Professional Documents
Culture Documents
Functional Thin Films and Nanostructures For Sensors. Synthesis, Physics, and Applications, 2009, p.224
Functional Thin Films and Nanostructures For Sensors. Synthesis, Physics, and Applications, 2009, p.224
Functional Thin Films and Nanostructures For Sensors. Synthesis, Physics, and Applications, 2009, p.224
springer.com
To our families and parents.
Olena, Nadia
Michelle, Abi, Libby
Foreword
In recent years, there has been a convergence of fundamental materials science and
materials processing methods. This convergence, although highly interdisciplinary
in nature, has been brought about by technologies such as bandgap engineering and
related techniques that have led to application-specific devices such as lab-on-a-
chip and system-on-a-chip. The demand for reduced device size, device portability,
and low power dissipation coupled with high speed of operation continues to dictate
terms and conditions for the evolution of nanotechnology. The present trend in
approaches to systems manufacturing continues to focus on integration of multi-
functionalities on the same chip. These functionalities include, for example, on-
board laser sources, sensors, and amplifiers. Both the military and civilian markets
continue to drive the research and development component. In recent years, the
emergency preparedness guidance systems have added excitement and curiosity to
this expanding industry. The outgrowth of technologies of interest for emergency
preparedness includes the development of terahertz sources and detectors and sys-
tems for detection of explosives and concealed weapons, among others.
Sensors made from bulk materials have been around for a long time. Enormous
advances in the processing technologies of thin films have led to the ability to
manufacture application-specific functional thin films. These include transparent
electrodes and antireflection films such as indium tin oxide, which serve as inter-
face components between humans and electronic devices, or optical circuit ele-
ments used in optical communication networks, or as contacts and antireflection
coatings in solar cells. Products are also being developed with magneto-optical,
electrochromic, or UV material for their use as functional thin films in optics.
Photonic crystals contain a variety of functional thin films; they require processing
of thin films under very stringent control of their structure and properties.
For microelectromechanical systems (MEMS), in addition to silicon-based tech-
nology, ferroelectric thin films are being used in the fabrication of microactuators
and micromotors, capacitors, and other thin-film devices. Functional thin films are
being used in the manufacture of devices such as surface acoustic wave (SAW)
devices for high-frequency telecommunications filtering, infrared detectors, pres-
sure sensors, accelerometers, force sensors, vibration, thickness, and chemical sen-
sors and biosensors. The reduction in size from bulk to micro- and nanostructured
transducers, while promising high sensitivity, high speed, and increased selectivity,
vii
viii Foreword
requires new design considerations that should consider factors such as integration
with other devices and device lifetime.
Functional thin films offer an enormous infrastructure for a highly interdisciplin-
ary integration of inorganic/semiconducting, organic/bio, and electronic/optoelec-
tronic sensor systems. The field is constantly evolving and will continue to do so
by absorbing novel materials approaches such as carbon nanotubes, high Tc super-
conductors, ferroelectrics, and thermoelectrics.
The chapters in this book are designed to give the reader the big picture, from
the design phase to the implementation and realization of a transducer. Every effort
has been made to include the state-of-the-art in each chapter. The intended audience
is scientists, researchers, and engineers, however, graduate students will find the
book to be very useful in their research and understanding of sensors and beyond.
The editors and contributors are leading researchers in industry and academia in
their subject areas.
Newark, New Jersey N. M. Ravindra
February 2008
Series Preface
The book series Integrated Analytical Systems offers the most recent advances in all
key aspects of development and applications of modern instrumentation for chemi-
cal and biological analysis. The key development aspects include: (i) innovations in
sample introduction through micro- and nanofluidic designs; (ii) new types and
methods of fabrication of physical transducers and ion detectors; (iii) materials for
sensors that became available due to the breakthroughs in biology, combinatorial
materials science, and nanotechnology; and (iv) innovative data processing and
mining methodologies that provide dramatically reduced rates of false alarms.
A multidisciplinary effort is required to design and build instruments with previ-
ously unavailable capabilities for demanding new applications. Instruments with
more sensitivity are required today to analyze ultratrace levels of environmental
pollutants, pathogens in water, and low vapor pressure energetic materials in air.
ix
x Series Preface
Sensor systems with faster response times are desired to monitor transient in vivo
events and bedside patients. More selective instruments are sought to analyze
specific proteins in vitro and analyze ambient urban or battlefield air. For these and
many other applications, new analytical instrumentation is urgently needed. This
book series is intended to be a primary source of both fundamental and practical
information of where analytical instrumentation technologies are now and where
they are headed in the future.
Looking back over peer-reviewed technical articles from several decades ago,
one notices that the overwhelming majority of publications on chemical analysis
has been related to chemical and biological sensors and has originated from depart-
ments of chemistry in universities and divisions of life sciences of governmental
laboratories. Since then, the number of disciplines has dramatically increased
because of the ever-expanding needs for miniaturization (e.g., for in vivo cell analy-
sis, embedding into soldier uniforms), lower power consumption (e.g., harvested
power), and the ability to operate in complex environments (e.g., whole blood,
industrial water, or battlefield air) for more selective, sensitive, and rapid determi-
nation of chemical and biological species. Compact analytical systems that have a
sensor as one of the system components are becoming more important than indi-
vidual sensors. Thus, in addition to traditional sensor approaches, a variety of new
themes has been introduced to achieve an attractive goal of analyzing chemical and
biological species on the micro- and nanoscale.
Preface
Anyone with the most cursory knowledge of sensors must have had a chance to use
such devices at some point in their life or career. Whether to collect data in a lab
course, to automate an otherwise tedious process, to improve the efficiency of a
delicately tuned process, or to do something as mundane as taking a family picture,
sensors have become an integral part of our environment and our daily lives.
Charge-coupled devices also known as CCD photodetector arrays, for example,
have revolutionized photography, astronomy, spectroscopy, X-ray diffraction, and
medical imaging to name but a few. A number of scientific discoveries have been
enabled by CCDs including the possibility to determine molecular and lattice struc-
tures at intermediate stages of a chemical synthesis or a structural transformation.
At the core of the widespread adoption of sensors are their rapidly decreasing foot-
print and cost and increased functionality. The miniaturization of solid-state devices
in general and sensors in particular was made possible thanks to significant trans-
formations and a large number of incremental and disruptive inventions in the area
of thin-film and nanostructure science and fabrication technologies.
Thin films and nanostructures can play multiple roles in a sensor including
structural support, reliability enhancement, filtering, and transduction. Thin films
and nanostructures are called functional when they fulfill a function other than
structural support. These micro- and nanostructured materials have applications
that extend far beyond sensing to data storage, lighting, displays, hydrophobic coat-
ings, decoration, and a large number of other fields that are outside the scope of this
book. In this book, these materials are discussed in the context of transduction and
how they contributed to the current sensor revolution.
Sensor design and fabrication are multidisciplinary and require broad and deep
knowledge in diverse areas of science and engineering such as materials science,
physics, chemistry, biology, and mechanical and electrical engineering. Covering a
subject with so many roots in diverse scientific and engineering disciplines is
undoubtedly a daunting task and any author who attempts it will do so with signifi-
cant trepidation. Aware of the challenge at hand, the editors of this book attempted,
ambitiously, to cover in one volume an account of general sensor theory, design
considerations related to the use of functional thin films and nanostructures, and
specific case studies of functional thin films and nanostructure applications in
sensing. Part of our motivation in taking on this task is that no such work, to our
xi
xii Preface
knowledge, has been published. Having said this, we are strongly familiar with the
large body of publications in this area that we refer to in this book and we are
keenly indebted to the works of many authors in putting this book together.
This book is devoted to teaching the new sensor designer the key steps involved
in developing sound transducer technology from materials selection, to design for
performance, to process development, and finally to integration. Throughout the
chapters, the authors emphasize and highlight the important role played by func-
tional thin films in solving problems and discuss how to take advantage of such
materials to build superior devices. The book is also intended to provide the more
experienced designers with a condensed summary of sensor design methodology
and excellent references that will prove useful in future sensor design endeavors.
To put all of the shared design and fabrication knowledge into perspective and add
a touch of reality to the concepts discussed in Chapters 1 through 4, Chapters 5
through 8 are completely dedicated to putting the theory into practice and demon-
strating the whole design process using a number of concrete applications.
February 2008 Anis Zribi
Jeffrey Fortin
Editor Biographies
xiii
Contents
2 Transduction Principles........................................................................... 17
Jeffrey Fortin
6 Chemical Sensors: New Ideas for the Mature Field ............................. 103
Radislav A. Potyrailo
xv
Contributors
Jeffrey Fortin
GE Global Research Center
Micro and Nano Structures Technologies
Chang Liu
Northwestern University
Department of Mechanical Engineering
Audry Nelson
GE Sensing, TelAire
Radislav Potyrailo
GE Global Research Center
Vicktor Ryzhii
University of Aizu
Computer Solid State Physics Laboratory
Wei-Cheng Tian
GE Global Research Center
Micro and Nano Structures Technologies
Magnus Willander
Linköping University
Institute of Science and Technology
Qingxiang Zhao
Linköping University
Institute of Science and Technology
Yiping Zhao
University of Georgia
Department of Physics and Astronomy
Anis Zribi
United Technologies Corporation Fire and Security, Kidde Detection
Technology Research, Development and Engineering Colorado Springs,
CO, USA
xvii
Chapter 1
Sensor Design Guidelines
Anis Zribi
Sensors are ubiquitous in our environment and play essential roles in our everyday
life. Our own view of the world is defined by our senses that enable us to perceive
stimuli from the environment through a network of biological sensors. Tiny hairs in
our inner ears detect the deflection of a membrane as it vibrates in response to
acoustic waves and make it possible for us to hear; photoreceptors in our eyes
enable us to see objects and discern their colors; chemical receptors on the tongue
(known as taste buds) allow us to differentiate between salty, sweet, bitter, and sour.
This fascinating network of biological sensors caters to our organs’ needs to control
certain biological processes and our needs for security and safety.
Driven by the need to better understand our world, to increase the productivity
of industrial processes and machines, and to improve our quality of life, scientists
and engineers constantly seek to develop the necessary measurement tools. These
tools are sensors and instruments that are often inspired by biological sensors and
their functioning principles. Such devices have become essential for advancing
A. Zribi
United Technologies Corporation Fire and Security, Kidde Detection Technology Research,
Development and Engineering 4820 Centennial Blvd, Suite 145, Colorado Springs 80919,
CO, USA
e-mail: anis.zribi@fs.utc.com
A. Zribi and J. Fortin (eds.), Functional Thin Films and Nanostructures for Sensors, 1
Integrated Analytical Systems,
DOI: 10.1007/978-0-387-68609-7_1, © Springer Science + Business Media, LLC 2009
2 A. Zribi
Sensor Architecture
Power Block
book is on the transducer block and the role that functional thin films and nanos-
tructures are increasingly playing in high-performance transduction.
Depending on the type of transducer, two basic types of sensors can be distin-
guished: quantitative (analog or digital) and threshold or binary. The two are quite
different in function and in application. A quantitative sensor produces an output
value that is a direct and continuous function of a measurand value. For example, a
thermocouple might have a potential differential of 10 mV at room temperature and
a potential differential of 20 mV 10° above room temperature. Any differential
potential value between these two is possible depending on the particular tempera-
ture to which the sensor is exposed. Threshold sensors, on the other hand, have only
two states, often called “on” and “off”. Perhaps the most familiar example of a
threshold sensor is a smoke detector which is triggered to the on position if a fire
erupts and the signal is used to trip an alarm.
The wealth of sensor and transducer technologies available to measure the same
measurands (Mi) increases the challenge of evaluating and comparing the perform-
ance of sensor devices. Therefore, it is critical to define a set of performance criteria
that the designers can use to develop sensors that meet customer specifications and the
user can use to appraise and contrast the various options. Table 1.1 summarizes the
list of performance attributes that are most commonly used to assess sensor tech-
nologies and a more detailed description of these parameters is provided later in
this chapter.
1 Sensor Design Guidelines 5
N
Mimin Mimax
Mi
ng
reasi
Inc
Mij
δN sing M i
Decrea
δMi
Mi
The dynamic range of a sensor is the span of measurands that constitute the overall
operating domain for the device. Within this interval, the sensor is supposed to
maintain its properties and reliability characteristics.
Response Curve
order to simplify the calibration procedures. Also, it is worth noting that customar-
ily the response of the sensor is normalized with respect to reference values (e.g.,
resistance at room temperature, resonance frequency of the membrane under known
conditions).
Considering all the possible measurands (Mi) and key noise parameters that can
affect the sensor output, the sensor response N can be expressed as
∂N
N = ∑ ΔMi (1.1)
i ∂M i
given that the different measurands affect the output independently of each other.
Sensitivity
∂N p ∂N p ∂N p − 2 ∂N1 p
i
Stransducer = ... = Π S kj . (1.3)
∂M i ∂ M p −1 ∂ M p −1 ∂M i k = 1
The amplification A is an intrinsic property of the amplifier and the analog filter
sensitivity is defined in a similar manner to the transducer block by Equation (1.3):
∂Y2 (1.4)
Sfilter = .
∂Y1
1 Sensor Design Guidelines 7
∂Y2 ∂Y1 ∂N
Si total = . . (1.6)
∂Y1 ∂N ∂Mi
Response Time
The response time of the device is a cumulative quantity incorporating the response
times of the respective building blocks. It is defined as the transient response of the
sensor when it experiences a step change in the measurand. If ti are the response
times of the various blocks, then the total response time of the sensor is:
t total = t transducer + t filter + t A / D (1.8)
In most diffusion-based chemical and biological sensors, tchemical transducer is the domi-
nant response time in Equations (1.7) and (1.8). Therefore, we typically can ignore
the contributions from electronics and physical transducers. However, this may not
be the case in thin-film or nanostructure-enabled sensors as the response time of the
material is considerably reduced because of the designed nanomorphology of the
material. This simplification is also not an option for physical sensors where the
response time of the transducer is on the same order of magnitude as the other sen-
sor subsystems.
The response time of the whole device is usually estimated as the time required for
a transient output signal to reach a fraction (e.g., 70%) of its steady-state change.
Resolution
The sensor resolution is the smallest change in input that leads to a detectable
change in output. The most general definition is given by Equation (1.10):
N N (1.10)
Re solution = Lim = noi
M i → noise Si total Si tot
8 A. Zribi
Clearly for nonlinear sensors, the resolution is dependent on the operating point and
can vary if operating conditions vary. Maximum resolution is attained at maximum
sensitivity and minimum noise.
Accuracy
The accuracy of a transducer is the maximum deviation of its output from the value
of the unknown measurand as determined by a gold standard technique. It is a
quantitative indication of the degree of conformity of the sensor to a standard.
Accuracy represents a systematic error or a bias in the sensor and measuring it and
correcting for it are difficult. Transducer calibration is required to account for this
type of error and partially correct it. Long-term changes in the performance of the
transducer because of materials’ aging and wear will affect its accuracy.
Precision
Hysteresis characterizes the lagging of the sensor response behind the variation of the
measurand. This can be attributed to the sensing material memory and/or to the trans-
ducer properties. As a performance specification, hysteresis is defined as the maximum
difference between the upscale and downscale readings on the same artifact during a
full-range traverse in each direction. It is often reported as the ratio (or percentage) of
the difference between the upscale and downscale readings to the full scale.
Drift can be defined as the slow unpredictable change of the sensor output at
constant input. Drift can affect both the signal and noise levels and it can emanate,
for example, from residual stress relaxation, residual diffusion, material aging, and
degradation. Drift in sensors is defined for a specific time interval of interest.
Selectivity
Sensor design is one of the most interdisciplinary technical areas requiring both
breadth and depth of knowledge in materials science, physics, chemistry, biology,
mathematics and statistics, electronics, and packaging. The successful design of a
sensing system requires very good communication between scientists and engi-
neers of different backgrounds.
Much as in any design activity, it is very difficult and undesirable to bind creativity
by a set of design rules or a design methodology. However, past a first stage, focused
on brainstorming, transduction mechanism and materials down-selection and feasi-
bility analysis, a successful design team must have a clear objective and a guiding
design methodology to steer their effort. The typical sensor design steps include:
• Collection and analyses of the device specifications
• Transducer selection/invention
• Materials selection/invention
• Sensor design and modeling
• Prototyping
• Measurement of materials properties
• Prototype testing and model calibration
• Design iteration(s)
• Final device fabrication
• Technology transition to manufacturing
• Fabrication process scaleup
The most challenging and least regulated steps are the first three and they are least
covered by the literature. In the next sections, we provide some guidance, concepts,
and ideas to help the new sensor designer make faster progress towards his or her
ultimate goal.
Assuming that the design team is armed with a clear set of specifications for the
desired sensor performance and the operating and storage conditions, the next steps
10 A. Zribi
ought to target the selection or invention of the transducer(s) that can convert the
measurand into an electrical/optical signal. The questions that need to be answered
by the team at this stage are:
1. What specific physical, chemical, and biological properties does the measurand
possess that differentiate it from other potential sensor inputs?
2. For each of the identified characteristics, what are the potential confounding
inputs, which in the future will be identified as noise sources?
