on the nanoscale.

Figure 1.5 SEM micrographs of four CTFs made of amorphous silicon by sputtering. The
column inclination angle χ , defined via Fig. 1.6, is indicated for all four CTFs. The column
inclination angles in these micrographs are close to 90 deg, but those for STFs are usually
lower in value.

Downloaded From: http://ebooks.spiedigitallibrary.org/ on 09/25/2013 Terms of Use: http://spiedl.org/terms

 Overview 7

Figure 1.6 Coordinate system, the vapor incidence angle χv , and the column inclination
angle χ .

1.2.2 Growth mechanics

Columnar thin films are the direct result of self-shadowing at the length scale of
the arriving adatoms (which are atoms on the growing film surface before they at-
tain their final free energy state). CTFs are fabricated under low-adatom-mobility
conditions—where the sticking coefficient for adatoms is essentially unity and
when the vapor arrives from a specific direction, either totally (e.g., in evapora-
tion from a point source at 10−5 Torr) or on average (e.g., in sputtering from a
planar source at 10−3 to 10−1 Torr). Only two issues need to be addressed in order
to understand the origin of CTFs:

(i) whether the arriving adatoms form a continuous, and thus uniform, net-
work or do they cluster; and
(ii) if they form clusters, then do the clusters grow independently or do they
compete for growth much like the leaves of a tree vying for sunlight?

Extensive research has revealed that clusters with associated honeycomblike

density deficit regions (commonly called void networks) are formed. Furthermore,
the clusters do compete for growth, leading to power-law-shaped columns that are
hierarchical cluster aggregates resembling a cauliflower when viewed from above
their top surface. This has been seen experimentally [29–31] and has also been
modeled via ballistic aggregation [32] and molecular dynamics [33] simulation
studies.
For deposition normal to the substrate surface (i.e., χv = 90 deg), the adatoms
arrive on rectilineal trajectories and form clusters 1- to 3-nm in size [26, 31]. The
sizes of these clusters have a statistical variation, and the larger clusters tend to
capture more of the arriving adatoms. But the clusters do not simply get larger;
rather, more clusters of similar size-distribution continue to form atop the clus-
ters formed earlier, and the clusters tend to aggregate hierarchically, which leads
to competitive columnar growth [26]. The edges of the dominant columns have a
capture radius that extends beyond the perimeter of the cluster/column boundary,

Downloaded From: http://ebooks.spiedigitallibrary.org/ on 09/25/2013 Terms of Use: http://spiedl.org/terms

 8 Chapter 1

thus making columnar expansion a possibility. This statistically favorable expan-

sion of the tallest columns leads to shadowing of the smaller adjoining columns.
As the top surface area is virtually fixed, at least one column must contract for
every column that expands. In the absence of any additional moderating mecha-
nism, the columns evolve eventually to yield a cauliflowerlike surface morphology
observable with conventional scanning electron microscopes for films greater than
∼ 1 µm thickness. This has been classified as the Zone 1 morphology in the well-
known structure zone model (SZM) [34, 35]. The column diameters expand ac-
cording to a power-law dependence in which the power-law exponent is a function
of the deposition conditions [29, 36].
The columns can become noncompetitive in their growth evolution [28]. These
parallel columnar structures constitute the so–called matchstick morphology, since
they resemble a bundle of wooden matchsticks when viewed from the top and the
sides [7, 37]—not unlike the parallel pins in the toy Round Pin Point™ in Fig. 1.7

Figure 1.7 The parallel steel pins of the toy Round Pin Point™ replicate the matchstick
morphology of Zone M quite well.

Figure 1.8 In contrast to the steel pins in the toy of Fig. 1.7, the primary and the secondary
trunks as well as the aerial roots of a banyan tree (Ficus indica) are irregularly arranged and
are, perhaps, closer to the morphology of CTFs.

Downloaded From: http://ebooks.spiedigitallibrary.org/ on 09/25/2013 Terms of Use: http://spiedl.org/terms

 Overview 9

as well as the trunks and the aerial roots of the banyan tree in Fig. 1.8. The recent
version [38] of the SZM, shown in Fig. 1.9, incorporates the matchstick morphol-
ogy in Zone M. This zone has been observed in sputtered thin films of amorphous
silicon, amorphous germanium, silicon carbide (SiC), and tungsten oxide (WO3 ),
where the moderating mechanism appears to be ion bombardment at normal inci-
dence to the substrate, near and above the sputtering threshold (∼ 30 to 50 eV for
most materials). Under such low-energy bombardment conditions, the most loosely
bound adatoms, at the edges of the clusters and columns, are very likely forward-
sputtered toward the underlying shadowed regions. Then as the top surfaces of the
favored columns become rougher and more convoluted with evolutionary develop-
ment, the increase in sputtering yield as a function of χv [39] drives the top surface
and column edges toward the steady-state morphology characteristic of Zone M.
Furthermore, the interior morphology of the individual matchsticks is expected to
resemble the noncompetitive morphology of the transition zone, termed Zone T by
Thornton [35].
Matchstick morphology also appears when films are deposited at oblique inci-
dence angles, especially when χv is small (i.e., at glancing angles). As χv increases,
the cluster density also increases due to decreased self-shadowing. In consequence,

Figure 1.9 A recent version of the SZM for CTFs [38]. The process variables are the re-
duced temperature T /Tm (where T is the substrate temperature and Tm is the melting point
of the bulk material), the argon pressure in the PVD chamber, and the ion bombardment
energy. This figure is based on one in Thornton’s highly cited publication [35] but modified
in order to include the matchstick morphology of Zone M that lies between Zones 1 and T.
Zones 2 and 3 are irrelevant for STFs, since high adatom mobility eliminates the effects of
self-shadowing and clustering. (Adapted from Messier and Trolier-McKinstry [38] with per-
mission of Elsevier.)