3. What type of energy conversion is required to transform the input signal (mechani-
cal, chemical, biological, electromagnetic) to a device output signal (optical or
electrical)?
4. What are the candidate transduction schemes that can be used for the target
measurand?
The final transduction scheme may involve one or multiple transducers operating
in series or parallel (array of detectors). Considering the following criteria can
further refine the list of candidate transduction mechanisms.
1. Simplicity of the transducer and robustness to failure
2. Materials requirements dictated by operating environment
3. Development time
4. Cost
The transducer material properties are important from performance and sensor reli-
ability perspectives. The suite of functional materials available to sensor and
micro-/nanoinstrument designers is rapidly expanding and numerous techniques
have been developed to integrate a large variety of organic and inorganic materials
and their alloys. The wealth of options puts the designer face to face with the chal-
lenge of selecting the best-suited materials given a transducer design concept.
Numerous considerations come into play when selecting a transducer material
and they fall into three categories: performance, process compatibility, and reliability.
Although it is very difficult to discuss the performance selection criteria of a trans-
ducer in general terms, process compatibility and reliability criteria are common to
all transducers and they are of chemical (chemical resistance, photo definability,
adhesion) and physical nature (rheology, mechanical, dielectric). Depending on the
transducer length scale, whether it falls in the bulk (>100 mm), micro (<100 mm),
or nano range (<100 nm), materials properties and their behaviors and responses to
environmental conditions and stimuli are significantly different.
Over the years, bulk materials properties have been well documented and often
standardized, however, submicron materials properties are still not as well docu-
mented and many are being investigated (Srikar and Spearing 2003). Some of the
1 Sensor Design Guidelines 11
As device dimensions scale down to hundreds of microns and single digit microns,
structural and functional materials’ dimensions that make up the device are
approaching the size of a few atomic layers and molecules, a range now familiar to
most of us as the nanoregime (<100 nm). In this nanoworld, the surface-to-volume
ratio is extremely high and diverges as dimensions shrink down. In this dimensional
range, surface effects (curvature, surface energy) and structural defects dominate
materials’ properties. This can be further explained by the fact that intermolecular
forces such as Van Der Waals, London dispersion forces, ionic interactions, hydro-
gen bonds, and dipole–dipole interactions prevail at the nanoscale. In liquids con-
fined to nanodimensions, short-range order has been already observed and reported
in numerous studies and liquid thin-film thickness begins to take discrete values.
The implication is that the physical and chemical properties of nanomaterials
differ greatly from their bulk counterparts and they are often strong functions of
thermal fluctuations. These properties include the effective viscosity, diffusion
coefficients, melting point, glass transition temperature, refractive index, mechanical
properties (elastic modulus), and the thermal conductivity.
Melting Point
where Tmbulk is the bulk melting point, Tm(d) is the melting point of a cylindrical
particle of diameter d, ssl is the surface tension of the solid–liquid interface, ΔHf
is the bulk enthalpy of fusion, and rs is the density of the solid.
The presence of a higher fraction of atoms near surfaces and interfaces is charac-
teristic of thin films and nanostructures. The proximity of atoms to a surface or an
interface creates an atomic environment, different from the bulk, where surface free
energy plays a bigger role in the elastic and plastic behavior of materials. Film
thickness and grain size, for example, have been proven to affect the deformation
mechanisms in metallic films significantly (Lilleodden et al. 2001). Metallic films
with a thickness of a micron or less exhibit very different plastic behaviors depend-
ing on the grain size. Initially, films with larger grains exhibit pronounced hardness
whereas fine-grained films show a soft behavior. When exposed to a load producing
1 Sensor Design Guidelines 13
3 ⎡ 2G 1 ⎤
Eˆ b = Eb + ⎢K − (9 L + 18 M + 8 N )⎥
4a ⎣ 3 Eb ⎦
(1.13)
1 ⎡1 2G ⎤
Eˆ shear = Eshear + ⎢ ( K + 6 m ) − 1 (3 M + 4 N )⎥ ,
a ⎣5 3 Eb ⎦
where Eb and Eshear are, respectively, the bulk and shear moduli of the material, K
and m are functions of the surface energy, Γ1 a first-order term in the power series
expansion of the surface energy density as a function of the surface strain, and L,
M, and N are the third-order elastic constants of the material. Equation (1.13)
clearly indicates that the contribution of the surface energy to the elastic modulus
is inversely proportional to the characteristic length of the structure. This term
increases the elastic modulus of materials in the nanoregime and is responsible for
the high stiffness of nanomaterials. An immediate consequence of high stiffness is
that nanostructures possess much higher mechanical resonance frequencies than
bulk structures. This property has been implemented by numerous investigators to
develop resonators for various applications including trace chemical and biological
detection (Calleja et al. 2005), high sensitivity pressure calibrators, and others.
Viscosity
Functional thin films and nanostructures can be solid or liquid phase. In the micro-
and nanosize regime, fluidic droplets and films differ from bulk fluids because of
their pronounced inhomogeneity. Micro- and nanofluids have been reported in the
literature (Pozhar 2000) to exhibit various rheological behaviors that are often
contradictory. These behaviors are based on fluidic models with scarce experimental
data and numerous assumptions regarding the nature and magnitude of interactions
of the fluid molecules with their surrounding.
One of the most accurate models of the viscosity of confined, inhomogeneous
molecular fluids has been formulated by Pozhar and Gubbins and thus named the
14 A. Zribi
where
5 m
h=( )
16s 2 pb
Optical Properties
with dimensions larger than 10 nm, electrons still behave as do particles and the
classical size effect based on Drude’s model is applicable. For these length scales,
the dielectric function needs to be corrected for electron scattering. As the dimen-
sions of the thin film or particle near 10 nm, electrons behave more as waves and
Drude’s model is only applicable after introducing some energy range modifica-
tions that affect the dielectric function.
Various models (Wood 1982; Kawabata and Kubo 1966; Genzel et al. 1975;
Cocchini 1985; Bassani et al. 1985) built on different assumptions have been devised
to capture the absorption behavior of nanoparticles and thin films at a range of fre-
quencies with different degrees of success. Although many of these models predict
the correct trends for the absorption spectra, they often disagree numerically with
experimental data and diverge as the characteristic dimensions approach bulk length
scales (Huang and Lue 1994). In his publication, Huang derived an expression for
the real and imaginary parts of the dielectric function for small metallic particles
using Lindhard’s equation. Huang demonstrated with his model very good agree-
ment with experimental data for structures with sizes confined to less than 10 nm.
The indication is that absorption spectra of metallic structures in the nanoscale shift
towards the blue part of the spectrum and absorption peaks tend to broaden. These
findings can be accounted for and taken advantage of when designing optical com-
ponents with dimensions confined to the nanometer range. The optical properties of
metal nanoparticles and more specifically noble metal nanoparticles will remain a
continuous subject of research (Scarrafrdi et al., 2005, Scaffardi and Tocho, 2006)
because of the potential applications in many fields such as spectrally selective coat-
ings, nonlinear optics, and heterogeneous catalysis.
Summary
References
Jeffrey Fortin
Abstract This chapter presents the most common fundamental transduction principles
used in microsensors. Each section provides an overview of the theory and then
gives an example of a sensor that uses the transduction principle being described. A
classification of measurands is presented as well as the most common transduction
techniques including piezoresistance, piezoelectricity, capacitive, resistive, tunneling,
thermoelectricity, optical and radiation-based techniques, and electrochemical.
Introduction
This chapter presents the most common fundamental transduction principles used
in microsensors. Each section provides an overview of the theory and then gives an
example of a sensor that uses the transduction principle being described.
Wikipedia defines a transducer as follows.
A transducer is a device, usually electrical, electronic, or electro-mechanical,
that converts one type of energy to another for the purpose of measurement or
information transfer. Most transducers are either sensors or actuators. In a broader
sense, a transducer is sometimes defined as any device that senses or converts a
signal from one form to another. (www.Wikipedia.com)
In a similar definition a transducer is defined as a device providing a usable output
in response to a specific measurand, where the measurand is defined to be the physi-
cal quantity, property, or condition that is to be measured (Norton 1982). It is further
stated here that when one is designing a sensor or trying to choose the appropriate
transduction technique there are a few questions one can ask, including: What is the
measurand? What is the principle of transduction? What is the sensing element?
What are the limits of the measurand to which the transducer will need to respond?
J. Fortin
GE Global Research Center. 1 Research Circle, KW C314, Niskayuna, NY12309
e-mail: fortinje@research.ge.com
A. Zribi and J. Fortin (eds.), Functional Thin Films and Nanostructures for Sensors, 17
Integrated Analytical Systems,
DOI: 10.1007/978-0-387-68609-7_2, © Springer Science + Business Media, LLC 2009
18 J. Fortin
Piezoresistivity
Piezoresistivity is, in its most basic form, the change in a material’s resistance
resulting from a change in stress in the material. The word piezo is derived from
the Greek word piezein, which means to press or squeeze. Many materials exhibit
the piezoresistive effect and it is typically quantified by what is termed the gauge
factor. The gauge factor is the change in resistance per given strain per starting
resistance and can be described via the following equation,
DR
GF = . (2.1)
R.Strain
The gauge factor for silicon decreases with increasing impurity concentrations and
this can be predicted by model. Controlled doping is typically accomplished via ion
implantation of the specific doping ion into the silicon to define the piezoresistor.
An alternative technique is to deposit a film containing the doping ion over a pat-
terned Si surface and then use a temperature treatment to drive the dopant into the
depth of the Si.
The piezoresistive gauge factor decreases as temperature increases, and this can
also be predicted. The coefficients increase linearly with the inverse of temperature.
One can find a deep description of the mathematics in Sze (1994).
When deciding if piezoresistance can be used as a transduction measurand for a
particular measurand one only has to determine if the sensor can be designed such
that the measurand can produce a stress on a portion of the device where a piezore-
sistor can be located. Examples of measurands that are quantified via piezoresist-
ance include: pressure, vibration, acceleration, and magnetic field. Once it has been
decided that the measurand could be quantified via piezoresistance one must deter-
mine if it is the best approach that will meet all the specifications of the application,
as described in Chapter 1.
A great example of a transducer that uses the piezoresistive effect is a silicon-
based MEMS pressure sensor. Silicon-based pressure sensors have been around
since the late 1950s and they are a very mature technology. GE Sensing offers a
multitude of Si-based pressure sensors for a variety of applications including blood
pressure sensing, tire pressure sensing, industrial process measurement, and so on.
An overview of one pressure sensor is presented here that is designed to measure
tire pressure (www.GESensing.com).
This sensor is approximately 1 by 1 mm and is an absolute pressure sensor,
meaning the reference cavity is a vacuum and, once calibrated, the sensor gives the
absolute pressure inside the tire. As with most pressure sensors, the electronic readout
technique utilizes a Wheatstone bridge.
The sensor element is a thin silicon membrane with embedded piezoresistors.
The piezoresistors are formed via an implantation step and the proper doping level
is chosen to provide the highest gauge factor. The piezoresistors are positioned in
the areas of the membrane that see the highest strain due to the force of the pressure
bending the membrane (Fig. 2.1).
20 J. Fortin
Reference Cavity
Piezo- Resistor
Fig. 2.1 Side view of GE Sensing silicon pressure sensor. (Courtesy of GE Sensing.)
P1 Applied
Pressure
Reference Cavity
P2
Pressure
Fig. 2.2 Side view representing diaphragm deflection of GE Sensing silicon pressure sensor.
(Courtesy of GE Sensing.)
Piezoelectricity
+ + + + + + + + + + + + + + + +
− − − − − − − − − − − − − − − −
(a) (b)
Fig. 2.3 (a) Piezoelectric material with no externally applied strain has no net charge on its
surfaces. (b) The same material under strain produces charges at surface that can be measured
Test results for the microphone showed a sensitivity of 0.75 uV/Pa with a linear
dynamic range from 47.8 to 169 dB and a resonant frequency of 50.8 kHz.
There are many ways to convert a change in an external measurand into a change
in an electrical signal directly, particularly in a MEMS device, where there are moving
parts that can be accessed electrically or integrated thin films that often have electrical
properties based on the environment to which they are exposed.
Resistance: One key transduction technique that converts the measurand to a
change in resistance is piezoresistance, which, because of its widespread use in
microdevices was covered separately above. Other resistance-based transduction
techniques also rely on the measurand interacting with a film or bulk structure and
hence changing an electrical property. For example, certain polymers have a moisture-
or gas-sensitive resistivity. Another example is in the measurement of temperature,
where because resistance of a material is a function of temperature it can be directly
used to measure temperature; this type of device is referred to as a thermistor.
An example of a device that uses a materials change in resistance due to expo-
sure to the measurand can be seen in the work of Valentini et al. (2004). This device
uses carbon nanotubes (CNTs) as the functional transducer material. Carbon nano-
tubes present extremely high surface-to-volume ratios and have recently seen
significant attention for their gas adsorption properties (Treacy 1996). Valentini et al.
used an interdigital electrode structure made from platinum deposited and patterned on
top of a silicon nitride film. The CNTs were then grown from a catalyst between
the Pt electrodes to heights of approximately 200 nm. The results showed that the
resistance of the film on CNTs decreased when in contact with NO2 and increased
when in contact with NH3, ethanol, water vapor, and C6H6. The detection limit for
NO2 was shown to be as low at 10 ppb.
Capacitance: In this transduction technique the measurand interacts with the
device to change the capacitance value of a capacitor. This change can be induced
by changing the effective distance between the two plates or electrodes of the
capacitor or by changing the dielectric constant of the insulator material. Examples
of both are given below.
The capacitive transduction technique that can be used to measure pressure is
capacitance. In a typical silicon-based capacitive pressure sensor the design is similar
to a piezoresistive pressure sensor as described above. Instead of implanting pie-
zoresistors into the diaphragm, the diaphragm itself is used as one plate of a capaci-
tor. Alternatively a metal layer can be placed on or embedded in the diaphragm. The
second plate or electrode is located at the bottom of the gap (see Fig. 2.5). In the
sensor shown in Fig. 2.4 the substrate is a degenerately doped silicon wafer and the
membrane has been wafer-bonded under vacuum to a patterned oxide layer. Contact
is made to the substrate through an opening in the oxide and directly to the also
2 Transduction Principles 23
Pressure
Reference
Ground Capacitor
Si diaphragm
Si Wafer
(a) (b)
Fig. 2.5 (a) G-CAP2TM Humidity Sensor; (b) response curve for G-CAP2TM capacitive humidity
sensor. (Courtesy of GE Sensing.)
Thermoelectricity
Overview of Theory
Wire A
Wire B
T2 T1
V
REFERENCE
JUNCTIONS
RADIATION
DETECTING
JUNCTIONS
RADIATION
ABSORBER
Cold Junction
Diaphragm Hot Junction
Fig. 2.8 Cross-section of the GE thermopile IR sensing chip. (Courtesy of GE Sensing, www.
GESensing.com.)
Overview of Theory
Electrochemical
Overview of Theory
Summary
Fundamentally transduction is taking energy from one form and transferring it into
another and quantifying that energy change or energy input. As one can see there are
a number of measurands that can be quantified via a variety of transduction
techniques. There are often multiple transduction approaches to quantify a measur-
and and one must fully understand the specifications of the application in order to
down-select approaches. The final approach may not necessarily be clear and experi-
mentation or innovation may be required to determine or define the best approach.
2 Transduction Principles 29
References
Eaton WP, Smith JH (1997) Micromachined pressure sensors: review and recent developments.
Smart Mater. Struct., 6:530–539
Garcia C, Zhukov A, Zhukova V, Ipatov M, Blanco JM, Gonzalez J (2005) Effect of tensile stresses
on GMI of Co-rich amorphous microwires. IEEE Trans. Magnetics, 41(10):3688–3690
Gardner JW (1994) Microsensors: Principles and Applications. Wiley, West Sussex, England
Han M, Liang DF, Deng LJ (2005) Review paper, sensors development using its unusual proper-
ties of Fe/Co-based amorphous soft magnetic wire. J. Mat. Sci., 40:5573–5580
Halfner E (1969) The piezoelectric crystal unit-definitions and methods of measurement. Proc.
IEEE 57, No. 2
Horowitz S, Nishida T, Cattafesta L, Sheplak M (2006) Solid State Sensors, Actuators, and
Microsystems Workshop, Hilton Head Island, SC, June 4, 2006
Humenyuk I, Torbiero B, Assie-Souleille S, Colin R, Dollat X, Franc B, Martinez A, Temple-
Boyer P (2006) Development of pNH4-ISFETS microsensors for water analysis.
Microelectronics J., 37:475–479
Janata J (2003) Electrochemical microsensors. Proc. IEEE, 91(6):864–869
Ko S, Kim Y, Lee S, Choi S, Kim S (2003) Micromachined piezoelectric membrane acouostic
device. Sensors and Actuators A, Physical, 103:130–134
Liu CH, Kenny TW (2001) A high-precision, wide-bandwidth micromachined tunneling acceler-
ometer. J. MEMS, 10(3):425–433
Moskovits M (2005) Surface-enhanced Raman spectroscopy – A brief retrospective. J. Raman
Spectroscopy, 36(6/7):485–496
Norton H (1982) Sensor and Analyzer Handbook. Prentice Hall, NJ, pp. 18–24
Ried R, Kim E, Hong D, Muller R (1993) piezoelectric microphone with on-chip CMOS circuits.
J. MEMS, 993 (23):111–120
Royer M, Holmen J, Wurm M, Aadland O (1983) ZnO on Si integrated acoustic sensor. Sensors
and Actuators, A: Physical, 4:357–362
Stuart DA, Haes AJ, Yonzon CR, Hicks EM, Van Duyne RP (2005) Biological applications of
localised surface plasmonic phenomenae. IEE Proc. – Nanobiotechnol., 152(1):13–32
Sze SM (1994) Semiconductor Sensors. Wiley, New York
Exceptionally high Young’s modulus observed for individual carbon nanotubes”, Treacy, M.M.J.
(NEC Res. Inst., Princeton, NJ, USA); Ebbesen, T.W.; Gibson, J.M., Nature, v 381, n 6584, 20
June 1996, p 678–680
Valentini L, Cantalini C, Armentano I, Kenny JM, Lozzi L, Santucci S (2004) Highly sensitive
and selective sensors based on carbon nanotubes thin films for molecular detection. Diamond
Relat. Mat., 13:1301–1305
White RM (1987) A sensor classification scheme. IEEE Trans. Ultrason. Ferroelec, Freq. Contr.
UFFC-34:124
Zribi A, Iorio L, Lewis D (2005) Oil-free stress impedance pressure sensor for harsh environ-
ments. IEEE. Vol 2005, p 1275–1277
Chapter 3
Growth and Synthesis of Nanostructured
Thin Films
Yiping Zhao
Introduction
Fabrication and synthesis of nanostructured thin films (NSTFs) are essential for
exploring their properties and creating advanced applications. In general, the nano-
structured thin film (NSTF) is defined as an assembly of a thin layer of nanostruc-
tured objects such as nanoparticles, nanorods, nano-tubes, nanowires, nanoporous
networks. It is different from individual or bundles of nanometer scale objects. In the
literature, there are two kinds of nanostructured thin films (NSTFs) that have been
used without any differentiation. One refers to an ultrathin film with submicrometer
thickness (£100 nm), either having a continued morphology or a discontinued
islandlike morphology. This kind of NSTF is usually prepared by conventional
thin-film deposition techniques such as physical vapor deposition (PVD), chemical
Y. Zhao
University of Georgia, Department of Physics and Astronomy, 221 Riverbend Research
South Laboratory, 220 Riverbend Road, Athens, GA 306028
e-mail: zhaoy@physast.uga.edu
A. Zribi and J. Fortin (eds.), Functional Thin Films and Nanostructures for Sensors, 31
Integrated Analytical Systems,
DOI: 10.1007/978-0-387-68609-7_3, © Springer Science + Business Media, LLC 2009
32 Y. Zhao
Table 3.1 Process Conditions and NSTF Structures for NSTF Fabrication or Synthesis
Fabrication or Synthesis Conditions NSTF Structures
Approaches: bottom-up or top-down Morphology: nanoparticle, nanorod, nanotubes,
nanowires, nanoporous network, or mixtures
Mechanism: physical or chemical process Orientation: aligned or misaligned
Environment: gas or liquid phase Regularity: ordered or disordered
Temperature: high or low Layered structure: single layer or multilayer,
single material or different materials
Other factors, such as pressure, carrier gas,
plasma, substrate, etc.
vapor transport method, the catalyst-assisted fabrication method, the physical vapor
deposition method, and the chemical vapor deposition method.
The thermal vapor transport method is used for depositing nanorod or nanowire thin
films directly from the vapor of the materials under the supersaturating condition
(Xia et al. 2003; Wang 2004). The deposition can be a physical process through
nucleation, and it can also involve chemical reactions during deposition such as
oxidation. The experimental setup for a typical thermal vapor transport deposition
system is shown in Fig. 3.1.
The system consists of at least a furnace to control the temperature of the fabri-
cation chamber, a source to provide the material, and a substrate to collect the
NSTF. In most cases, a carrier gas is introduced to transport the vapor from the
source to the substrate. Also a vacuum pump is used so that a desired chamber pres-
sure can be obtained. In some cases, neither the carrier gas nor the vacuum pump
is used. The source vapor can be introduced directly from the crucible into the
chamber or by other thermal sources outside the chamber (by carrier gas) or by
other techniques such as laser ablation, chemical reduction, and so on.
The important processing parameters are temperature (both the chamber and the
substrate), the partial pressures of the carrier gas and the vapor source, the flow and
type of carrier gas, and the growth time. Usually the substrate temperature is lower
than the source temperature. Fig. 3.2 shows several examples of nanorod thin films
fabricated by the thermal vapor transport methods (Huang et al., 2001; Yang et
al. 2002). Under most growth conditions, randomly aligned and networklike nano-
rod or nanowire thin films are formed. With fine-tuned growth conditions, well-
aligned nanorod arrays can be fabricated. In general, the growth temperature is
relatively high.
Furnace
Fig. 3.1 A general experimental setup for thermal vapor transport method (Wang 2004).
(Reprinted by permission from the Annual Review of Physical Chemistry, Volume 55, copyright
2004, Annual Reviews, www.annualreviews.org)
34 Y. Zhao
Fig. 3.2 (a) Randomly ordered ZnO nanorods array (Huang et al. 2002b). (Reprinted by permis-
sion from Advanced Materials 13, 113 (2001), copyright 2001, Wiley.) (b) Aligned ZnO nanorod
array (Yang et al. 2002) (Reprinted by permission from Advanced Functional Materials, 12, 323
(2002), copyright 2002, Wiley) fabricated through thermal vapor transport method
(a) (b)
300 nm
Fig. 3.3 Aligned ZnO nanorod array fabricated through catalyst assisted method (a) Au catalyst
(Ng et al. 2003) (Reused with permission from Hou Tee Ng, Applied Physics Letters, 82, 2023
(2003) copyright 2003, American Institute of Physics), and (b) nio catalyst (Lyu et al. 2002).
(Reprinted by permission from Chemical Physics Letters 363, Seung Chul Lyu, Low temperature
growth and photoluminescence of well-aligned zinc oxide nanowires, 134, copyright 2002,
Elsevier)
Fig. 3.4 (a) Schematic illustration of the VLS process; (b) vertical Si nanowire array grown on
a (111) Si wafer (Wu et al. 2002). (Reprinted by permission from Chemistry-A European Journal,
8, 1260 (2002), copyright 2002, Wiley)
Fig. 3.5 A schematics illustrating the general steps and physical mechanism for a PVD process
since the 1960s (Levitt, 1970), and it employs a catalyst to promote the anisotropic
crystal growth. The catalyst usually forms eutectic droplets with the reactants, and
the size of the eutectic droplet determines the size of the nanorods. The eutectic
alloy droplet will sit on top of the nanorods until the rods cease growth due to other
reasons such as poisoning or end of growth as shown in Fig. 3.4a (Wu et al. 2002).
Fig. 3.4b also shows an example of an aligned Si nanorod array on a Si (100) sur-
face fabricated by the VLS method (Wu et al. 2002). In general, a specific catalyst
must be chosen for each material, and usually the growth temperature is relatively
high.
Fig. 3.6 Three major processes for PVD: (a) thermal evaporation, (b) sputtering growth, and
(c) pulsed laser deposition
physical means, (2) the vapor transports from its source to the substrate, and (3) the
vapor condenses on the substrate to form the thin film. There are several ways to
resolve the key step for the PVD process, that is, to convert the solid phase into the
gas phase. In general, three major processes have been utilized: (1) the thermal
evaporation method, (2) the sputtering growth method, and (3) pulsed laser deposition.
The basic experimental setups for these three fabrication techniques are illustrated
in Fig. 3.6.
For the thermal evaporation method, the source material is placed into a crucible,
and the crucible can be heated either by resistance or by an electron beam to its
melting temperature so that there is enough vapor coming from the source that can
be deposited onto the substrate. For the sputtering growth, a source target is used,
and inside the vacuum chamber high-energy ions are generated by ionization.
Those high-energy particles collide with the source target and knock out the atoms
or molecules from the source target. The ejected atoms or molecules condense onto
a substrate to form a thin film or NSTF. A similar idea is applied for pulsed laser
deposition; instead of using high-energy ions, a high-intensity, short-wavelength,
pulsed laser beam is used to bombard the target.
In most cases, the PVD methods are used to produce islandlike ultrathin films
especially with films of Au, Pt, or Ag for plasmonic applications (Bartlett et al.
2004), or to produce nanocluster films by co-deposition of metal and dielectrics
(Biswas et al. 2003, 2004). Recently, a so-called glancing angle deposition (GLAD)
technique has been developed to fabricate aligned nanorod array structures, and the
details of GLAD are discussed in the section, “Glancing Angle Deposition.”
3 Growth and Synthesis of Nanostructured Thin Films 37
The chemical vapor deposition (CVD) technique is a widely used method for
depositing thin films and nanostructures for a large variety of materials. In a typical
CVD process, as shown in Fig. 3.7, reactant gases or precursors diluted in a carrier
gas at room temperature enter the reaction chamber. The gas mixture is heated as it
Fig. 3.8 Aligned multiwall carbon nanotube arrays fabricated through CVD technique (Wei et al.
2002). (Reprinted by permission from Natures 416, B. Q.Wei, R. Vajtai, Y. Jung, J. Ward, Y.
Zhang, P.M. Ajayan, G. Ramanath, Organized Assembly of Carbon Nanotubes, copyright 2002,
Macmillan)
38 Y. Zhao
approaches the deposition surface. Depending on the process and operating conditions,
the reactant gases may undergo homogeneous chemical reactions in the vapor phase
before striking the surface, and the reactions result in a condensed phase thin film form-
ing onto the substrates with the corresponding volatile reactants being pumped from the
chamber. In order to promote the chemical reaction, either a plasma or laser beam can
be introduced during the process. CVD primarily is a thin-film deposition technique.
With the help of different catalysts, the CVD has become a major technique to make
NSTFs over the past 20 years. For example, many different nanostructures such as
carbon nanotube arrays (Fig. 3.8; Wei et al. 2002), inorganic oxide nanorods or
nanowires, and others can be fabricated using the CVD technique.
A unique CVD process is called the atomic layer deposition (ALD) process. ALD
is based on sequential, self-limiting surface reactions as illustrated in Fig. 3.9 (Kim
2003). For one growth cycle, only one layer of precursors adsorbs onto the substrate
due to self-limiting surface adsorption, and then a reduction step is performed leaving
a monolayer of the desired atom/molecule on the substrate. Thus, one can deposit one
layer of atoms or molecules onto the substrate. Because the process is controlled by
surface adsorption, this unique growth technique can provide atomic layer control and
allow conformal films to be deposited on very high aspect ratio structures.
Liquid phase fabrication techniques refer to the synthesized NSTFs under a wet envi-
ronment. These processes usually occur at relatively low temperatures but involve
chemical reactions. In the following section, several liquid phase fabrication methods
are discussed, but due to the limited knowledge of the author, there may be other
methods that have not been included.
3 Growth and Synthesis of Nanostructured Thin Films 39
Sol–Gel Methods
The sol–gel process is a versatile solution process for making ceramic and glass
NSTFs (Caruso and Antonietti 2001; Pomogailo 2005). The sol–gel process
involves the transition of a system from a liquid “sol” into a solid “gel” phase. The
“sol” is made of solid particles with a diameter of few hundred nanometers that are
suspended in the liquid phase. The particles condense into a new phase (gel) in
which a solid macromolecule is immersed in a liquid phase (solvent). A common
example is the hydrolysis and condensation reactions of metal alkoxides to form
larger metal oxide molecules. After a series of hydrolysis and polymerization reac-
tions, the precursor forms a colloidal suspension, or a “sol”. Further processing of
the “sol” enables one to make ceramic materials in different forms. Thin films can
be produced by spin-coating or dip-coating, and nanoporous NSTFs can be formed
after removing the wet “gel”. With proper viscosity of a “sol”, ceramic fibers can
be drawn from the “sol” through the electrospinning process discussed in the sec-
tion, “Electrospinning Methods.”
The diblock copolymer method is very similar to the sol–gel method. A diblock
copolymer molecule consists of two immiscible polymer blocks A and B that
are covalently bonded at one end as shown in Fig. 3.10 (Harrison et al. 1997). If the
molecules are given sufficient mobility, they self-organize to minimize the free
energy of the material system arriving at one of several possible phase morpholo-
gies characterized by a small interaction area between the two unlike blocks.
Diblock copolymer molecules have been studied extensively in bulk, and complete
phase diagrams have been generated from both modeling and experiments. In bulk,
the phase morphology depends only upon the relative chain lengths of the two
blocks. A highly asymmetric diblock yields a spherical phase (spheres of the
smaller molecular weightblock A immersed in a matrix of block B) and an increas-
ing fraction of block A initiates the formation of a cylindrical phase, a gyroid phase,
and finally a lamellar phase is created for a symmetric diblock where A and B are
of similar chain lengths. Fig. 3.11 shows an example.
Spin coating is a simple but well-known technique to prepare thin films from liquid.
One of the most important applications of spin coating is to cast photoresistant
films for the lithography processes during microfabrication. A typical spin coating
40 Y. Zhao
(a)
A Block A B B
A B B
A A B
A B
A
A
Block B B
(b)
Fig. 3.10 (a) A block copolymer consists of homogeneous blocks, A (lighter) and B (darker),
connected by a covalent bond, drawn here for equal lengths of A and B blocks. (b) The resulting
morphology depends primarily on the relative length of the two blocks. Lamellae are typically
observed for similar lengths of A and B blocks. Shortening the B block produces morphologies
such as cylinders, and finally, for an even shorter B block, spheres (Harrison et al. 1997). (Reused
with permission from Journal of Vacuum Science and Technology B, 16, 544 (1998), Christopher
Harrison, Miri Park, Paul M. Chaikin, Richard A. Register, and Douglas H. Adamson, copyright
1998, AVS The Science and Technology Society)
Fig. 3.11 SEM micrographs of exposed and developed PS template for different annealing times.
Each sample was annealed undisturbed at 160°C under a vacuum for total time of (a) 3.5 h, (b) 6 h,
(c) 15 h, and (d) 34 h. Lower images have been filtered to enhance edges (Guarini et al. 2001).
(Reused with permission from Journal of Vacuum Science and Technology B, 19, 2784 (2001), K. W.
Guarini, C. T. Black, K. R. Milkove, and R. L. Sandstrom, copyright 2001, AVS The Science and
Technology Society)
3 Growth and Synthesis of Nanostructured Thin Films 41
process, as shown in Fig. 3.12, involves depositing a small volume of a liquid onto
the center of a substrate and then spinning the substrate at high speed (typically
>1000 rpm). The centripetal force during the rotation will cause the liquid to spread
to the edge of the substrate leaving a thin layer of liquid. After evaporation or
solidification occurs on the surface, this liquid layer becomes a thin film. Final film
thickness depends on the nature of the liquid such as the viscosity, drying rate,
percent solids, surface tension, and so on, and the spinning parameters. If the liquid
contains nanostructures, such as nanoparticles, colloids, nanorods, or even materi-
als that can form nanostructures, such as sol–gel liquids, block copolymers, and the
like, the resulting film after spin coating is a nanostructured thin film.
Fig. 3.13 Nanochannels formed in anodized aluminum oxide films: the upper row shows the
results due to random nucleation (Li et al. 1998). (Reused with permission from Journal of
Applied Physics, 84, 6023 (1998), A. P. Li, F. Müller, A. Birner, K. Nielsch, and U. Gösele,
copyright 1998, American Institute of Physics.) The lower row results from regular molding on
the Al thin foil (Masuda et al. 1997). (Reused with permission from Applied Physics Letters, 71,
2770 (1997), Hideki Masuda, Haruki Yamada, Masahiro Satoh, Hidetaka Asoh, Masashi Nakao,
and Toshiaki Tamamura, copyright 1997, American Institute of Physics)
atomic layer at a potential less than that needed to deposit the element on itself. The
details of electrochemical ALD are discussed in Stickney (2002).
The opposite process for deposition is etching. During the process of elec-
trochemical etching of a metallic specimen, the reduction and oxidation process
(redox process) takes place. All metals in contact with the solution have a pro-
nounced tendency to become ionized by releasing (losing) electrons and
becoming oxidized. In electrolytic or anodic etching, an electrical potential is
applied to the specimen by means of an external circuit. A typical setup consists
of the specimen (anode) and its counterelectrode (cathode) immersed in an
electrolyte. Two well-known nanoporous thin films can be fabricated through
electrochemical etching: the anodized aluminum oxide (AAO) nanochannels,
and the porous Si structures. Fig. 3.13 shows the AAO fabricated through
anodization.