Downloaded From: http://ebooks.spiedigitallibrary.org/ on 09/25/2013 Terms of Use: http://spiedl.org/terms

 10 Chapter 1

the columns become less separated and, thus, more competitive for growth evolu-
tion. This has also been seen in ballistic aggregation simulations [7].
In an extensive review of both experimental and ballistic aggregation model-
ing studies of obliquely deposited CTFs, van Kranenburg and Lodder [40] con-
cluded that elongated clusters and columns generally pointing in the direction of
the incoming vapor flux are a direct consequence of the adatomic self-shadowing
process; furthermore, when viewed from directly overhead, the length of the long
axis relative to the width of the cluster increases markedly for χv < 30 deg. The
columns become separated and begin to grow as noncompeting cylinders—with
elliptical cross-sections due to anisotropy in self-shadowing [7, 40]—as χv is de-
creased further. The columns become more separated in the vapor incidence di-
rection due to the increased shadowing effect in the longitudinal direction (paral-
lel to the vapor incidence plane), while shadowing in the transverse direction is
unaffected by changes in χv . This leads to a higher average mass density in the
transverse direction (referred to as “bundling”), and to a density anisotropy that
has been exploited commercially in magnetic thin films [40].

1.2.3 CTFs as dielectric continuums

As the columnar cross-section radius is less than or equal to 150 nm for a large
variety of CTFs, these films can be considered effectively as homogeneous or-
thorhombic continuums in the visible and infrared regimes, depending on the con-
stitutive parameters of the deposited material [41]. Generally thought of as dielec-
tric materials, their optical birefringence has long been appreciated and exploited
[42, 43].
If the plane z = 0 is designated as a reference plane parallel to the substrate,
three unit vectors,2 viz.,

uτ = ux cos χ + uz sin χ , (1.1)

un = −ux sin χ + uz cos χ , (1.2)
ub = −uy , (1.3)

assist in the macroscopic electromagnetic characterization of both CTFs and STFs,

with ux , uy , and uz as the three Cartesian unit vectors. The constitutive relations
of a CTF may be set up as functions of position vector r = xux + yuy + zuz and
angular frequency ω as

D(r, ω) = 0 = · E(r, ω), (1.4)

r
B(r, ω) = µ0 H(r, ω). (1.5)
2 Vectors are in boldface; dyadics are in normal face and double-underlined; column vectors and ma-
trixes are in boldface and enclosed within square brackets. Appendix A provides a simple introduc-
tion to dyadics and dyads. See Chapter 1 of Chen’s book [50] for a simple, useful, and comprehensive
treatment of these mathematical entities.

Downloaded From: http://ebooks.spiedigitallibrary.org/ on 09/25/2013 Terms of Use: http://spiedl.org/terms

 Overview 11

In these equations as well as hereafter, 0 = 8.854 × 10−12 F m−1 and µ0 = 4π ×

10−7 H m−1 are the permittivity and the permeability of free space (i.e., vacuum),
respectively, and an exp(−iωt) dependence on time t is implicit.
Extensive experimental research on CTFs [44–49] necessitates the orthorhom-
bic prescription

(ω) = a (ω)un un + b (ω) uτ uτ + c (ω)ub ub (1.6)

r

as the relative permittivity dyadic. In general, is biaxial, with the reduction

r
c = a appropriate for uniaxiality. Optical measurements indicate that some CTFs
can be uniaxial [45, 46], but biaxiality is probably the norm [47–49]. Biaxiality
may arise from the columnar cross-sections being noncircular, as discussed in
Sec. 1.2.2, and can have significant optical consequences relative to uniaxiality
for CTFs as well as STFs [51–53].
Increased adatomic self-shadowing with decrease in χv creates a density aniso-
tropy. This implies that not only should have the orthorhombic form (1.6), but
r
also that a , b , and c be dependent on the column inclination angle χ . A series
of optical characterization experiments on CTFs of the oxides of tantalum, tita-
nium, and zirconium [49], grown by thermal evaporation, suggests the following
parametric relationships:

2
a (ω, χv ) = a0 (ω) + a1 (ω)v + a2 (ω)v 2 , (1.7)


2 2
b (ω, χv ) = b0 (ω) + b1 (ω)v + b2 (ω)v , (1.8)


2 2
c (ω, χv ) = c0 (ω) + c1 (ω)v + c2 (ω)v , (1.9)
tan χ = m̄ tan χv . (1.10)

Here, the coefficient m̄ > 1 and the functions a0 (ω), etc., vary with the deposited
material species and the deposition conditions, while v = χv /(π/2) is the vapor
incidence angle expressed as a fraction of a right angle. For vapor incidence perp-
endicular to the substrate, v = 1 and the columnar cross-section is circular; hence,
a0 + a1 + a2 ≡ c0 + c1 + c2 . Equations (1.7)–(1.10) should hold at frequencies
substantially distant from material resonances [54, Ch. 13]. Other parametric rela-
tionships have also been derived from experimental data [55, 56].
In order to enhance biaxiality and linear birefringence, Hodgkinson and Wu
[57] modified the standard PVD technique by using two vapor fluxes, both from
the same source boat. In their serial bideposition technique, both vapor fluxes are
directed in the xz plane, but one flux is directed at an angle χv and the other at
π − χv . Furthermore, the vapor fluxes are switched on and off alternately, so that
each flux is responsible for the deposition of ∼ 3 nm thickness only. The resulting
columns are upright; that is, χ = 90 deg. Uniaxiality should arise, however, if one
of the two vapor fluxes is directed in the xz plane and the other in the yz plane [58].
Also, two source boats can be used to deposit different material species, either
concurrently or alternately [16].