Electrospinning Methods
Point Charge
1
Syringe
Pendant Drop
Envelope Cone
y0
(a) (b)
Fig. 3.14 (a) Schematic drawing of the electrospinning process with a flat ground collector. (b)
Numerical solution showing the equipotential lines and the direction of the electrostatic forces
(arrows) between a point charge and an infinite plate (Theron et al. 2001). (Reprinted by permis-
sion from Nanotechnology, 12, 384 (2001), copyright 2001, AIP)
Fig. 3.15 Left: SEM of electrospun type III collagen (human placenta) (Matthews et al. 2002).
(Reprinted by permission from Biomacromolecules 3, 232 (2002), copyright 2002, American
Chemical Society.) Right: SEM images of aligned nanofibers that were collected on a carbon tape
attached to the edge of the disc collector (Theron et al. 2001). (Reprinted by permission from
Nanotechnology, 12, 384 (2001), copyright 2001, AIP )
loaded into a syringe. A high voltage is created between the capillary tube and
the collection plate so that an electrically charged liquid jet is generated from the
end of the capillary tube. Before reaching the collection plate, the liquid jet
evaporates or solidifies and is collected as an interconnected web on the plate.
The critical parameters for electrospinning are voltage, the distance between the
collection plate and the capillary tube, the properties of the liquid, such as vis-
cosity, evaporation, or solidification rate, syringe pumping speed, and so on.
Two examples are shown in Fig. 3.15. With the process conditions properly
tuned along with the proper designing of the collection method, capillary tubes,
porous nanofibers, aligned nanofibers, tubelike fibers, and so on, are all
achievable.
The general processing steps for direct replication are shown in Fig. 3.16. The
template is a nanoporous material, such as anodized alumina. Electrochemical
plating is usually used to fill the channels. Thus, a back electrode is first depos-
ited by thermal evaporation of Au or other metals. With the help of this elec-
trode, other materials can be electrochemically plated into the channels. After
chemically dissolving the template material, a nanostructure that replicates the
template is formed. Many metallic nanorod structures have been fabricated by
this method. Note that during the electrochemical plating, one could alterna-
tively change the materials plated into the channel. Thus, multilayered nanorod
structures can be fabricated.
3 Growth and Synthesis of Nanostructured Thin Films 45
(a) Nanoporous template (b) Back electrode coating (c) Electroplating desired materials (d) Disolving the template
Fig. 3.16 The direct replication process for template-based synthesis technique
(a) Nanorod template (b) Surface coating of nanorods (c) Removing the nanorod template
Fig. 3.17 A typical partial replication process: using nanorods as template to generate nano-
tubes
On the contrary, the partial replication method will not fill the channel fully. Rather,
as shown in an example of Fig. 3.17, the channel will be partially filled. In Fig.
3.17, the template is an aligned nanorod array. By using a CVD or ALD technique,
a uniform layer of desired material can be conformally coated on the surface of the
nanorods with a controllable thickness. This kind of transition structure can
be called a core-shell structure. After removing the top layer through a dry etching
process such as plasma etching or reactive ion etching, and chemically dissolving
the nanorod template, an array of nanotubes of the desired material will form.
However, the template is not limited to the nanorods, nanoparticles, such colloids,
or nanowires, and can serve as such a template to make either core-shell structures
or shelllike or tabulatelike structures.
sputtering growth, or pulsed laser deposition, and the like, can be readily converted
into a GLAD system. In fact, the GLAD technique is the extension of the com-
monly used oblique angle deposition (OAD) which has been practiced for many
years by the thin-film deposition community. Most of its fundamental growth
mechanisms are similar to the oblique angle deposition although deviation may be
expected.
The experimental setup for oblique angle deposition is shown in Fig. 3.18a. The
collimated evaporation beam has an incident angle q (³70° in general) with respect
to the substrate surface normal. The incoming vapor flux can be treated as a vector
F as shown in Fig. 3.18b. The flux has two components: a vertical component F^ =
F cosq and a lateral component (a vector) F|| with | F | = F sinq. The substrate will
||
receive the vapor flux from both vertical and lateral directions.
During the deposition onto a flat substrate, the impinging atoms will randomly
form islands on the substrate as shown initially in Fig. 3.19a. As deposition pro-
ceeds, the nucleated islands will act as further growth centers, and all the tall
islands will receive more impinging atoms as compared to the shorter ones due to
a so-called shadowing effect. This competition process will only leave the tallest
islands to grow into columns and a nanocolumnar film will be formed (Fig. 3.19b).
The lateral component F|| is the source for the shadowing effect.
For oblique angle deposition, F|| remains constant during the deposition, and a
columnar film with tilt angle b will be formed. Fig. 3.20 shows the cross-sectional
SEM images of Si thin films deposited at different incident angles. At q = 0°, a
continuous and uniform thin film is formed; at q = 30°, small columns begin to
grow; at q = 60°, the columnar structure becomes more obvious; at q = 80°, obvi-
ous columnar structures are formed into a film. In general, the column tilt angle b
is less than the vapor incident angle q which follows the empirical tangent rule,
Fig. 3.18 (a) Experimental setup for oblique angle deposition. (b) The incident flux F can be
decomposed into two different components, F⊥ is the flux perpendicular to substrate, and F|| is the
flux parallel to the substrate
3 Growth and Synthesis of Nanostructured Thin Films 47
Fig. 3.19 The shadowing effect during oblique angle deposition: (a) initial nucleation to form
shadowing centers, and (b) columnar structures formed due to the shadowing effect
Fig. 3.20 SEM cross-section view of Si films fabricated at different incident angles. The scale
bars are 100 nm
tan b = 1/2 tan q for small q (Nieuwenhuizen and Haanstra 1996; Lisfi and Lodder
1 − cos q ⎞
2001), or the cosine rule, b = q − arcsin ⎛⎜ ⎟ (Trait et al. 1993).
⎝ 2 ⎠
48 Y. Zhao
Fig. 3.21 Multilayer Fe/Ni structure deposited by OAD: (a) SEM top view, and (b) SEM cross-
section view
Thin films deposited by oblique angle deposition have the following character-
istics: the films are in general porous and they possess nanocolumnar structures.
The nanocolumns tilt away from the surface normal to the incident evaporation
beam direction. The size and density of the nanocolumns change as a function of
the incident angle q (Fig. 3.20). Furthermore, if one changes the deposition source
with a fixed deposition configuration as shown in Fig. 3.1, multilayered columnar
films can also be grown. Fig. 3.21 shows the SEM top view and cross-sectional
view of a Ni/Fe two-layer nanocolumnar film deposited by OAD. This structure
also shows the thermoelectric effect upon heating.
(a) (b)
F
Fig. 3.22 (a) Experimental setup for GLAD; (b) cancellation of the F|| term due to the rotation.
Only F^ contributes effectively to the increasing height of the nanocolumns
combination of the two rotations along with manipulation of the deposition rate, the
nanocolumns can be sculptured into a C-shape, S-shape, zigzag shape, matchstick,
helical, or vertical columns (Robbie and Brett 1997; Young and Kowal 1959;
Motohiro and Taga 1989; Azzam 1992; Robbie et al. 1996; Messier et al. 1997,
2000; Liu et al. 1999; Malac and Egerton 2001; Malac et al. 1999; Zhao
et al. 2002a, b; Ye et al. 2002). In fact, the most intriguing aspect of GLAD is that
the structures of the nanorods can be well designed by programming the substrate
rotations. The following gives examples of various nanostructures fabricated in
such a way.
Fig. 3.23 (a) The program sequence for vertically aligned nanocolumn growth, (b) SEM top
view, and (c) SEM cross-sectional view of the Si nanorods grown by continuous rotation. The
scale bars are 200 nm (Zhao et al. 2002a)
If the substrate rotation speed is intentionally slowed at each rotation interval, there
will be enough vapors deposited onto the substrate to form a column with the pre-
ferred orientation. The result of this slow rotation gives a continuous formation of
nanocolumns on top of each other and along different directions, that is, the forma-
tion of helical nanorod structures (Zhao et al. 2002a). By controlling the time
interval between each rotation step as well as the incident angle, one can control the
length, diameter of the pitch, diameter of the rod, and the number of turns of the
helical nanorod structure. For example, to fabricate square spiral posts (or square
spirals) with a circumference of L for one pitch, we fix the flux incident angle at
85° and program the stepper motor with the following sequence.
1. Grow a layer with a thickness of L/8 without rotating the substrate.
2. Rotate 90° at a fast azimuthal rotation rate.
3. Repeat steps 1 and 2.
4. Grow a layer with a thickness of L/4 without rotating the substrate.
5. Rotate 90° at a fast azimuthal rotation rate.
6. Repeat steps 4 and 5 to the desired pitch number for the square spirals.
The expected square spiral structure is shown in Fig. 3.24a. The Si square spiral
arrays on bare Si substrates fabricated through this method are shown in Figs. 3.24b
and c (Zhao et al. 2002a). Both the top-view and cross-sectional SEM images show
that the spirals are uniformly distributed across the whole surface with almost the
3 Growth and Synthesis of Nanostructured Thin Films 51
Fig. 3.24 (a) The sequence for fabricating a two-turn square spiral (see the detailed description
in the text), and SEM images of two-turn square Si spirals; (b) top-view; and (c) cross-section of
Si spirals on a bare Si substrate. The arrows on (c) indicate the growth sequence in a spiral, which
is similar to what we expect in (a) (Zhao et al. 2002a)
Fig. 3.25 (a) The program sequence for square nanospirals, and (b) the SEM cross-sectional
image of ten-turn square Si spirals. Due to short pitch, the nanospirals appear well separated from
each other
same length. However, each spiral is closely packed to the adjacent spiral which
means that the whole film should act as a spiral bed instead of individual spirals. The
structure of the spiral is described by the arrows in Fig. 3.24c. From the substrate to
the top of the spiral, the arrow changes direction ten times; the first two direction
changes are small, which correspond to the L/8 arms and the remaining eight turns
correspond to the eight L/4 arms representing two complete turns of the square spi-
ral. This structure is similar to what we expect in Fig. 3.24a. The diameter of the Si
arm is about 50 nm, and the diameter of the spiral is about 200–500 nm. The twist
angle of the spiral is determined by the incident flux angle because the fabrication
process is mostly under a fixed angle deposition. The resulting tilt angle (similar to
the experiment shown in Fig. 3.20 was determined to be 55° ± 2°. By adjusting the
L parameter, smaller-sized spiral nanostructures can be formed. Fig. 3.25 shows the
programmed azimuthal rotation for square spiral structures and an example of a ten-
turn spiral film. The deposition conditions are almost exactly the same as the previ-
ous deposition except for the value of L = 200 nm.
52 Y. Zhao
Fig. 3.26 The program sequence examples for (a) zigzag, and (b) C-shape nanorods
Fig. 3.27 The SEM (a) top-view, and (b) cross-section images of multilayer Si spiral/straight
nancolumns grown on a bare Si(100) substrate. There are two layers of six-turn spirals and two
layers of 300 nm long columns (Zhao et al. 2002a)
Fig. 3.28 The SEM cross-section images of double handed TiO2 triangular spirals (van Popta
et al. 2005). (Reused with permission from Journal of Applied Physics, 98, 083517 (2005), Andy
C. Van Popta, copyright 2005, American Institute of Physics)
to the program shown in Fig. 3.25a, then continuously rotate the substrate at a faster
speed to build the straight nanocolumns according to the program shown in Fig.
3.23a. One would then repeat the whole process. For the film deposited in Fig. 3.27,
the growth conditions are R = 0.32 nm/s, L = 120 nm, and the rotational speed for
the columns is 0.0019 rev/s. There are two layers of six-turn spirals and two layers
of 300 nm long columns. The two spiral sections and two column sections are
almost the same. By changing the computer program in a similar manner, other
kinds of multilayered nanostructures can be fabricated. For example, by changing
the rotation directions, van Popta et al. (2005) have fabricated double-handed trian-
gular helical structures from TiO2 as shown in Fig. 3.28.
In the same spirit, multilayered nanostructures with different materials can be fab-
ricated using a multilayer deposition system; that is, the source can be changed
during the deposition. Two simple examples are shown in Fig. 3.29: one is a match-
sticklike Cu/Si nanorod structure, and the other one is a Ni rod/Si spiral structure.
The morphology of these two structures is relatively simple since one material
forms a layer of simple morphology. However, by programming the substrate rota-
tion and the source material simultaneously, one can fabricate even more compli-
54 Y. Zhao
Fig. 3.29 The SEM cross-section images of (a) Cu/Si matchsticklike nanorod structure, and (b)
Ni rod/Si spiral structure
Fig. 3.30 The program sequence for growth of Ni/Si multilayer nanospiral structure and expected
morphology
cated structures. Fig. 3.30 shows the program sequence to make a Si/Ni multilayer
spiral (He et al. 2007):
1. Grow a layer of Si nanorods of 800 nm in deposition thickness without rotating
the substrate.
2. Close the source shutter, rotate the substrate 90°; at the same time, replace the
Si source by a Ni source.
3. Grow a layer of Ni nanorods of 800 nm in deposition thickness without rotating
the substrate.
4. Close the source shutter, and rotate the substrate 90°; at the same time, replace
the Ni source by the Si source.
5. Repeat steps 1, 2, 3, and 4 to obtain the desired two-turn and eight-arm square
nanosprings.
3 Growth and Synthesis of Nanostructured Thin Films 55
Fig. 3.31 The SEM (a) top view, and (b) cross-section images of Ni/Si multilayer nanospiral
structure. The arrows on (b) indicate the growth sequence in a spiral. (c) The TEM image of a
bundle of Ni/Si multilayer nanospirals. The dark parts represent Ni and light parts are Si
(He et al. 2007)
The flux incident angle was fixed at 86°, and the substrate was rotated azimuth-
ally in the same direction. The resulting Si/Ni multilayer spiral structures are shown
in Fig. 3.31. The SEM top view and cross-sectional view show that the nanorods
are similar to the square spiral as shown in Fig. 3.24, but the TEM image shows
different contrasts due to different materials; the dark parts are Ni nanorods, and the
lighter parts are Si nanorods.
For the nanostructure fabrication described in the previous section, we have only
utilized either a continuous rotation or interrupted rotation. However, if we intro-
duce two different rotation speeds during one evolution, then the relative phase
difference between these two rotation speeds can be used to tune the topology of
the nanostructures (Zhao et al. 2002b; Ye et al. 2002). One obvious way to incor-
porate the phase factor is to introduce a symmetric rotation with two different rota-
tion speeds. In order to do so, we divide each revolution during deposition into 2N
phase sectors where N represents the N-fold symmetry of the desired shape of the
columns. The rotation speed was alternately changed between wl and wh (wh>> wl)
for adjacent sectors. Because wh is much higher than wl, within the sector with
rotational speed wh, the sample will receive a smaller amount of vapor per radian
than that with wl. Thus, the shape of the nanorods can be tuned by changing the
ratio of wh/wl as well as the phase sector N.
Fig. 3.32 shows a program sequence for a twofold symmetry rotation GLAD.
Different slopes in different phase sectors indicate the difference in rotation speeds.
Fig. 3.33 shows the resulting Si nanorod structures with two-, three-, and fourfold
symmetry rotation (Zhao et al. 2002b). The examples show that different symmetrical
rotations alter the lateral arrangement of the nanorod arrays while also changing the
shape or aggregation of the nanorods. Due to rotational symmetry, all the nanorods
are aligned with each other and are perpendicular to the substrate surface.
56 Y. Zhao
Fig. 3.33 The SEM top view images of different GLAD samples with different rotational sym-
metries. The upper row lists films deposited on bare Si(100) substrates, and the lower row lists
films deposited on the colloid substrates which have point defects (Zhao et al. 2002b)
If the phase sectors are not divided symmetrically, different sectors will receive
different amounts of vapor; one should expect that the tilting angle of the nanorods
can be modulated (Ye et al. 2002). One simple case is to divide one revolution into
two different phase sectors as shown in Fig. 3.34. In one sector, the substrate rotates
3 Growth and Synthesis of Nanostructured Thin Films 57
0
0 T/2 T
Program Time
at speed wl for an angular range of f, and in the remaining sector, the substrate
rotates at speed wh for an angular range of 2p – f . Because the asymmetry in the
f direction is intentionally introduced, there will be a net F|| after a complete revo-
lution, and there will be an effective vapor incident angle a different from q. The
apparent lateral flux F¢|| will be changed according to the sizes of sectors and the
ratio of the rotational speeds:
⎡ f / 2 F|| cos ϕ −f / 2
F|| cos ϕ ⎤ 2 F|| ⎛ 1 1⎞ f
F¢ || = ⎢ ∫ dϕ + ∫ dϕ ⎥ / t = ⎜ − ⎟ sin . (3.1)
⎢⎣ −f / 2 w l f /2
wh ⎥⎦ t ⎝ wl wh ⎠ 2
f 2p − f (3.2)
t= + .
wl wh
⎛ 1 1⎞ f ⎛ wh ⎞
2 F sin q ⎜ − ⎟ sin ⎜⎝ w − 1⎟⎠
F¢ || ⎝ wl wh ⎠ 2 f (3.3)
tan a = = = 2 tan q sin l
.