Downloaded From: http://ebooks.spiedigitallibrary.org/ on 09/25/2013 Terms of Use: http://spiedl.org/terms

 12 Chapter 1

1.2.4 Primitive STFs with nematic morphology

A seminal event occurred in 1966 that eventually led to the emergence of the STF
concept in 1994 [2]. While a CTF was growing, Nieuwenhuizen and Haanstra de-
liberately altered χv to prove that columnar morphology “cannot be a result of the
method of preparation itself” [15]. The resulting change in χ was accomplished
while the film thickness grew ∼ 3 nm, the transition being practically abrupt in the
visible and infrared regimes. Some two decades later, Motohiro and Taga demon-
strated that χ can be abruptly altered easily during growth [16], which was con-
firmed a few years later [59]. This ability is the basis for realizing STFs with bent
nematic morphologies.
Thus, primitive STFs with zigzag and chevronic morphologies came into ex-
istence. The similarity of CTFs to crystals had long been noticed in the optical
literature [42], so that the primitive STFs with nematic morphology can be consid-
ered as stacked crystalline plates. This has been astutely exploited for designing,
fabricating, and testing various optical devices [43, 60]. Furthermore, serial as well
as simultaneous bideposition of CTFs and chevronic STFs are now routine in the
manufacture of wave plates for the automobile industry [61–63].

1.2.5 Chiral STFs

Another seminal event toward the emergence of the STF concept had already oc-
curred in 1959. Although that event went largely unnoticed, all credit for peri-
odic STFs with chiral (i.e., handed) morphology should be accorded to Young and
Kowal [12]. Without actually seeing the anisotropic morphology of CTFs via scan-
ning electron microscopy or otherwise, these two pioneers consciously [64] ro-
tated the substrate about the z axis constantly during growth to create thin films
of calcium fluoride with morphology predicted to display transmission optical ac-
tivity. Most likely, they were the first researchers to deliberately engineer thin-film
morphology for producing a nontrivial STF—one with a fully three-dimensional
morphology.
The concept of voids in thin films was to become common 12 years later. Re-
markably, however, Young and Kowal actually used that word to conjecture that
“the [optical] activity of a helically deposited film could be due to the co-operative
action of a helically symmetrical arrangement of crystallites, crystal growth or
voids.” Furthermore, they conjectured that the columnar direction could change
virtually instantaneously and continuously with changes in the position and the
orientation of the substrate. Happily, the Young–Kowal technique of rotating the
substrate, the helicoidal morphology realized thereby, and the transmission optical
activity of chiral STFs, were rediscovered in the last decade [65–67].

1.2.6 Sculptured thin films

Recognition finally came, during the 1990s, that

Downloaded From: http://ebooks.spiedigitallibrary.org/ on 09/25/2013 Terms of Use: http://spiedl.org/terms

 Overview 13

(i) a very wide variety of columnar morphologies is possible through control

of two fundamental axes of substrate rotation;
(ii) multisections with different columnar morphologies and/or different ma-
terials are possible;
(iii) a wide range of potential applications exists; and
(iv) preparation–property–application connections can be truly engineered by
coupling theoretical and experimental results.

The concept of STFs then emerged in short order [2, 3].

STFs are modifications of CTFs in which the column direction can be changed
almost abruptly and often, even continuously, during growth. When CTFs are ob-
liquely deposited, a wide variety of STF morphologies tailored at the nanoscale
are realizable by simple variations of two fundamental axes of rotation, either sep-
arately or concurrently [6, 37, 65, 68–75]. These fundamental axes lead to two
canonical classes of STFs that have been termed

(i) sculptured nematic thin films (SNTFs) [6], and

(ii) thin-film helicoidal bianisotropic mediums (TFHBMs) [2, 65].

More complex shapes and even multisections, in which either the material or the
shape or both are changed from section to section along the z axis, have been
executed [76, 77].
SNTF morphologies include such simple two-dimensional shapes as slanted
columns, chevrons, and zigzags as well as the more complex C- and S-shapes; see
Fig. 1.1. The substrate has to be rotated about the y axis, which lies in the substrate
plane and is perpendicular to the vapor incidence direction, while χv is varied
either episodically or continuously [6]. One concern with this approach is related
to the fact that the density of a CTF is highly dependent upon χv [5, 6, 70] and,
therefore, density variations arise as an SNTF grows. As discussed in Sec. 8.2.3,
the compensation of these variations is an area of future research.
TFHBMs are fabricated by tilting the substrate at some oblique angle to the in-
cident vapor flux (i.e., χv < 90 deg), followed by substrate rotation about the z axis.
Helicoidal morphologies result for constant rotational velocity about the z axis
[12, 65]. By varying the rotational velocity in some prescribed manner throughout
a rotational cycle, a slanted helicoidal structure occurs with the slant angle con-
trollable over all χ above its minimum value for static glancing angle deposition.
Furthermore, by making the rotational cycle + 360 deg, where = 0 deg is
some chosen phase-shift angle, it is possible to engineer a wide range of superhe-
lixes with controlled handedness [5, 37]. The mass density as a function of film
thickness is expected to remain constant since χv is fixed for TFHBMs, so long
as the columns attain a steady-state diameter in the early nucleation and growth
stages.
However, as seen in Fig. 1.2, the column width of the superhelix increases
with growth evolution, which begs the scientific question: Why? In many reports

Downloaded From: http://ebooks.spiedigitallibrary.org/ on 09/25/2013 Terms of Use: http://spiedl.org/terms

 14 Chapter 1

on STFs, the columns shown by cross-section scanning electron microscopy usu-

ally appear parallel with cross-section diameters ranging between 10 and 300 nm
and with a similar range in separation distances. Since most optical applications
of STFs will require that the column diameter, the separation distance between
columns, and the connectivity between columns all remain constant during growth,
this steady-state column size condition has practical advantages. For instance,
when an STF deposition cycle is repeated for a multiple-period TFHBM prepared
by both evaporation and sputtering at constant angular velocities of substrate ro-
tation, it is important that the last turn has the same pitch and porosity as all pre-
vious turns. This morphological reproducibility leads to, for instance, predictable
optical activity spectrums [67, 73], since the column sizes and shapes allow the
STFs to effectively act as rotationally nonhomogeneous continuums at optical fre-
quencies [4]. This can only occur if the films evolve from the initial clusters into
steady-state, noncompetitive, parallel columns within a small fraction of the total
film thickness, and constant cross-section as well as constant porosity is maintained
for the remainder of the film growth.