F⊥ ⎛ f 2p − f ⎞ 2 ⎛w ⎞
F cos q ⎜ + 2p + ⎜ h − 1⎟ f
⎝ wl w h ⎟⎠ ⎝ wl ⎠
Because the experimental incident angle q is fixed, the column density should
be constant whereas the effective incident angle a is changed by changing f and
wh/wl. We expect the tilt angle b will change continuously; the continuous tuning of
58 Y. Zhao
the tilt angle has been demonstrated experimentally (Ye et al. 2002). A special case
is if the substrate is rotated uniformly, wh/wl = 1; then tan a= 0. The deposited
columns have no preferential deposition angle, and the columns stand up along the
normal of a substrate.
Almost all the nanocolumnar films shown above are deposited onto bare flat Si
substrates. In these cases, the random nucleation is the deterministic mechanism for
the lateral distribution of the nanocolumns. However, if we can use one of the major
growth mechanisms of GLAD, the shadowing effect, properly, we can grow a high
aspect ratio regular array of nanocolumns. To achieve this goal, we can use a sub-
strate with a proper two-dimensional nanotemplate as shown in Fig. 3.35a. The
features on the template will act as shadowing centers, and the deposition particles
will only accumulate onto the shadowing centers under the proper geometric depo-
sition conditions especially when the vapor incident angle q is larger than the critical
angle qc. The critical angle qc, as shown in Fig. 3.35a, is determined by the geo-
metry of the template, tan q c = L / h . Fig. 3.35b also shows an example of a regular
helical Si spring grown onto regular W plugs fabricated by photolithography. Clearly,
the use of a template provides a lateral control of the nanocolumnar growth.
All the above fabrication methods concentrate on a single degree of freedom for the
substrate manipulation which is azimuthal rotation. In fact, there are two other
rotational degrees of freedom one can utilize: the polar rotation to change the
particle incident angle q, and the nutation to rotate the substrate to face a different
(a)
D L q<qc
q>qc
qc
h
Fig. 3.35 (a) The regular array template and the geometry of the GLAD: the definition of the
critical incident angle qc; (b) SEM cross-section view of regular Si spirals grown onto regular W
plugs fabricated by photolithography method. The scale bar is 1 mm
3 Growth and Synthesis of Nanostructured Thin Films 59
Fig. 3.36 (a) The SEM cross-section view of three-dimensional porous Si structure fabricated by
combining azimuthal and polar rotations (detailed sequence is described in the text). (b) A cap-
ping layer formed on top of the nanorod surface (Robbie and Brett 1997). (Reused with permis-
sion from Journal of Vacuum Science and Technology A, 15, 1460 (1997), K. Robbie and M. J.
Brett, copyright 1997, AVS The Science and Technology Society)
phase of the incident flux. The polar rotation has been demonstrated to change the
size and density of the nanocolumns as shown in Fig. 3.20 and the nutation has not
been used for nanocolumn fabrication.
In principle, one can control all three rotational degrees of freedom for rotation
to fabricate the desired nanostructures. Fig. 3.36a shows an example of a porous Si
thin film fabricated by combining azimuthal rotation and polar rotation. To fabri-
cate this film, the sequence of the deposition is as follows.
1. Open shutter.
2. Set q = 85°, and deposit for 200 s.
3. Change q to 75°, and deposit for 100 s.
4. Change angle q to 20°, and deposit for 100 s.
5. Close shutter.
6. Change angle q from 20° to 85°.
7. Repeat steps 1–6 for another six times (seven layers).
During the deposition, the azimuthal rotation proceeded with a rotation speed of
0.5 rev/s, and the deposition rate was 0.5 nm/s. Other similar structures can be
designed through the two motor controls. For example, by programming the inci-
dent angle, one can put a capping layer onto the nanorod structures (Fig. 3.36b;
Robbie and Brett 1997).
By slightly changing the configuration of the GLAD setup, we can deposit aligned
nanostructures onto cylindrical optical fibers (Fan and Zhao 2005). The experimental
setup is shown in Fig. 3.37. Inside a PVD chamber, a cylindrical object such as an
optical fiber is installed coaxially onto a stepper motor (Motor2) through a pin vise.
60 Y. Zhao
Motor2
Motor1 Azimuthal
Polar rotation
rotation
Rod mount:
pin vise, etc.
Uniform rod
q
Evaporation
source
Fig. 3.37 Experimental setup for nanorod deposition onto a cylindrical object (Fan and Zhao
2005)
The object can be rotated in both polar (by Stepper Motor1) and azimuthal (by
Stepper Motor2) directions controlled by a computer. In order to deposit aligned
nanorods onto the surface of a cylindrical object, the axis of the object is slightly
rotated to a polar angle q < 10° by Motor1. The source material is evaporated onto
the object at a constant rate. Therefore, during the deposition the incoming vapors
can be shadowed by nanostructures (including initial islands) on the surface, and
the cylindrical object is rotated at a constant speed about its axis by Motor2 to
expose the entire surface to the vapor in order to achieve uniform coating.
Fig. 3.38 shows the Ag nanorod deposited onto a silica optical fiber with core
diameter of 98.6 m. Other materials such as Cu, Si, Ni, Al, and TiO2 have also been
fabricated as nanorods onto optical fibers using this same method. These prelimi-
nary results show that it is possible to integrate well-aligned nanorod structures onto
optical fiber-based sensors; therefore, a nanostructure-based high-sensitivity sensor is
possible. With a slight modification of the deposition configuration, one can also
put nanorod arrays onto tapered objects. For example, Fig. 3.39a shows an Ag
nanorod array on a tapered micropipette surface. With programming both Motor1
and Motor2, one can also deposit multilayer film/nanorod (Fig. 3.39b) and nano-
rod/nanorod (Fig. 3.39c) structures onto optical fibers. These results demonstrate
that it is feasible to directly incorporate complicated nanoscale architectures onto
optical fibers. The nanorods can also be deposited onto the tip of an optical fiber.
3 Growth and Synthesis of Nanostructured Thin Films 61
Fig. 3.38 Aligned Ag nanorods on an optical fiber (Fan and Zhao 2005)
Fig. 3.39 (a) Ag nanorods on tapered micropipette tip; (b) Ag film/nanorod two layered struc-
ture; and (c) Ag nanorod/Si nanorod multilayered structure on an optical fiber. The arrows point
to the Ag/Si interface (Fan and Zhao 2005)
Fig. 3.40 The experimental configuration for depositing nanorods on the cross-section of an
optical fiber and an example of Ag nanorods on the core of an optical fiber (Fan and Zhao 2005)
Fig. 3.40 shows the deposition geometry and the resulting Ag nanorods on the core
of the surface. These results demonstrate that it is feasible to directly incorporate
complicated nanoscale architectures onto optical fibers using GLAD.
62 Y. Zhao
Summary on GLAD
We use numerous examples to demonstrate that the GLAD technique has the fol-
lowing advantages in terms of fabricating NSTFs. (1) It can form vertical aligned
nanorod arrays. (2) The size and density of the nanorods can be controlled by the
vapor incident angle. (3) There is virtually no materials limit. As long as the mate-
rial can be evaporated, it can form vertical aligned nanorod structures. (4) The
shape, alignment, and orientation of the nanorods can be easily changed by pro-
gramming the rotation procedures. (5) The shadowing effect introduces a self-
alignment effect where templates can be used to mediate the size and orientation of
the nanorods. (6) Three-dimensional-shaped nanorod structures can be sculptured
by computer programming. (7) Multilayered nanostructures are achievable using a
multilayer deposition system. (8) It is a low-temperature PVD process, and it is
compatible with microfabrication processes. For example, Brett et al. have inte-
grated nanostructures fabricated by GLAD into microfluidic devices (Kiema et al.
2005). The most intriguing fact of GLAD is that the structures of the nanorods can
be well-designed in such a way. This cannot be achieved by other nanostructure
fabrication techniques. However, the potential of GLAD techniques for device fab-
rication has yet to be fully explored.
References
Azzam RMA (1992) Chiral thin solid films: Method of deposition and applications. Appl. Phys.
Lett., 61:3118–3120.
Bartlett PN, Baumberg JJ, Coyle S, Abdelsalam ME (2004) Optical properties of nanostructured
metal films. Faraday Discuss., 125:117–132.
Biswas A, Aktas OC, Schürmann U, Saeed U, Zaporojtchenko V, Faupel F, Strunskus T (2004)
Tunable multiple plasmon resonance wavelengths response from multicomponent polymer-
metal nanocomposite systems. Appl. Phys. Lett., 84:2655–2657.
Biswas A, Marton Z, Kanzow J, Kruse J, Zaporojtchenko V, Faupel F, Strunskus T (2003)
Controlled generation of Ni nanoparticles in the capping layers of Teflon AF by vapor-phase
tandem evaporation. Nano Lett., 3:69–73.
Caruso RA, Antonietti M (2001) Sol-gel nanocoating: An approach to the preparation of structured
materials. Chem. Mater., 13:3272–3282.
Fan JG, Zhao YP (2005) Direct deposition of aligned nanorod array onto cylindrical objects.
J. Vac. Sci. Technol. B, 23:947–953.
Guarini KW, Black CT, Milkove KR, Sandstrom RL (2001) Nanoscale patterning using self-
assembled polymers for semiconductor Applications. J. Vac. Sci. Technol. B, 19:2784–2788.
Harrison C, Park M, Chaikin PM, Register RA, Adamson DH (1997) Lithography with a mask of
block copolymer microstructures. J. Vac. Sci. Technol. B, 16:544–552.
He YP, Fu JX, Zhang Y, Zhao YP, Zhang LJ, Xia AL, Cai JW (2007) Multilayered Si/Ni nano-
springs and their magnetic properties. Small, 3:153–160.
Huang MH, Wu YY, Feick H, Tran N, Weber E, Yang PD (2001) Catalytic growth of zinc oxide
nanowires by vapor transport. Adv. Mater., 13:113–116.
Huang ZM, Zhang YZ, Kotaki M, Ramakrishna S (2003) A review on polymer nanofibers by
electrospinning and their applications in nanocomposites. Composites Sci. Technol.,
63:2223–2253.
3 Growth and Synthesis of Nanostructured Thin Films 63
Wu YY, Yan HQ, Huang M, Messer B, Song JH, Yang PD (2002) Inorganic semiconductor
nanowires: Rational growth, assembly, and novel properties. Chem. Eur. J., 8:1260–1268.
Xia YN, Yang PD, Sun YG, Wu YY, Mayers B, Gates B, Yin YD, Kim F, Yan HQ (2003) One-
dimensional nanostructures: Synthesis, characterization, and applications. Adv. Mater.,
15:353–389.
Yang PD, Yan HQ, Mao S, Russo R, Johnson J, Saykally R, Morris N, Pham J, He RG, Choi HJ
(2002) Controlled growth of ZnO nanowires and their optical properties. Adv. Funct. Mater.,
12:323–331.
Ye DX, Zhao YP, Yang GR, Zhao YG, Wang GC, Lu TM (2002) Manipulating the column tilt
angles of nanocolumnar films by glancing angle deposition. Nanotechnology, 13:615–618.
Young NO, Kowal J (1959) Optically active fluorite films. Nature, 183:104–105.
Zhao YP, Ye DX, Wang GC, Lu TM (2002b) Novel nano-column and nano-flower arrays by
glancing angle deposition. Nano Lett., 2:351–354.
Zhao YP, Ye DX, Wang PI, Wang GC, Lu TM (2002a) Fabrication Si nano-columns and square
springs on self-assembly colloid substrates. Int. J. Nanosci., 1:87–97.
Chapter 4
Integrated Micromachining Technologies
for Transducer Fabrication
Wei-Cheng Tian
Introduction
In general, a micromachined sensing element consists of two major parts: the trans-
ducer and the sensing material. Various micro- and nanotechnologies can be used to
fabricate the sensing element. These micro- and nanotechnologies fall into one of two
categories: top-down and bottom-up fabrication approaches. In general, the top-down
approach involves semiconductor manufacturing processes and the bottom-up
approach requires chemical and/or biological techniques. The dimensions of struc-
tures fabricated using the top-down approach range from tens of nanometers to a few
hundred microns whereas bottom-up approaches can produce structures down to a
few angstroms in size. These characteristics make top-down approaches generally
more suitable for device platform and bottom-up approaches suitable for sensing
material development. For example, a micromachined diaphragm using a top-down
approach may be used as a device platform and a gas-specific film may be coated
using a bottom-up approach for sensing material. With the adsorption of the gas
onto the sensing film, the resonant frequency of the diaphragm will be altered due
to its mass or stress change.
W.-C. Tian
GE Global Research Center, 1 Research Circle, KW C1324, Niskayuna, NY12309
e-mail: tianw@research.ge.com
A. Zribi and J. Fortin (eds.), Functional Thin Films and Nanostructures for Sensors, 65
Integrated Analytical Systems,
DOI: 10.1007/978-0-387-68609-7_4, © Springer Science + Business Media, LLC 2009
66 W.-C. Tian
Lithography and pattern transfer are two major steps in the fabrication process of a
micromachined transducer. In this section, numerous lithographic techniques,
Section “Lithography,” and pattern transfer approaches, Section “Pattern Transfer,”
for various materials and applications are discussed and compared
Lithography
Optical Approaches
Table 4.1 Summary of Various Lithography Technologies for Micro- and Nanopatterning
Lithography Type Technologies
Optical approaches Ultraviolet (Haefliger and Boisen 2005), deep ultraviolet (Rothschild
2005; Kim and Kim 2005; Liu et al. 2005), extreme ultraviolet
(Collins and Edwards 2006; Gallatin et al. 2003; Silverman 2005;
Tanaka et al. 2005), interference lithography (Chan et al. 2005;
Gutierrez-Rivera et al. 2006; Sewell et al. 2005; Solak 2006),
immersion lithography (Dario et al. 2005; Smith et al. 2005),
zone-plate-array lithography (McGeoch 2005; Menon et al. 2005b)
Charged-particle- Electron beam lithography (Ahmed 1986; Baek et al. 2005; Classen
beam approaches et al. 1992; Kim HS et al. 2006; Myers et al. 2006; Zlobin 2005;
Weber et al. 2005), electron projection lithography (Dhaliwal et al.
2001; Doering et al. 2005; Eguchi et al. 2005; Kasahara et al. 2005;
Koba et al. 2005), ion beam lithography (Gamo and Namba 1984)
Other lithographic X-ray (Di Fabrizio et al. 2004; Jian et al. 2003; Liu et al. 2006; Ueno
approaches et al. 1999), LIGA (Ehrfeld et al. 1988; Guckel 1998; Rogner et al.
1992) (X-ray and LIGA can also be considered optical), scanning
probe lithography (Quate 1997; Soltys et al. 2003) (scanning probe
is also electron-based), hybrid lithography (Cleaver and Ahmed
1984; Steen et al. 2005, 2006; Yang 2005; Yasin et al. 2005)
Mercury arc lamps and excimer lasers are conventional light sources for litho-
graphic tools and emission lines at either 405 or 365 nm wavelengths are commonly
used. At these wavelengths, the minimum resolved line width can vary from 1 um
to a few micrometers (Haefliger and Boisen 2005). To further increase line width
resolution, advanced lithographic tools using deep ultraviolet (DUV) light and
extreme ultraviolet lithography (EUV) with wavelengths shorter than 248 nm are
required. Submicron line widths down to tens of nanometers are achievable with
these techniques (Rothschild 2005; Collins and Edwards 2006; Gallatin et al. 2003;
Silverman 2005; Tanaka et al. 2005; Salib et al. 2005; Liu et al. 2005). EUV lithog-
raphy systems use light with 10–14 nm wavelengths producing feature sizes as
small as 12 nm (Naulleau et al. 2005). Because the wavelengths used for EUV
lithography approach X-ray wavelengths, it is also called soft X-ray lithography.
Photoresist patterning using interference lithography has also been demonstrated
(Chan et al. 2005; Gutierrez-Rivera et al. 2006; Sewell et al. 2005; Solak 2006). An
interference pattern between two or more coherent light sources is formed and
applied to the photoresist layer coating the substrate. Interference lithography
offers advantages over other lithography techniques because of its ability to define
grid patterns over large areas in a single, fast, mask-less exposure.
Immersion lithography is used to extend the resolution of optical lithography
below 100 nm by replacing an air gap between the optics and the wafer with a
liquid (Dario et al. 2005; Smith et al. 2005). This results in dividing the diffraction-
limited feature size by the refractive index of the liquid which is chosen to be
greater than one. The shorter the wavelength of the light used in immersion lithog-
raphy the smaller is the minimum achievable feature size.
68 W.-C. Tian
Charged-Particle-Beam Approaches
Fig. 4.1 Thick microstructures fabricated using LIGA (Guckel 1998). (Reproduced by permis-
sion from IEEE, copyright 1998, IEEE.)
70 W.-C. Tian
When a voltage bias is applied between a sharp probe tip and the substrate, an
electric field is generated around the tip and can be used to locally oxidize Si or to
expose resist (Quate 1997). This local oxidation process is powerful because of its
fine resolution (sub-50 nm) and the robust oxide etch mask that is created. In addi-
tion, the electrons generated around the tips have lower energy (<50 eV) compared
to EBL (100 eV to 100 KeV). This lower energy prevents the scattering in the resist
and backscattering from the substrate and thus enhances the resolution.