1.3 Time-harmonic electromagnetic fields

Although nonoptical applications of STFs are certainly possible [3, 78, 79], opti-
cal applications have been the chief drivers of STF research [17, 18, 77, 80, 81].
Moreover, optical monitoring of STF growth appears necessary in a production
setting. Therefore, the linear electromagnetic properties—mostly for application
in the visible and the infrared regimes—of STFs are recounted in this section.
1.3.1 Linear constitutive relations
By definition, the morphology of a single-section STF in any plane z = z1 can be
made to coincide with the morphology in another plane z = z2 with the help of a
suitable rotation. In other words, the local morphology is spatially uniform, but the
global morphology is unidirectionally nonhomogeneous. Naturally, this leads to
the concept of local constitutive properties of an STF. The global constitutive prop-
erties of an STF can be connected to the local ones by means of rotation dyadics.
The most general linear medium is bianisotropic; i.e., both D(r, ω) and B(r, ω)
are linear functions of both E(r, ω) and H(r, ω), in a direction-dependent man-
ner [82]. Accordingly, the linear frequency-domain (or time-harmonic) constitutive
relations of a single-section STF are set up as

D(r, ω) = 0 S(z) · (ω) · S T (z) · E(r, ω)
ref

+ α (ω) · S (z) · H(r, ω) ,
T
(1.11)
ref

B(r, ω) = µ0 S(z) · β (ω) · S T (z) · E(r, ω)
ref

+ µ (ω) · S (z) · H(r, ω) ,
T
(1.12)
ref

Downloaded From: http://ebooks.spiedigitallibrary.org/ on 09/25/2013 Terms of Use: http://spiedl.org/terms

 Overview 15

where the superscript T denotes the transpose. These equations model the STF as a
bianisotropic continuum. Whereas the relative permittivity dyadic (ω) and the
ref
relative permeability dyadic µ (ω) represent the reference (or local) electric and
ref
magnetic properties, respectively, the dyadics α (ω) and β (ω) delineate the
ref ref
reference magnetoelectric properties. All four of these constitutive dyadics have
the same form as the right side of Eq. (1.6). Gyrotropic terms (of the type v × I ,
where v is some vector and I is the identity dyadic) can be added to any or all
of the four reference constitutive dyadics [83]. For a purely dielectric STF, both
α (ω) and β (ω) are equal to the null dyadic, and µ (ω) is equal to the
ref ref ref
identity dyadic.
Nominally, all columns in an STF twist and bend identically as z changes, a
feature that is captured by the rotation dyadic S(z). This dyadic is some composi-
tion of the following three elementary rotation dyadics:

S (z) = ux ux + (uy uy + uz uz ) cos ξ(z)

x
+ (uz uy − uy uz ) sin ξ(z), (1.13)
S (z) = uy uy + (ux ux + uz uz ) cos τ (z)
y
+ (uz ux − ux uz ) sin τ (z), (1.14)
S (z) = uz uz + (ux ux + uy uy ) cos ζ(z)
z
+ (uy ux − ux uy ) sin ζ(z). (1.15)

The angular functions of z in these equations may be specified piecewise: either

S(z) = S (z) or S(z) = S (z) for SNTFs, while S(z) = S (z) for TFHBMs. Al-
x y z
though TFHBMs need not be periodically nonhomogeneous along the z axis, it is
easy to fabricate them with periods chosen anywhere between 50 and 2000 nm.
Chiral STFs are generally analyzed as periodic dielectric TFHBMs with ζ(z) =
π z/ , with 2 as the structural period [84].
Multisection STFs are cascades of single-section STFs fabricated in an inte-
grated manner [3]. The five dyadics S(z), (ω), µ (ω), α (ω), and β (ω)
ref ref ref ref
are defined differently for each section, the transitions between the sections being
virtually abrupt and, therefore, optically inconsequential [76, 77].

1.3.2 Electromagnetic wave propagation

The foregoing section makes it clear that an STF is transversely homogeneous (at
least in the wavelength regimes of interest) but nonhomogeneous along the z axis.
Therefore, electromagnetic wave propagation in an STF is best handled using 4×4
matrixes and column vectors of size 4. At any given frequency, with the transverse
wavenumber κ and the angle ψ fixed by excitation conditions, the following spatial

Downloaded From: http://ebooks.spiedigitallibrary.org/ on 09/25/2013 Terms of Use: http://spiedl.org/terms

 16 Chapter 1

Fourier representation of the electric and the magnetic field phasors is useful:


E(r, ω) = e(z, κ, ψ, ω) exp iκ(x cos ψ + y sin ψ)

 . (1.16)
H(r, ω) = h(z, κ, ψ, ω) exp iκ(x cos ψ + y sin ψ)

Substitution of this representation into the source-free Maxwell curl postulates

∇ × E(r, ω) = iωB(r, ω), (1.17)