Hybrid lithography, either by combining electron/ion beam lithography or opti-
cal/electron beam lithography, has been successfully implemented (Cleaver and
Ahmed 1984; Steen et al. 2005, 2006; Yang ; Yasin et al. 2005). These techniques
use different lithographic methods, for example, optical lithography and EBL, on a
single resist and marry the advantages of optical lithography (high throughput) and
EBL (high resolution) (Steen et al. 2006).
Pattern Transfer
Lithographic methods transfer patterns either from a hard or a soft mask (computer
layout charged-particle-beam lithography) to the resist coating the substrate, which
can be semiconductor and polymeric materials. The resist typically provides an
etch mask for further subtractive (etching) or additive (deposition) processes used
to build the transducer. In this section, various micro- and nanopattern transfer
technologies for numerous microsensor platform materials are introduced.
Semiconductor (Si, SiC, GaN, etc.) and ceramic materials (e.g., glass) are common
substrates for micromachined transducers (Peterson 1982). Fabrication techniques
that involve etching of bulk wafers are known as bulk micromachining whereas
fabrication techniques that involve deposition of thin and thick films of materials
(e.g., polycrystalline silicon, silicon nitride) are known as surface micromachining.
Several pattern transfer methods, such as plasma etching, wet etching, laser abla-
tion, ion milling, and sand blasting for both bulk and surface micromachining are
discussed in this section.
Plasma etching is one of the most effective methods for pattern transfer and is
independent of crystal orientation. A photoresist, oxide or metal layer, is typically
used as the etch mask depending on the chemistries of the used etching technolo-
gies. Fluorine-based plasma etching with a resist mask can be employed to fabricate
high aspect ratio (thick) structures whereas chlorine-based plasma etching with a
metal etch mask can be implemented for creating submicrometer (thin) structures
(Rakhshandehroo et al. 1998; Tian and Pang 2001, 2002, 2003, Tian et al. 2000).
Many plasma etching processes using fluorine chemistry have been developed
to produce various microstructures. The advantages of fluorine etching include high
4 Integrated Micromachining Technologies for Transducer Fabrication 71
etch rates and high selectivity to etch masks. However, fluor atoms react with
silicon spontaneously and result in isotropic silicon etching to cause undercutting
underneath the mask. A process, known as the Bosch process, utilizes a multiplexed
etch (sulphur hexafluoride, SF6)/passivation (Octafluorocyclobutane, C4F8) cycle to
alleviate undercutting issues (Robert Bosch GmbH 1990 and 1991). Wafer cooling
is another solution to reduce undercutting problems associated with fluorine-based
plasma etching (Tachi et al. 1991).
Chlorine-based plasma etching is ion-assisted rather than spontaneous and is
therefore anisotropic (Rakhshandehroo et al. 1998). The sidewalls of microstruc-
tures fabricated by chlorine-based etching are very vertical and no sidewall passiva-
tion or wafer cooling is required compared to fluorine-based etching. The disadvantage
of chlorine-based etching is a lower etch rate and selectivity to the etch mask in
comparison to fluorine-based etching.
Wet chemical etchants, such as potassium hydroxide (KOH), tetramethylammo-
nium hydroxide (TMAH), or ethylene diamine pyrocatechol (EDP), in conjunction
with a heavily doped layer used as an etch stop in semiconductor materials can also
be used to form microstructures (Kovacs et al. 1998). A p++ doped Si layer with a
thickness ranging from 3–15 um can be utilized to define the thickness of the
microstructures because the etchants will not attack heavily doped semiconductors.
Other etch stop examples include silicon nitride or a multiple-layer film combining
silicon oxide and silicon nitride. A plasma etch step can be combined with various
etch stop technologies to fabricate freestanding microstructures. Although struc-
tures fabricated by wet etching generally provide a better surface roughness and
lower cost compared to plasma etching technologies, the crystal orientation
dependence limits wet etching abilities to produce complex geometries.
Glass is also a very common choice for device substrates. Wet etching processes
using chemicals such as hydrofluoric acid (HF) or buffer hydrofluoric acid (BHF),
have been applied to micromachine glass substrates (Corman et al. 1998). However,
these etching methods are isotropic and the etch mask cannot resist long for build-
ing deep trenches or channels. Dry etching techniques, such as laser ablation
(Klank et al. 2002, Liu et al. 1997, Schaffer et al. 2001), ion milling, or sand blast-
ing (Steve 2001, Wensink et al. 2000) can be used to pattern structures in glass and
overcome wet etching limitations.
Ablation is defined as the removal of material from the surface of a substrate by
vaporization, chipping, or other erosive processes. Various laser systems (carbon
dioxide, CO2 or neodymium-yttrium aluminium garnet, Nd-YAG, laser) have been
successfully employed for micro-/nanomachining (Klank et al. 2002; Liu et al.
1997). Very high aspect ratios can be achieved by this technique but the associated
capital investment may be too high.
Ion milling entails high-energy inert ion bombardment on the substrate to physi-
cally break the bonding between molecules at the surface of the substrate and then
remove the substrate molecules by momentum transfer (Reyntjens et al. 2001). The
ion energy needed to break the molecular bonding of the substrate material is typi-
cally much larger compared to the chemical bond energy of substrate materials, so
the material dependence etch rate is minimal. One disadvantage of ion milling is
72 W.-C. Tian
that achievable etch rates are typically lower compared to other etching
techniques.
Sandblasting is a technology where solid particles are accelerated in a focused
beam and used to hit the hard surface of a substrate at very high speed to smooth,
clean, or shape the surface of the substrate. Recently, it has been introduced for
micro-/nanomachining and has been used to demonstrate through wafer vias with
high aspect ratio(Schlautmann 2001; Wensink et al. 2000). Although the cost of this
technology is low, the ability to pattern very fine features is limited to the microm-
eter range and the sidewall profile is typically tapered. These three dry etching
technologies can be used not only for glass, but also for other semiconductor mate-
rials such as silicon (Si).
Fig. 4.2 SEM photos of a top view of 10 nm diameter metal dots (Chou 2001). (Reproduced by
permission from MRS Bulletin.)
Electrode-Based Platform
Electrode-based transducers are the most popular because of their simple fabrication
processes and ease of sensing film integration. Metal thin films are typically depos-
ited and patterned on top of a substrate to form electrodes in various patterns. The
sensing material (thin film or nanostructures) is then locally incorporated in between
electrodes or on top of the electrodes and the sensor output is measured in the form
of impedance (Han et al. 2005; Lapicki et al. 2005; Park et al. 2005; Sun et al. 2006),
voltage (Lapicki et al. 2005; Park et al. 2005), conductance (Hernandez et al. 2005;
Sayago et al. 2005; Smardzewski et al. 2004), or frequency (Ippolito 2005) change.
Examples of electrode designs include interdigitated electrodes (Han et al. 2005;
Smardzewski et al. 2004; Ippolito 2005) as well as simple square contact pads
(Hernandez et al. 2005). The substrate material selection is often dictated by the
application. A glass or quartz substrate can be used for chemiresistive measurement
(Han et al. 2005; Lapicki et al. 2005; Park et al. 2005; Sun et al. 2006) and a gallium
arsenide (GaAs) substrate can be used for Hall effect sensing (Lapicki et al. 2005).
Other substrates, such as alumina for conductance measurement and lithium nio-
bate (LiNbO3) for surface acoustic wave sensing, have also been reported recently
(Ippolito 2005; Sayago et al. 2005).
Semiconductor-Based Platform
MEMS-Based Platform
Fig. 4.3 Top view of a FET-based transducer (Kim DS et al. 2006). (Reproduced by permission
from Elsevier, copyright 2006, Elsevier.)
76 W.-C. Tian
Nanostructure Assembly
Immersion coating assembly is used to integrate functional thin films and nanostruc-
tures with interdigitated microelectrode (IME) transducers. The IME transducer is
fabricated with metallization followed by lithography. This IME substrate is subse-
quently immersed into the prepared coating sol or the solution of mixed nanoparti-
cles and thiols (Han et al. 2005; Lapicki et al. 2005; Park et al. 2005; Sun et al.
2006). The thickness of the functional material can be controlled by the withdrawing
rate, immersion time, or the drying temperature. The multilayer can be obtained by
several repetitions of coating steps. Alcohol gas sensing using IME transducers with
titanium oxide (TiO2) film and detection of volatile organic compounds or nitroaro-
matic compounds using nanoparticles/thiols on IME transducers have been demon-
strated (Han et al. 2005; Lapicki et al. 2005; Park et al. 2005; Sun et al. 2006).
Layer-by-layer (LBL) assembly is also a common technique for functional materials
integration with the transducer. This assembly technique provides a simple, versatile,
and inexpensive approach for thin-film formation by alternating the deposition of
oppositely charged species from an aqueous solution (Hammond 2004). LBL has
been successfully used to integrate sensing materials such as polyelectrolytes and
TiO2 nanoparticle films for gas and humidity sensing (Kim et al. 2004), polyacrylic
acid/polyethylene oxide film for pH sensing (Lutkenhaus et al. 2005), and den-
drimer/metallic nanoparticles film for biosensing (Goulet et al. 2005). A combina-
tion of lithographic approach and LBL assembly is demonstrated to spatially
pattern the polystyrene particles (150 and 64 nm) with feature sizes of 5 to 20 um
(Cui et al. 2004).
4 Integrated Micromachining Technologies for Transducer Fabrication 77
Table 4.3 Summary of Integration Methods of Sensing Materials and Micromachined Transducers
Integration Technology Functional Material
Nanostructure assembly Immersion coating assembly
Immersion coating assembly (Sun et al. 2006) Titanium oxide (TiO2)
Immersion (Han et al. 2005) Nanoparticles and thiols
Layer-by-layer (LBL) assembly
LBL (Hammond 2004) Various materials
LBL (Kim et al. 2004) Polyelectrolytes and TiO2 nanoparticles
LBL (Lutkenhaus et al. 2005) Polyacrylic acid (PAA); polyethylene
oxide (PEO); dye
LBL (Goulet et al. 2005) Dendrimer/metallic nanoparticles on glass
LBL and photolithography (Cui et al. 2004) Polystyrene nanoparticles
Other self-assembly
Magnetically aligned (Hernandez et al. 2005) Iron palladium (FePd) alloy nanowires
Biological-based self-assembly Zirconium dioxide (ZrO2) nanoparticles
(Zhao G et al. 2005) with heme proteins on functional
glassy carbon electrode
Nanostructured deposition
Electrochemical Deposition (Park et al. 2005) Nanoporous platinum oxide (PtO)
Deposition metal–insulator–metal ensemble Alkanethiol–Au nano cluster
(Smardzewski 2004)
Sputtering (Ippolito 2005) Zinc oxide/indium oxide (ZnO/InOx) on
XZ lithium niobate (LiNbO3) substrate
Sputtering (Zhao Z et al. 2005) 60% Palladium (Pd) 40% gold (Au)
Sputtering (Mardare et al. 2005) Chrome (Cr)-doped titanium oxide
Plasma enhanced chemical vapor deposition Metal-containing diamondlike
(PECVD) for carbon-silicon and sputtering nanocomposite
for metal (Takeno et al. 2005)
Pulsed laser deposition (Modi and Lacy 2005) ZnO nanorods
Pulsed laser deposition (Vainos 2004) Purely epitaxial nanocomposite
Glancing angle deposition (Kim DS et al. 2006) Silicon dioxide
Thermal evaporation of metallic tungsten oxide Tungsten oxide (WO3)
powder followed by thermal annealing
(Jayatissa and Gupta 2005)
MBE and lithography (Lapicki et al. 2005) Indium antimonide (InSb)
Polymerization
Electropolymerization (Bangar et al. 2005; Fan Monomer nanowires
and Lu 2005; Hammond 2004)
Spin coating and UV polymerization (Schmidt Molecularly imprinted polymer
and Haupt 2005)
Spotting
Carbon nanotubes (CNTs) were dissolved in ethanol Single-walled carbon nanotube (SWCNTs)
and sprayed by air brush (Sayago et al. 2005) with palladium doping
Ink jet spotting (Braun et al. 2006) Proteoliposomes
Pipetting (Zribi et al. 2005) Polystyrene sulfonic acid and carbon
nanotubes
78 W.-C. Tian
Nanostructure Deposition
Polymerization
The electropolymerization of the functional thin film took place from the anode to
the cathode under an applied current and can be designed with individual address-
ability. By simply adding biomolecules in the solution, biofictionalization of con-
ducting polymer nanowires can also be implemented (Bangar et al. 2005). Other
polymerization techniques for functional thin film, such as molecularly imprinted
polymer mimicking the behavior of natural antibodies, can be prepared by spin
coating followed by UV polymerization (Schmidt and Haupt 2005).
Spotting
A hydrogen sensor based on carbon nanotubes (CNTs) with Pd doping has been
reported (Sayago et al. 2005). The CNTs were deposited on an alumina substrate
via airbrush. The CNTs were dissolved in ethanol and sprayed on alumina using
dry air carrier gas. The results showed CNTs are p-type semiconductors and the
resistance increased with reducing gas concentration.
A reliable functionalization of biofilm onto a transducer and a successful detec-
tion with quantitative data have been reported by Braun et al. (2006). Application
of proteoliposomes onto MEMS cantilever arrays using ink jet spotting was dem-
onstrated. The authors claimed that this technique can be potentially used to meas-
ure membrane protein-based receptor–ligand interactions and conformational
changes.
Another example using spotting to incorporate functional thin film into a sensor
platform is MEMS resonant gas sensors (Zribi et al. 2005), as shown in Fig. 4.4.
The sensor consists of a resonant nitride membrane and a nanostructured gas-
sensing film with high adsorption and specificity to analytes. The sensing film
solution is deposited via pipetting with a few microliters onto the backside of the
membrane. Thin films of polystyrene sulfonic acid and single-wall carbon nano-
tubes have been demonstrated to detect moisture and CO2 separately using this
transducer.
Fig. 4.4 The photograph on the left shows the top view of the transducer. The drawing on the
right shows the cross-section of the device (Zribi et al. 2005). (Reproduced with permission from
Elsevier, copyright 2005, Elsevier.)
80 W.-C. Tian
Summary
References
Ahmed H (1986) Advances in high resolution and high throughput in electron beam lithography.
Vacuum, 37:375.
Baek IB et al. (2005) Electron beam lithography patterning of sub-10 nm line using hydrogen
silsesquioxane for nanoscale device applications. Journal of Vacuum Science and Technology
B: Microelectronics and Nanometer Structures, 23:3120–3123.
Bangar M et al. (2005) Individually addressable conducting polymer nanowires for FET based
label-free sensing, 208th Meeting of The Electrochemical Society – Meeting Abstracts,
200502:2510.
Becker H, Gartner C (2000) Polymer microfabrication methods for microfluidic analytical appli-
cations. Electrophoresis, 21(1):12–26.
Braun T et al. (2006) Conformational change of bacteriorhodopsin quantitatively monitored by
microcantilever sensors. Biophysical Journal, 90:2970–2977.
Chan TYM et al. (2005) Photonic band gap templating using optical interference lithography.
Physical Review E – Statistical, Nonlinear, and Soft Matter Physics, 71:046605.
Chong HA, Choi J-W, Beaucage G, Nevin JH, Lee J-B, Puntambekar A, Le JY (2004) Disposable
smart lab on a chip for point-of-care clinical diagnostics. IEEE Proceedings, 92:154–173.
Chou SY (2001) Nanoimprint lithography and lithographically induced self-assembly. MRS
Bulletin, 26:512.
Classen A et al. (1992) High voltage electron beam lithography of the resolution limits of SAL
601 negative resist. Microelectronic Engineering, 17:21–24.
Cleaver JRA, Ahmed H (1984) Combined electron and ion beam lithography system. Journal of
Vacuum Science & Technology B: Microelectronics Processing and Phenomena,
3:144–147.
Collins L, Edwards C (2006) Tricks of the light. IEE Electronics Systems and Software,
4:42–45.
Corman T et al. (1998) Deep wet etching of borosilicate glass using an anodically bonded silicon
substrate as mask. Journal of Micromechanics and Microengineering, 8:8–87.
Cui T et al. (2004) Lithographic approach to pattern multiple nanoparticle thin films prepared by
layer-by-layer self-assembly for microsystems. Sensors and Actuators, A: Physical,
114:501–504.
Dario G et al. (2005) First microprocessors printed with immersion lithography. Microlithography
World, 14:4, 6, 18.
Dhaliwal RS et al. (2001) PREVAIL – Electron projection technology approach for next-genera-
tion lithography. IBM Journal of Research and Development, 45:615–636.
Di Fabrizio E et al. (2004) X-ray lithography for micro- and nano-fabrication at ELETTRA for
interdisciplinary applications. Journal of Physics Condensed Matter, 16:S3517–S3535.
4 Integrated Micromachining Technologies for Transducer Fabrication 81
Doering H-J et al. (2005) Proof-of-concept tool development for projection mask-less lithography
(PML2). Progress in Biomedical Optics and Imaging – Proceedings of SPIE, Emerging
Lithographic Technologies IX, 5751:355–365.
Eguchi H et al. (2005) Image placement of large window-size membrane for EPL and LEEPL
mask. Proceedings of SPIE – The International Society for Optical Engineering, Photomask
and Next-Generation Lithography Mask Technology XII, 5853:910–920.
Ehrfeld W et al. (1988) LIGA process: Sensor construction techniques via x-ray lithography.
Technical Digest, IEEE Solid State Sensor and Actuator Workshop, 1–4.
Electrochemical impedance spectroscopy study of mixed thiol monolayers on gold, Xing, Y.F.
(Institute of Material Research and Engineering); Li, S.F.Y.; Lau, A.K.H.; O’Shea, S.J. ,
Journal of Electroanalytical Chemistry, v 583, n 1, Sep 1, 2005, p 124-132
Fan L et al. (2005) The development of functional endohedral metallofullerene materials.
Fullerenes Nanotubes and Carbon Nanostructures, 13:155–158.
Fan ZY, Lu JG (2005) Gate-refreshable nanowire chemical sensors. Applied Physics Letters,
86:123510.
Gallatin GM, Houle FA, Cobb JL (2003) Statistical limitations of printing 50 and 80 nm contact
holes by EUV lithography. Journal of Vacuum Science and Technology B: Microelectronics
and Nanometer Structures, 21:3172–3176.
Gamo K, Namba S (1984) Ion beam lithography. Ultramicroscopy, 15:261–269.
Gerlach AK, Guber G, Heckele AE, Herrmann M, Muslija A, Schaller T (2002) Microfabrication
of single-use plastic microfluidic devices for high-throughput screening and DNA analysis.
Microsystem Technologies, 7:265–268.
Goulet PJG et al. (2005) Surface-enhanced Raman scattering on dendrimer/metallic nanoparticle
layer-by-layer film substrates. Langmuir, 21:5576–5581.
Guckel H (1998) High-aspect-ratio micromachining via deep x-ray lithography. Proceedings of
the IEEE, 86:1586–1593.
Gutierrez-Rivera LE et al. (2006) Nanosieves fabricated by interference lithography and electro-
forming. Proceedings of SPIE – The International Society for Optical Engineering, Device and
Process Technologies for Microelectronics, MEMS, and Photonics IV, 6037:603710.
Haefliger D, Boisen A (2005) Contact printed masks for 3D microfabrication in negative resists.
Proceedings of the IEEE International Conference on Micro Electro Mechanical Systems
(MEMS), Proceedings of the 18th IEEE International Conference on Micro Electro Mechanical
Systems, MEMS 2005 Miami – Technical Digest, 556–559.
Hammond PT (2004) Form and function in multilayer assembly: New applications at the nanos-
cale. Advanced Materials, 16:1271–1293.
Han Li et al. (2005) Nanoparticle-structured sensing array materials and pattern recognition for
VOC detection. Sensors and Actuators, B: Chemical, 106:431–441.
Hernandez S et al. (2005) Electrodeposition of FePd alloy thin films and nanowires. 208th
Meeting of the Electrochemical Society – Meeting Abstracts, MA 2005-02:1221.
Holger B, Ulf H (2000) Hot embossing as a method for the fabrication of polymer high aspect
ratio structures. Sensors and Actuators, 83:130–135.
Ippolito SJ (2005) Highly sensitive layered ZnO/LiNbO3 SAW device with InOx selective layer
for NO2 and H2 gas sensing. Sensors and Actuators, B: Chemical, 111–112:207–212.
Jayatissa AH, Gupta T (2005) Fine-grain nanocrystalline tungsten oxide films for gas sensor
applications. Ceramic Transactions, Ceramic Nanomaterials and Nanotechnology III –
Proceedings of the 106th Annual Meeting of the American Ceramic Society, 159:119–124.
Jian L et al. (2003) SU-8 based deep x-ray lithography/LIGA. Proceedings of SPIE – The
International Society for Optical Engineering, 4979:394–401.
Juang Y-J, Lee LJ, Koelling KW (2002) Hot embossing in microfabrication. Part I: Experimental,
Polymer Engineering & Science, 42(3):539–550.
Kasahara H et al. (2005) Performances by the electron optical system of low energy electron beam
proximity projection lithography tool with a large scanning field. Journal of Vacuum Science
and Technology B: Microelectronics and Nanometer Structures, 23:2754–2757.
82 W.-C. Tian
Kim DS et al. (2006) An extended gate FET-based biosensor integrated with a Si microfluidic chan-
nel for detection of protein complexes. Sensors and Actuators B: Chemical, 117:488–494.
Kim HS et al. (2006) Low energy electron beam microcolumn lithography. Microelectronic
Engineering, 83:962–967.
Kim JB, Kim K-S (2005) Novel photobleachable deep UV resists based on single component non-
chemically amplified resist system. Macromolecular Rapid Communications, 26:1412–1417.
Kim JH et al. (2004) Fabrication of nanoporous and hetero structure thin film via a layer-by-layer
self assembly method for a gas sensor. Sensors and Actuators, B: Chemical, 102:241–247.
Klank H, Kutter JP, Geschke O (2002) CO2-laser micromachining and back-end processing for
rapid production of PMMA-based microfluidic systems. Lab on a Chip, 2:242–246.
Koba F et al. (2005) EPL performance in 65-nm node metallization technology and beyond.
Progress in Biomedical Optics and Imaging – Proceedings of SPIE, Emerging Lithographic
Technologies IX, 5751:501–508.
Kovacs GTA, Maluf NI, Peterson KE (1998) Bulk micromachining of Si. Proceedings of IEEE,
86:1536–1551.
Lapicki A et al. (2005) Functionalization of micro-hall effect sensors for biomedical applications
utilizing superparamagnetic beads. IEEE Transactions on Magnetics, 41:4134–4136.
Lee LJ, Madou MJ, Koelling K, Daunert W, Lai S, Koh S, Juang CG, Lu Y-J, Yu Y (2001) Design
and fabrication of CD-like microfluidic platforms for diagnostics: Polymer-based microfabri-
cation. Biomedical Microdevices, 3:339–351.
Liu G et al. (2006) Fabrication of 3D photonic crystal by deep x-ray lithography. Proceedings of
SPIE – The International Society for Optical Engineering, 6110:61100R.
Liu X, Brenner K-H, Wilzbach M, Schwarz M, Fernholz T, Schmiedmayer J (2005) Fabrication
of alignment structures for a fiber resonator by use of deep-ultraviolet lithography. Applied
Optics, 44:6857–6860.
Liu X, Du D, Mourou G (1997) Laser ablation and micromachining with ultrashort laser pulses.
IEEE Journal of Quantum Electronics, 33:1706–1716.
Lutkenhaus JL et al. (2005) Elastomeric flexible free-standing hydrogen-bonded nanoscale assem-
blies. Journal of the American Chemical Society, 127:17228–17234.
Mardare D et al. (2005) Chromium-doped titanium oxide thin films. Materials Science and
Engineering B: Solid-State Materials for Advanced Technology, Symposium D: Functional
Oxides for Advanced Semiconductor Technologies, 118:187–191.
Martinsson H et al. (2005) Current status of optical maskless lithography. Journal of
Microlithography, Microfabrication and Microsystems, 4:1–15.
McGeoch MW (2005) Maskless EUV lithography via optically addressed modulator. Progress in
Biomedical Optics and Imaging – Proceedings of SPIE, Emerging Lithographic Technologies
IX, 5751:349–354.
Menon R et al. (2005a) Maskless optical lithography using MEMs-based spatial-light modulators.
Progress in Biomedical Optics and Imaging – Proceedings of SPIE, MOEMS Display and
Imaging Systems III, 5721:53–63.
Menon R et al. (2005b) Maskless lithography. Materials Today, 8:26–33.
Metal-insulator-metal ensemble (MIME) chemical detectors, Smardzewski, R.R. (Geo-Centers,
Inc., Gunpowder Branch); Jarvis, N.L.; Snow, A.W.; Wohltjen, H.; Mackay, R.A. , 2004 NSTI
Nanotechnology Conference and Trade Show - NSTI Nanotech 2004, v 3, 2004 NSTI
Nanotechnology Conference and Trade Show - NSTI Nanotech 2004, 2004, p 163-164
Modi N, Lacy F (2005) Piezoelectric microcantilevers of nanoscale thickness for detection of
cells. 2005 NSTI Nanotechnology Conference and Trade Show – NSTI Nanotech 2005
Technical Proceedings, 2005 NSTI Nanotechnology Conference and Trade Show – NSTI
Nanotech 2005 Technical Proceedings, 1:501–504.
Myers BD et al. (2006) Variable pressure electron beam lithography (VP-eBL): A new tool for
direct patterning of nanometer-scale features on substrates with low electrical conductivity.
Nano Letters, 6:963–968.
Naulleau P et al. (2005) EUV microexposures at the ALS using the 0.3-NA MET projection
optics. Progress in Biomedical Optics and Imaging – Proceedings of SPIE, Emerging
Lithographic Technologies IX, 5751:56–63.
4 Integrated Micromachining Technologies for Transducer Fabrication 83
Park S et al. (2005) pH-sensitive solid-state electrode based on electrodeposited nanoporous plati-
num. Analytical Chemistry, 77:7695–7701.
Peterson KE (1982) Silicon as a mechanical material. Proceedings of IEEE, 70:420–457.
PZT stack etch for MEMS devices in a capacitively coupled high density plasma reactor”,
Werbaneth, Paul (Tegal Corporation); Almerico, John; Jerde, Les; Marks, Steve, Proceedings
of SPIE - The International Society for Optical Engineering, v 4592, 2001, p 267-273
Quate CF (1997) Scanning probes as a lithography tool for nanostructures. Surface Science,
386:259–264.
Rakhshandehroo MR, Weigold JW, Tian WC, Pang SW (1998) Dry etching of Si field emitters
and high aspect ratio resonators using an inductively coupled plasma source. Journal of
Vacuum Science and Technology B, 16:2849–2854.
Resnick DJ, Sreenivasan SV, Willson CG (2005) Step and flash imprint lithography. Materials
Today, 8(2):34–42.
Reyntjens S et al. (2001) A review of focused ion beam applications in microsystem technology.
Journal of Micromechanics and Microengineering, 11:287–300.
Robert Bosch GmbH (1990, 1991) Trench etch process for a single wafer RIE dry etch reactor,
U.S. Patent 4,855,017 and 4,784,720 (1991) and German Patent 4241045C1 (1990).
Rogner A et al. (1992) LIGA technique – What are the new opportunities. Journal of Micro-
mechanics and Microengineering, 2:133–140.
Rothschild M (2005) Projection optical lithography. Materials Today, 8:18–24.
Salib M, Michaeli A, Lazar A, Bechor H, Settle M, Krauss TF (2005) Photonic crystal fabrication
on silicon-on-insulator (SOI) using 248 nm and 193 nm lithography, Proceedings of SPIE –
The International Society for Optical Engineering, Nanophotonics for Communication:
Materials and Devices II, 6017:60170B.
Sayago I et al. (2005) Hydrogen sensors based on carbon nanotubes thin films. Synthetic Metals,
148:15–19.
Schaffer CB, Brodeur A, Garcia JF, Mazur E (2001) Micromachining bulk glass by use of femto-
second laser pulses with nanojoule energy. Optics Letters, 26(2):93–95.
Scheer HC, Schulz H, Lyebyedyev D (2001) Strategies for wafer-scale hot embossing lithography.
Proceedings of SPIE, 4349:86–89.
Schmidt RH, Haupt K (2005) Molecularly imprinted polymer films with binding properties
enhanced by the reaction-induced phase separation of a sacrificial polymeric porogen.
Chemistry of Materials, 17:1007–1016.
Sewell H et al. (2005) 32 nm node technology development using interference immersion lithog-
raphy. Progress in Biomedical Optics and Imaging – Proceedings of SPIE, Advances in Resist
Technology and Processing XXII, 5753:491–501.
Silverman PJ (2005) Extreme ultraviolet lithography: Overview and development status. Journal
of Microlithography, Microfabrication and Microsystems, 4:1–5.
Smardzewski RR et al. (2004) Metal-insulator-metal ensemble (MIME) chemical detectors. 2004
NSTI Nanotechnology Conference and Trade Show – NSTI Nanotech 2004, 3:163–164.
Smith BW et al. (2005) 25 nm immersion lithography at a 193 nm wavelength. Proceedings of
SPIE – The International Society for Optical Engineering, Optical Microlithography XVIII,
5754:141–147.
Solak HH (2006) Nanolithography with coherent extreme ultraviolet light. Journal of Physics
D: Applied Physics, 39:R171–R188.
Soltys J et al. (2003) Study of tip-induced Ti-film oxidation in atomic force microscopy contact
and non-contact mode. Acta Physica Polonica A, 103:553–558.
Steen SE et al. (2005) Looking into the crystal ball: Future device learning using hybrid E-beam
and optical lithography. Progress in Biomedical Optics and Imaging – Proceedings of SPIE,
Emerging Lithographic Technologies IX, 5751:26–34.
Steen S et al. (2006) Hybrid lithography: The marriage between optical and e-beam lithography.
A method to study process integration and device performance for advanced device nodes.
Microelectronic Engineering, 83:754–761.
Sun L et al. (2006) Preparation and gas-sensing property of a nanosized titania thin film towards
alcohol gases. Sensors and Actuators, B: Chemical, 114:387–391.
84 W.-C. Tian
Tachi S, Tsujimoto K, Arai S, Kure T (1991) Low temperature dry etching. Journal of Vacuum
Science and Technology B, 9:796–803.
Takeno T et al. (2005) Metal-containing diamond-like nanocomposite thin film for advanced
temperature sensors, Materials Science Forum. PRICM 5: The Fifth Pacific Rim International
Conference on Advanced Materials and Processing, 475–479:2079–2082.
Tanaka Y et al. (2005) Evaluation of pattern fidelity in EUVL using a high-numerical-aperture
small-field EUV exposure tool (HiNA). Progress in Biomedical Optics and Imaging –
Proceedings of SPIE, Emerging Lithographic Technologies IX, 5751:733–740.
Tian W-C, Pang SW (2001) Released submicrometer Si microstructures formed by one-step
dryetching. Journal of Vacuum Science and Technology B, 19:433–438.
Tian W-C, Pang SW (2002) Freestanding microheaters in Si with high aspect ratio microstruc-
tures. Journal of Vacuum Science and Technology B, 20:1008–1012.
Tian W-C, Pang SW (2003) Thick and thermally isolated Si microheaters for microfabricated
preconcentrators. Journal of Vacuum Science and Technology B, 21:274–279.
Tian W-C , Weigold JW , Pang SW (2000) Comparison of Cl 2 and F-based dry etching for highaspect
ratio Si Microstructures etched with an inductively coupled plasma source . Journal of
Vacuum Science and Technology B , 18 : 1890 – 1896 .
Ueno H et al. (1999) Study on fabrication of sub-micron structures for MEMS using deep X-ray
lithography. Proceedings of the International Symposium on Micro Machine and Human
Science, 87–92.
Vainos NA (2004) Laser grown photonic structures. Proceedings of SPIE – The International
Society for Optical Engineering, ROMOPTO 2003: Seventh Conference on Optics,
5581:1–11.
Weber W et al. (2005) Electron beam lithography for nanometer-scale planar double-gate transis-
tors. Microelectronic Engineering, Proceedings of the 30th International Conference on
Micro- and Nano-engineering, 78–79:206–211.
Wensink H, Berenschot JW, Jansen HV, Elwenspoek MC (2000) High resolution powder blast
micromachining. Proceedings of IEEE MEMS 2000, Miyazaki, Japan, pp. 769–774.
Yang X (2005) Hybrid exposure strategy: Combining e-beam direct writing with optical lithogra-
phy for magnetic recording heads. Journal of Vacuum Science and Technology B:
Microelectronics and Nanometer Structures, 23:2624–2630.
Yasin S et al. (2005) UVIII for combined e-beam and optical exposure hybrid lithography,
Microelectronic Engineering, Proceedings of the 30th International Conference on Micro- and
Nano-engineering, 78–79:47–50.
Zhao G et al. (2005) Direct electrochemistry and electrocatalysis of heme proteins immobilized
on self-assembled ZrO2 film. Electrochemistry Communications, 7:724–729.
Zhao Z et al. (2005) Annealing enhanced hydrogen absorption in nanocrystalline PdAu sensing
films. Journal of Applied Physics, 97:124301.
Zlobin VA (2005) Development of electron beam lithography for nanoscale devices, Proceedings
of SPIE – The International Society for Optical Engineering, Opto-Ireland 2005:
Nanotechnology and Nanophotonics, 5824:23–32.
Zribi A et al. (2005) Micromachined resonant multiple gas sensor. Sensors and Actuators,
A: Physical, 122:31–38.
Chapter 5
Applications of Functional Thin Films
and Nanostructures in Gas Sensing
Audrey Nelson
Introduction
The BCC report (BCC research, 2005) C-245 from last year (Gas Sensors and Gas
Metering: Applications and Markets, by Edward Gobina, PhD, Published April
2005), reported that the global market for gas sensors and gas metering equipment
including secondary instrumentation was estimated at $2.8 billion in 2004 and
expected to rise at an average annual growth rate (AAGR) of 5.9% to $3.8 billion
in 2009. This rapidly growing industry has been pushing gas-sensing technology to
produce more reliable and more accurate instruments that are both easy to use and
economical. As these goals are met, the market may well open up and grow at even
faster rates.