∇ × H(r, ω) = −iωD(r, ω), (1.18)

followed by the use of the constitutive relations (1.11) and (1.12), leads to four or-
dinary differential equations and two algebraic equations. The phasor components
ez (z, κ, ψ, ω) and hz (z, κ, ψ, ω) are then eliminated to obtain the 4×4 matrix or-
dinary differential equation (MODE) [83]
d
[f(z, κ, ψ, ω)] = i[P(z, κ, ψ, ω)] [f(z, κ, ψ, ω)]. (1.19)
dz
In this equation,
 
ex (z, κ, ψ, ω)
 ey (z, κ, ψ, ω) 
[f(z, κ, ψ, ω)] =  
 hx (z, κ, ψ, ω)  (1.20)
hy (z, κ, ψ, ω)

is a column vector, and [P(z, κ, ψ, ω)] is a 4×4 matrix function of z that can be eas-
ily obtained using symbolic manipulation programs. The 4×4 system can reduce
in certain circumstances to two autonomous 2×2 systems—as for propagation in
the morphologically significant planes of single-section SNTFs [85].
An analytic solution of Eq. (1.19) can be obtained, provided [P(z, κ, ψ, ω)] is
not a function of z, i.e., when [P(z, κ, ψ, ω)] = [Pcon (κ, ψ, ω)]. This happens, of
course, for CTFs, and the solution procedure is described by Hochstadt [86]. The
exact analytic solution of Eq. (1.19) also has been obtained for axial propagation
(i.e., κ = 0) in periodic TFHBMs and chiral STFs [25, 84, 87]. A solution in terms
of a convergent matrix polynomial series is available for nonaxial propagation (i.e.,
κ = 0) in periodic TFHBMs [88–92].
More generally, only a numerical solution of Eq. (1.19) can be obtained. If the
matrix [P(z, κ, ψ, ω)] is a periodic function of z, a perturbative approach [93] can
be used to obtain simple results for weakly anisotropic STFs [88, 94]; coupled-
wave methods may come in handy otherwise [95–98].
But if [P(z, κ, ψ, ω)] is not periodic, the constitutive dyadics can be assumed
as piecewise constant—i.e., constant over slices of thickness z—and the approx-
imate transfer equation [83]

 z

f(z + z, κ, ψ, ω) exp i[P(z + , κ, ψ, ω)] z f(z, κ, ψ, ω) (1.21)
2

Downloaded From: http://ebooks.spiedigitallibrary.org/ on 09/25/2013 Terms of Use: http://spiedl.org/terms

 Overview 17

can be suitably manipulated with appropriately small values of z. This numerical

technique has been applied to chiral STFs [89, 98].
Finally, quasistatic solutions of Eq. (1.19) can be obtained in the same ways,
after taking the limit ω → 0 ab initio [99]. These are useful if applications of
STFs in the microwave and lower-frequency regimes are desired—for examples, as
interlayer dielectrics in integrated electronic circuits [100, 101] and for humidity
sensors that rely on capacitance change induced by altered humidity [102].

1.3.3 Structure–property relationships

Howsoever it is carried out, the solution of the matrix ordinary differential Eq.
(1.19) paves the way for predicting electromagnetic phenomenons as well as for
designing STF devices that are compatible with planar technology. But Eqs. (1.11)
and (1.12) incorporate the assumption of an STF as a unidirectionally nonhomoge-
neous continuum. This is valid in a macroscopic sense, i.e., in the visible and the
infrared regimes, because STFs with appropriate morphological length scales can
be fabricated. However, the relationship of the nanostructure to the macroscopic
properties must be understood—or at least modeled—for intelligent design and
fabrication of STF devices.
The concept of local homogenization has been used to devise a nominal model
[103]. The deposited material as well as the void regions are nominally thought
of as parallel ellipsoids in any xy plane. The Bruggeman formalism [41] is then
used to estimate (ω), etc., in terms of the shape factors of the ellipsoids, the
ref
bulk constitutive properties of the deposited material, and the porosity of the STF.
The possibility of the void regions being infiltrated by some material can also be
handled [104, 105]. Furthermore, this model can incorporate the frequency depen-
dence of the constitutive properties of the deposited materials [106].
The devised model has been used [106–108] to comprehensively study the
plane-wave responses of dispersive chiral STFs on axial and nonaxial excitation,
with focus on the spectrums of various measures of transmission optical activity.
The dependencies of these quantities on the column inclination angle, periodicity,
porosity, and two ellipsoidal shape factors have been deduced. When fully cou-
pled with Eq. (1.6) as well as those such as Eqs. (1.7)–(1.10), and after calibration
against experimentally obtained reflectance/transmittance data [109], the nominal
model may turn out to be a powerful design tool and process-control paradigm. It
has already been applied to assess the piezoelectric tunability of lasers and optical
filters made of chiral STFs [110, 111].

1.4 Optical applications of STFs

Although optical, electronic, acoustic, thermal, chemical, and biological applica-
tions of STFs were forecast early on [3], the potential of these nanostructured ma-
terials has been most successfully actualized in linear optics thus far. Several types
of optical filters, sensors, and electrically addressable displays are in various stages
of development, but are definitely past their embryonic stages.

Downloaded From: http://ebooks.spiedigitallibrary.org/ on 09/25/2013 Terms of Use: http://spiedl.org/terms

 18 Chapter 1

1.4.1 Optical filters

Chiral STFs display the circular Bragg phenomenon in accordance with their peri-
odic nonhomogeneity along the z axis [89]. Briefly, a structurally right/left-handed
chiral STF only a few periods thick almost completely reflects normally incident
right/left circularly polarized (RCP/LCP) plane waves with wavelength lying in
the so-called Bragg regime; while the reflection of normally incident LCP/RCP
plane waves in the same regime is very little. Figure 1.10 presents the measured
and the predicted transmittance spectrums of a structurally left-handed chiral STF
made of titanium oxide, showing the almost complete blockage of an incident LCP
plane wave and the high transmission of an incident RCP plane wave at free-space
wavelengths in the neighborhood of 620 nm.
The bandwidth of the Bragg regime and the peak reflectivity therein first in-
crease with the thickness of the chiral STF, and then saturate. Once this saturation
has occurred, further thickening of the film has negligible effects on the reflection
spectrum. The Bragg regime is also marked by high levels of optical activity [17,
66, 73], which, however, does not scale with the film thickness and is also highly
dependent on the orientation of the incident electric field phasor [113].

Figure 1.10 Measured and predicted transmittance spectrums of a structurally left-handed

chiral STF for normal incidence (κ = 0). The transmittance TLR is the intensity of the LCP
component of the transmitted plane wave relative to the intensity of the RCP component
of the incident plane wave, etc. Dispersion was not taken into account when predicting the
transmittances from the solution of Eq. (1.19). (Adapted from Wu et al. [112] with permis-
sion.)

Downloaded From: http://ebooks.spiedigitallibrary.org/ on 09/25/2013 Terms of Use: http://spiedl.org/terms

 Overview 19

More than one Bragg regime is possible when a plane wave is obliquely inci-
dent (i.e., κ = 0) [89], but it is the normal-incidence case that appears to be of the
greatest value in the context of planar technology. The major successes reported
are as follows:

• Circular polarization filters. The circular Bragg phenomenon can be em-

ployed to realize circular polarization filters. A normally incident, circularly
polarized plane wave of one handedness can be reflected almost completely,
while that of the other handedness is substantially transmitted, if absorption
is small enough and the film is sufficiently thick, in the Bragg regime. This
has been experimentally demonstrated [112]. As of now, the Bragg regime
can be positioned at virtually any free-space wavelength between 450 and
1700 nm.
Calculations reveal that polarization insensitivity, for application in laser
mirrors, can be realized with a cascade of two otherwise identical chiral
STFs but of opposite structural handedness [114, 115]. Furthermore, step-
wise chirping can widen the bandwidth [116, 117], and tightly interlaced
chiral STFs may be attractive for bandwidth engineering [118]. Finally, dis-
persive characteristics can allow more than one Bragg regime [119], as ex-
emplified by the calculated reflectance spectrums shown in Fig. 1.11.

Figure 1.11 Calculated reflected spectrums of a structurally right-handed chiral STF

half-space for normal incidence (κ = 0). The reflectance RLR is the intensity of the LCP
component of the reflected plane wave relative to the intensity of the RCP component of
the incident plane wave, etc. Dispersion is responsible for the circular-polarization-sensitive
Bragg regimes centered at 147- and 349-nm wavelengths. (Adapted from Wang et al. [119]
with permission of Elsevier.)

Downloaded From: http://ebooks.spiedigitallibrary.org/ on 09/25/2013 Terms of Use: http://spiedl.org/terms

 20 Chapter 1

A handedness inverter for light of only one of the two circular polariza-
tion states was designed [120], and then fabricated as well as tested [121].
As the first reported two-section STF device, it comprises a chiral STF and
a CTF functioning as a half-wave plate. Basically, it almost completely re-
flects, say, LCP light, while it substantially transmits incident RCP light after
transforming it into LCP light, in the Bragg regime.
• Spectral hole filters. A two-section STF was proposed as a spectral hole filter
[122]. Both sections are chiral STFs of the same structural handedness and
identical thickness L. Their structural periods 21 and 22 are chosen such
that 2L(−1 −1
2 − 1 ) = 1. A narrow transmission band or passband then
appears for circular polarized plane waves of the same handedness as the
two chiral STF sections.
A more robust three-section STF was also proposed as a spectral reflec-
tion hole filter. Its first and third sections are identical chiral STFs, while
the thin middle section is a homogeneous layer that acts like a phase defect
[123, 124]. This design was implemented to obtain a 11-nm-wide spectral
hole centered at a free-space wavelength of 580 nm in the reflectance spec-
trum [125].
An even better design became available shortly thereafter, wherein the
middle layer was eliminated, but the lower chiral STF was twisted by 90 deg
with respect to the upper chiral STF about the z axis. The twist performed
satisfactorily as the required phase defect [76]. With much thicker chiral
STFs on either side of the phase defect, calculations revealed that ultranarrow
spectral holes (
0.1 nm bandwidth) can be obtained in the transmittance
spectrum [126, 127]; but their performance may be impaired by attenuation
within the thick sections. Cascades of phase defects of different types are
expected to deliver more robust performances in comparison to single defects
[128].

Most recently, slanted chiral STFs have been introduced [98] to couple the
circular Bragg phenomenon to the Rayleigh-Wood anomalies exhibited by surface-
relief gratings [129]. This coupling occurs when the helicoidal axis is inclined with
respect to the z axis, and suggests the use of these new types of STFs as narrowband
circular polarization beamsplitters.
SNTFs can also be pressed into service as optical filters—for linearly polarized
plane waves. Rugate filters have been realized as piecewise uniform SNTFs to
function as narrowband reflectors [18]. S̆olc filters of the fan and the folded types
are also possible with the same technology [43, 130]. The major issue for further
research and development is the control of mass density, and hence (ω) with
ref
χv when fabricating continuously nonhomogeneous SNTFs.
The future of multisection STF devices in optics appears bright because of a
recent feat of Suzuki and Taga [77], who were able to deposit a cascade of six
different sections of combined thickness ∼ 2 µm.

Downloaded From: http://ebooks.spiedigitallibrary.org/ on 09/25/2013 Terms of Use: http://spiedl.org/terms

 Overview 21

1.4.2 Optical fluid sensors

The porosity of STFs makes them attractive for fluid-concentration-sensing appli-

cations [131, 132], because their optical response properties must change in ac-
cordance with the number density of infiltrant molecules. In particular, theoretical
research has shown that the Bragg regime of a chiral STF must shift accordingly,
thereby providing a measure of the fluid concentration [131]. Qualitative support
for this finding is provided by experiments on wet and dry chiral STFs [133].
Furthermore, STF spectral hole filters can function as highly sensitive fluid
concentration sensors. Proof-of-concept experiments with both circularly polarized
and unpolarized incident light have confirmed the redshift of spectral holes upon
exposure to moisture [105].

1.4.3 Displays

Liquid crystals (LCs) can be electrically addressed and are therefore widely used
these days for displays [134, 135]. Although STFs are not electronically address-
able, the alignment of nematic LCs forced into the void regions of chiral STFs
has been shown to respond to applied voltages [136]. Thus, STF–LC composite
materials may have a future as robust displays.
Another interesting possibility, in the same vein, is to grow carbon (and other)
nanotubes by chemical reactions involving fluid catalysts and precursors [137] in-
side highly porous STFs. The growing nanotubes would have to conform to the
structure imposed by the STF skeleton, and the nanotube–STF composite material
thus formed could be useful for field emission devices.

1.4.4 Optical interconnects

STF technology is compatible with the planar technology of electronic chips. Chi-
ral STFs have the potential to simultaneously guide waves with different phase
velocities in different directions [138, 139], and could therefore function as op-
tical interconnects, leading to efficient use of the available “real estate” in elec-
tronic chips. Furthermore, the helicoidal structure of chiral STFs would resist ver-
tical cleavage and fracture. Simultaneous microrefrigeration enabled by the porous
STFs would be a bonus.

1.4.5 Optical pulse-shapers

The current explosive growth of digital optics communication has provided the
impetus for time-domain research on novel materials. As chiral STFs are very at-
tractive for optical applications, the circular Bragg phenomenon is being studied in
the time domain. A pulse-bleeding phenomenon has been identified as the under-
lying mechanism, which can drastically affect the shapes, amplitudes, and spectral
components of femtosecond pulses [140]. However, narrowband rectangular pulses
can pass through without significant loss of information [141]. The application of
STFs to shape optical pulses appears to be waiting in the wings.

Downloaded From: http://ebooks.spiedigitallibrary.org/ on 09/25/2013 Terms of Use: http://spiedl.org/terms

 22 Chapter 1

1.4.6 Biochips and biosensors

Endowed with porosity of nanoengineered texture, STFs can function as microre-

actors for luminescence-producing reactions involving biochemicals. Biolumines-
cent emission is bound to be affected by the reactor characteristics. If the reactor is
a chiral STF, its helicoidal periodicity can be exploited. The structural handedness
as well as the periodicity of chiral STFs have been theoretically shown to criti-
cally control the emission spectrum and intensity, while the polarization state of
the emitted light is strongly correlated with the structural handedness of the em-
bedded source filaments [142]. Optimization with respect to χv appears possible
[143, 144].
STFs can function as biosensors in other modalities as well. For instance,
groundwater reservoirs and flooded mines are known to harbor anerobic bacte-
ria that reduce metals [145, 146]. An optical fiber with its tip coated by a chiral
STF made of iron oxide could be introduced in an anerobic environment, wherein
the metal-reducing bacteria (if any) would reduce the chiral STF gradually. Prelim-
inary calculations indicate that spectral shifts of the optical signature of the chiral
STF would result [147]. Such shifts could be monitored above ground to estimate
the bacterial concentration.

1.4.7 STFs with transverse architectures

An exciting new development is the growth of STFs with transverse architectures

on micropatterned substrates [148], as exemplified by Fig. 1.12. Photoresist lines
and other features are initially deposited on the planar surface of a substrate, fol-
lowed by the PVD of an STF. In the resulting film, the microscale features in the
transverse plane are blended with nanoscale features in the growth direction. Such
films can be homogeneous over ∼ 500-nm pieces in the transverse plane. The over-
all appearance may be likened, somewhat roughly, to an arrangement of vertical
cores of one or more STFs placed on a lattice. Because complex periodic features
can be etched, the new development shall give a fillip to photonic bandgap engi-
neering.

Figure 1.12 SEM micrograph of an STF of silicon oxide deposited on a lithographically

patterned substrate. (This micrograph was provided by Mark W. Horn, Pennsylvania State
University.)

Downloaded From: http://ebooks.spiedigitallibrary.org/ on 09/25/2013 Terms of Use: http://spiedl.org/terms

 Overview 23

Remarks on STFs with transverse architectures in relation to photonic crystals

are in order: Chiral STFs have been grown on regular lattices by lithographically
patterning the substrates [74, 149]. Whereas slow rotation of the substrate results
in the growth of arrays of nano- or micro-helixes spaced as close as 20 nm from
their nearest neighbors, faster rotation yields arrays of increasingly denser pillars
[37]. When the spacing between the helixes is comparable to the wavelength, such
a film can function as a photonic crystal [150, 151]. Many such films have been
deposited with interesting characteristics in the visible and the infrared regimes
[152, 153], and the possibility of fabricating them on cheap polymeric substrates
is very attractive [154]. However, such films are not STFs as they effectively lack
transverse homogeneity at the scale of the wavelengths of interest because the inter-
columnar spacing is not sufficiently small.

1.5 Other applications

From their inception [3], STFs were expected to have a wide range of applications,
implementable after their properties came to be better understood. Their optical ap-
plications came to be investigated first, as detailed in Sec. 1.4. However, their high
porosity—in combination with optical anisotropy and possible two-dimensional
electron confinement in the nanostructure—makes STFs potential candidates also
as

(i) electroluminescent devices;

(ii) high-speed, high-efficiency electrochromic films;
(iii) optically transparent conducting films sculptured from pure metals; and
(iv) multistate electronic switches based on filamentary conduction.

That same porosity can be harnessed in microreactors and thermal barriers, as it is

accompanied by high surface area [78, 79, 155]. For the same reason, STFs may be
useful as nanosieves and microsieves for the entrapment of viruses or for growing
biological tissues on surfaces of diverse provenances. The potential of STFs as
biosubstrates is bolstered by many reports on altered adsorption of proteins and
cells on nanopatterned surfaces [156, 157].
These applications of STFs are still in their incipient stages, but some advances
have been made on the following two fronts:

• Interlayer dielectrics. With the microelectronics industry moving relent-

lessly towards decreasing feature sizes and increasingly stringent tolerance
levels, an urgent need exists for the use of low-permittivity materials as in-
terlayer dielectrics. Silicon dioxide, the current material of choice, has too
high a quasistatic permittivity. The porosity of STFs and nanoporous sil-
ica makes them attractive low-permittivity materials for microelectronic and
electronic packaging applications [101]. However, chiral STFs are likely to
have significant thermal and mechanical as well as electrical advantages
over nanoporous silica—because of (i) porosity with controllable texture,

Downloaded From: http://ebooks.spiedigitallibrary.org/ on 09/25/2013 Terms of Use: http://spiedl.org/terms

 24 Chapter 1

and (ii) helicoidal morphology. Also, STFs can be impregnated with various
kinds of polymers.
• Ultrasonic applications. The sciences of electromagnetics and elastody-
namics have an underlying mathematical unity. For that reason, all optical
applications described thus far in Sec. 1.4 possess ultrasonic analogs. In-
deed, ultrasonic wave propagation in chiral STFs is now theoretically well-
established [158–160], as also is the potential for its applications [161, 162].
Actual implementation would, however, require the fabrication of chiral
STFs with periods ∼ 20 µm [163], the development of which is still awaited.

1.6 Prognostications

Several of the emerging applications mentioned in Secs. 1.4 and 1.5 are barely past
conceptualization. Considerable research on them is warranted, before they be-
come commercially viable. Just a few of the accomplishments discussed in Sec. 1.4
have crossed the threshold of academic research, and now require several issues to
be addressed.
A key issue is that of environmental stability of STFs. The chemical stability of
STFs has not been examined in any detail yet, although the susceptibility of porous
thin films to moisture is known [133]. An indentation experiment on a chiral STF
[164] as well as the successful deposition of six-section STFs [77] strongly indicate
that mechanical stability must be investigated in depth. However, only a prelimi-
nary model for the mechanical loading of STFs exists at this time [165, 166]. Due
to porosity, internal stresses and the morphological stability of STFs in the absence
of external loads have to be examined carefully as well [167, 168].
Another key issue is that of efficiency. The vapor incidence angle χv , the
bulk constitutive properties of the deposited material (responsible, e.g., for a ,
b , and c ), and the substrate rotation parameters appearing in S(z) must be op-
timized to achieve desired performance characteristics. As examples, the photoca-
talytic efficiency of chiral STFs of tantalum oxide is known to be optimal when
χv = 20 deg [155], efficient bioluminiscent emission has been shown to require
χv
15 deg [143], while χv could be manipulated in order to maximize the band-
width of a Bragg regime [144]. A study on second-harmonic generation in uniax-
ial chiral STFs has underscored the criticality of χ (and therefore of χv ) for effi-
ciency [169]. Cross-polarized remittances are drastically reduced and the diversity
in the co-polarized remittances is enhanced by the incorporation of index-matched
layers at the entry and the exit pupils of circular polarization filters [112]. Further
improvements may require the simultaneous deposition of different types of mate-
rials in order to reduce absorption and dispersion in desired wavelength regimes in
optical filters based on the STF concept.
Nonlinear optics with STFs is practically virgin territory, despite two reported
forays into second-harmonic generation [169, 170]. Due to the numerous classes of
nonlinearity [171], the delineation of nanocrystallinity in STFs will be of primary

Downloaded From: http://ebooks.spiedigitallibrary.org/ on 09/25/2013 Terms of Use: http://spiedl.org/terms

 Overview 25

Figure 1.13 SEM micrograph of a chiral STF of parylene. (This micrograph was provided
by Mark W. Horn, Pennsylvania State University.)

importance. Likewise, understanding of nanodomains in magnetic STFs, as well as

of magnetoelectric effects in bianisotropic STFs, are topics of future research.
Although the demonstrated successes of the STF concept and technology are
few as yet, the electromagnetic and elastodynamic frameworks for STFs are rea-
sonably mature. They have sufficient allure that committed researchers shall con-
tinue, and new researchers will be attracted by their novelty and challenge.
But in order for STF research and use to be truly widespread, economical pro-
duction must be enabled. Computer-controlled deposition schemes have already
been implemented [68, 172]. At the time of this writing, STFs of large thickness
(> 3 µm) and large-area uniformity (75 mm diameter) have been deposited with
growth rates as high as 0.4 µm/min [148]. A better production technique must be
even more rapid and deliver higher yields, so that larger-scale fabrication becomes
possible. The recent emergence of polymeric STFs, as exemplified by Fig. 1.13, is
very welcome in this connection.
Furthermore, the films will have to be laterally uniform with growth evolution
(i.e., possess Zone M morphology), and χ may have to be lower than 20 deg. This
will likely involve concurrent bombardment of the growing film by energetic par-
ticles, while the bombardment flux and the vapor flux may be differently directed.
If PVD (or any variant) is to be industrially successful, then new architectures
for the evaporant flux source—whether discrete or continuous, single or multiple—
must be developed to deposit STFs on large substrates. Reliability of deposition
uniformity would be facilitated by computer-controlled source and substrate ar-
chitectures. In turn, they will require the development of in situ monitoring of the
deposition process and appropriate control models. These and related avenues for
manufacturing research need to be opened up.

Downloaded From: http://ebooks.spiedigitallibrary.org/ on 09/25/2013 Terms of Use: http://spiedl.org/terms