Thin-film technology plays a very important role in a variety of instrumentation
used for gas sensing, including microelectromechanical (MEMS) gas sensors and
A. Nelson
GE Sensing, TelAire, 6860 Cortona Dr Suite B, Goleta, CA 93117-5568
e-mail: Audrey.Nelson@ge.com
A. Zribi and J. Fortin (eds.), Functional Thin Films and Nanostructures for Sensors, 85
Integrated Analytical Systems,
DOI: 10.1007/978-0-387-68609-7_5, © Springer Science + Business Media, LLC 2009
86 A. Nelson
Introduction
Charles Fabry and Alfred Perot (Fabry-Perot Interferometer, 1889). It took 60 years
for the advantages of using the Michelson interferometer as part of a nondis-
persive spectrometer to be recognized and nearly as many years for the Fabry-Perot
interferometer.
In the early 1950s, Peter Fellgett and Pierre Jacquinot employed the Michelson
interferometer to measure all wavelengths simultaneously, achieving an improve-
ment factor known as the multiplex or Fellgett’s advantage. In addition the interfero-
meter did not need slits to separate frequencies as the dispersive spectrometer does,
and there was also less reflection loss because there were fewer mirror surfaces;
considerably more energy reached the radiation detector. The higher level of light
achieved a second improvement factor known as the throughput or Jacquinot’s
advantage. These improvements are the basis of the Fourier transform spectrometer
(FTS), which enabled, for example, the first measurement of near-infrared emission
from the night sky by J. Connes in 1960 (Connes 1961). Dispersive and nondisper-
sive spectrometers are widely used in laboratories for the qualitative and quantita-
tive analysis of both gases and liquids. The use of spectrometers in industry has
been limited by the delicacy of the instrument.
A much simpler type of nondispersive spectrometer uses a noninterferometric
optical filter. These are generally called NDIR (nondispersive infrared) systems and
are used in the measurement of concentrations of gases that are of great interest to
industry such as CO2, CO, H2O, and CxHy. The multiplex advantage in an NDIR
system is actually greater than that in an FTS, because in an NDIR device all wave-
lengths arrive in phase at the radiation detector, and so there is no signal power loss
due to destructive interference. NDIR systems have a large throughput advantage
because there are fewer angular restrictions on the radiation when it is directed
through the sample gas column than there are in interferometers and dispersive
spectrometers. Last, but not least, a nondispersive system with an optical filter does
not need prisms, gratings, or interferometers, and therefore the optical engine can
be rather simple. Such a system can be both easy to use and economical, and thus
the demand for these devices has rapidly expanded in the past few years. We focus
our attention on nondispersive gas sensor systems using optical filters.
Radiation
called “Planck’s constant” = 6.63 × 10−34 J s. Thus, the quantum energy of a photon
is also simply related to the inverse of the wavelength, with E (electron volts or eV)
= 1.24/l (microns or µm).
Each portion of the electromagnetic spectrum has quantum energies appropriate
for the excitation of certain types of physical processes. The energy levels for all
physical processes at the atomic and molecular levels are quantized, and if there are
available quantized energy levels in a particular form of matter that match the quan-
tum energy of the incident radiation, then the material will absorb that radiation,
and go to a higher energy state.
Gas molecules have a number of natural frequencies, which correspond to exci-
tations of quantized vibrations or rotations of the molecules. For a given molecule
and bonding structure, the natural frequencies and hence the photon energies that
may be absorbed are always the same. The magnitudes of absorption energies are
very low, in the tenths of electron volts, and so radiation with wavelengths in the
range of 2–20 µm (mid-IR range) are absorbed by molecules.
Common gases such as water vapor (H2O), carbon dioxide (CO2), carbon monoxide
(CO), and hydrocarbons (CxHy), have many absorption lines in the mid-infrared
region. Fig. 5.2 shows the portion of the absorption spectra for some of these gases
between wavelengths of 4.2 and 5.6 µm. We see for each gas there are groups of
absorption lines known as vibration–rotation bands. For example, the absorption
spectrum of CO2 has a vibration–rotation band around 4.26 µm. The spacing
between distinct rotational states is generally smaller than the spacing between
vibrational states by approximately a factor of (m/M) 1/2 » 0.003, where m is the
mass of the electron, and M is the mass of the molecule.
Absorption Law
When infrared light passes through a gas, the gas has a high probability of absorb-
ing the light if the wavelength of the light matches a wavelength in the vibration–
rotation band.
5 Applications of Functional Thin Films and Nanostructures 89
8E-10
6E-16
Mid IR
4E-16
2E-16
0 Wavelength
0 1 2 3 4 5 6 7 8 9 10 in microns
Visible
100
Transmittance%
90
80
70 H2O
CO
60 N2O
50 CO2
40 Wavelength
4.2 4.4 4.6 4.8 5 5.2 5.4 5.6 in microns
Transmittance spectra for N2O, CO, CO2,
and H20 vapor at 100 ppm-meters concentration
1.2 0
T(L)
dT/dC(L) −0.000005
1/ppm
0.6 −0.00002
−0.000025
0.4
−0.00003
0.2
−0.000035
0 −0.00004
0 25 50 75 100 125 150 175 200
Path length in mm
gas concentration, and between these extremes the magnitude of dT/dC peaks giv-
ing the optimal sensitivity when:
aCL = 1; L = 1 / (aC ). (5.4)
In practical applications, the light beam is not monochromatic, and the absorption
coefficient α varies as a function of wavelength. The light intensity transmitted
through a target gas of concentration C is then described by an integral of many
exponentials:
∫ i (l )e
( − a ( l )CL )
I (C ) = 0 dl , (5.5)
where i0 (l) is the intensity density per unit wavelength at wavelength l, and
i0 (l) = dI0/dl.
In principle, if we know α as a function of l, we can perform this integral.
In practice we can experimentally measure the ratio I(C)/I(0), where I(0) is the
transmitted intensity for zero target gas concentration, and I(C) is the transmitted
intensity as a function of target gas concentration C. I(C)/I(0) does not follow a
simple exponential as a function of C, but one can find a best numerical fit to the
experimental data to enable the extraction of C from a measurement of I(C)/I(0). In
some cases, I(0) may be estimated from an intensity measurement in a wavelength
band where α(l) is negligible. This technique allows I(0) to be measured simulta-
neously with I(C) in one gas volume.
5 Applications of Functional Thin Films and Nanostructures 91
Introduction
There are two types of mid-IR sources: quantum sources and thermal sources.
Quantum sources include lasers and LEDs. Early mid-IR laser sources are lead–
salt diode lasers (LSDL), made from IV–VI semiconductor materials, which operate
in the 3–30 mm spectral region. Because the LSDL requires a cryogenic cooling
system, it limits the range of uses. In recent years as the thin-film deposition tech-
nology underwent rapid development, many compact laser devices containing
antimonide compounds such as AlGaAsSb, InGaAsSb, and InAsSbP have been
developed for operation at higher temperatures. In 1994 Bell Labs developed the
first quantum cascade laser (QCL), which made a mid-IR room temperature operat-
ing laser possible (Faist et al. 1994; Gmachl 1998).
Because most laser sources are monochromatic, systems employing them do not
need an optical filter. The advantage of the laser source is high power (mW~W),
high sensitivity (ppb~ppt level), and very fast response time (~ns). However, most
Sample Cell
Reference Cell
Detector IR Source
laser sources need cooling systems and are often complex to implement because of
stability issues. QCLs are not quite mature and are still too expensive for most
industrial applications. Mid-IR LEDs that do not require cooling are also scarce.
We discuss LED sources in a later section in more detail.
Thermal sources, such as incandescent lamps, are still most commonly used in
NDIR systems in industry. In general, thermal sources have two main shortcom-
ings: the first is their broadband radiation, which results in wasted energy because
only a narrow band is absorbed by the target gas; the second is their slow response
time which precludes fast modulation. For fast modulation with a thermal source,
one must use an optical chopper in the system configuration.
There are several reasons why modulation of the light source is desirable. The
most important one is that an alternating signal has a built-in zero reference. This
helps minimize the sensitivity of the offset drift and also is especially useful for
small AC signals in the presence of large DC offsets. Furthermore, filtering is easier
to implement into an AC signal by tuning the amplifier to pass the modulating
frequency and to reject all other frequencies, allowing one to reduce the noise level
considerably.
Because thin-film technology is undergoing rapid development, more and more
micromachined IR lamps and miniature blackbody sources (Axetris) have been
developed. They provide a mechanical stability advantage over lamps. Because the
source element can be made with membranes a few microns thick of very low ther-
mal mass, these sources often have fast response times.
The waveguide is part of the optical system, which may also include optical
mirrors or lenses to guide the light path. Simple waveguides can be made of either
metal or plastic with a few thin-film layers coating the inside of the waveguide.
Gold layer finishing is always desirable because gold has the best reflectivity for IR
light. The main purpose of the waveguide is to guide the radiation along the proper
path to the detector while minimizing the loss of radiation energy. In many simple
systems the waveguide is combined with a gas sampling chamber.
Detector
There are two types of mid-IR detectors: quantum detectors also known as
photodetectors, and thermal detectors. A quantum detector responds to individual
photons, which are the quanta of radiation. A thermal detector responds to tempera-
ture changes caused by the radiation power incident upon it.
There are many types of thermal detectors available, including thermopiles,
pyroelectric, and bolometer. The most desirable features of thermal detectors are
the high linearity of their signal as a function of the infrared radiation power and
their uniform response independent of wavelength.
A thermopile is an array of thermocouples connected in series. A thermocouple
is a pair of junctions of dissimilar metals that produce voltage when one side of the
junctions has a different temperature from the other. The so-called reference junc-
tion is kept at a known temperature by bonding it to a mass with a stable temperature.
5 Applications of Functional Thin Films and Nanostructures 93
λ
Life-time Modula- Vendor
Sources Range Photo
(h) tion
(µm)
International
Light
IR lamp 40,000 <2 Hz 0.2–5
Technologies
[ILT]
Nano-
Thermal Intex
amorphous 25,000 <100Hz 1 – 20
source [Intex]
carbon Films
Surface ICx
textured Photonics
25,000 <10 Hz 2 – 20
thin metal [ICx
filament Photonics]
Laser
Lead-salt
3 – 30 Components
Diode 10,0001 10kHz2
[Laser Com-
Lasers
ponents]
Roithner
Quantum LaserTech-
Laser Antimonide
2,0003 10KHz niK GmbH
Lasers
2–4 [Roithner
Laser]
Alpes
Quantum
Lasers
Cascade 7,000 10KHz [Alpes
Lasers 4 – 12 Lasers]
1
The Laser Guidebook by Jeff Hecht, Chapter 19, p. 280.
2
Laser sources are capable of modulation frequencies as high as a GHz, but for
gas-sensing uses, practical issues reduce the effective modulation frequency to the
kHz range.
3
“Midinfrared interband cascade lasers at thermoelectric cooler temperatures”
Electronics Letters Volume 42, 1034 (2006).
detectors present a great challenge, because of the limitations of the narrow band-
gap materials employed. Typical bandgaps are around ¼ eV, which is not too far
from the typical room temperature thermal energy kT ~ 0.025 eV. In Table 5.2, brief
characteristics of commonly available IR detectors and manufacturers are listed.
More detailed data such as normalized detectivity, time constant, resistance or
impedance, responsivity, noise spectrum, and spectral response are available from
manufacturers’ datasheets. One can choose the proper detector for a specific system
to match the required operating temperature, wavelength region, and other impor-
tant characteristics.
5 Applications of Functional Thin Films and Nanostructures 95
Filter
An NDIR system with a broadband radiation source often uses an optical narrow
bandpass filter, which is a component with a wavelength-dependent transmission or
reflectivity. The optical filter allows only wavelengths in the desired gas absorption
band to reach the detector. This increases the signal-to-noise ratio and minimizes
the effects of interfering gases, source variability at uninteresting wavelengths, and
broadband noise.
Although antireflection coating techniques had been developed in the early
1930s, and techniques for design of thin-film narrowband optical filters were well
established by the 1970s, the theory and techniques of layer thickness control were
apparently rather poorly developed. Over the last few decades, thin-film growth
technology and equipment have advanced dramatically. Now reliable narrow band-
pass filters with uniform response are available from many optical companies.
Depending on the shape of the transmission curve, one distinguishes filters of
the following types (Table 5.3).
There is a variety of designs of narrowband thin-film filters. The most common
type is based on the Fabry–Perot interferometer, which consists of two highly
reflecting mirrors with a half-wave dielectric layer in the middle, forming a
96 A. Nelson
standing-wave cavity. Incident light passes through two coated reflecting surfaces.
The distance between the reflective coatings determines which wavelengths will
destructively interfere and which wavelengths will be allowed to pass through the
coated surfaces of the optical bandpass filters. In situations where the reflected
beams are in phase, the light will pass through the two reflective surfaces.
However, if the wavelengths are out of phase, destructive interference will block
most of the reflections, allowing almost nothing to be transmitted through the
optical bandpass filters. In this way, interference optical bandpass filters are able to
attenuate the intensity of transmitted light at wavelengths that are higher or lower
than desired. This structure can be made of metal-dielectric layers (where the
dielectric layer forms a spacer bounded by two metallic reflecting layers), all-die-
5 Applications of Functional Thin Films and Nanostructures 97
1.76 100
Percent transmission
80
60
Sapphire 0.1-5.0 40
20
0
1 .2 .3 .4 .5 .6 .7 .8 .9 1.0 2.0 3.0 4.05.0 10 20
Wavelength in microns
1.47 100
Percent transmission
80
60
BaF2 0.2-11 40
20
0
1 .2 .3 .4 .5 .6 .7 .8 .9 1.0 2.0 3.0 4.05.0 10 20
Wavelength in microns
3.42 100
Percent transmission
80
60
Silicon 1.1-10 40
20
0
1 .2 .3 .4 .5 .6 .7 .8 .9 1.0 2.0 3.0 4.0 5.0 10 20
Wavelength in microns
100
2.49
Percent transmission
80
60
40
ZnSe 0.6 – 20
20
0
1 .2 .3 .4 .5 .6 .7 .8 .9 1.0 2.0 3.0 4.0 5.0 10 20
Wavelength in microns
1.54 100
Percent transmission
80
60
40
KBr 0.5 - 25
20
0
1 .2 .3 .4 .5 .6 .7 .8 .9 1.0 2.0 3.0 4.0 5.0 10 20
Wavelength in microns
lectric layers (a solid thin-film spacer with all dielectric multilayers on either side),
or a solid etalon plate (a cleaved crystal spacer with thin-film reflector deposited on
either side).
Most available midinfrared narrow bandpass filters are multiple cavity filters
that consist of a stack of static Fabry–Perot filters. A double cavity (structure such
as |reflector|half-wave|reflector|half-wave|reflector|) or triple cavity are very commonly
used in midinfrared narrow bandpass filters. The advantage of the multiple cavity
filters is a significant improvement in the shape of the transmission curve, the
desired wavelengths are transmitted with higher efficiency, and there is a sharper
and stronger cutoff of undesired wavelengths.
Many materials have been used to create thin-film coating and filter material.
Table 5.4 lists some of the common IR materials and their characteristics.
98 A. Nelson
Introduction
Telaire (Telaire) is one of the world leaders in NDIR gas-sensing technology and is
a pioneer in the development of low-cost NDIR systems. Here I give a real example
of Telaire’s common NDIR system and then discuss mid-IR LED development
using high precision thin-film deposition molecular beam epitaxy (MBE).
Custom Designed
Infrared Filters
Reference Target Gas Dual Beam
Diffusion Membranes Micro-Machined
Thermopile Detector
Incandescent 1.25"
Infrared Source Patented Waveguide
Microprocessor
• Lamp modulation: By modulating the lamp power the circuitry can use AC
signal processing to reduce any background radiation bias, detector drift, or
electronic drift.
• Waveguide: By making the gas-sampling chamber into an optical waveguide, it
is possible to both minimize light losses and optimize beam path length to fit the
gas concentration range. The longer the path length, the more molecules of
target gas will be in the path between the light detector and the source. Therefore,
low-concentration range sensors are designed with longer path lengths than
high-concentration range sensors.
• ABC logic: Crowning the above design is a patented (U.S. patents) Ambient
Background Calibration (ABC) tool for CO2 sensors. This calibration tool uses the
fact that the CO2 content in the atmosphere is quite stable to automatically correct
for any long-term drift by comparing the sensor readings with ambient back-
ground during any regular (e.g., daily) quiet periods where the sensor is used.
NDIR systems such as the one described have limitations. Because the lamp
consumes a significant amount of power it can’t be used in certain applications
requiring extremely low power. The lamp also has a slow response time, and so it
can’t be used when a fast response time is necessary. Furthermore, lamps have
stability issues. One solution is to replace the lamp with an infrared LED source.
IR LED development has a long history. A few pioneering groups in this field are
at the Russian Ioffe Institute and at Lancaster University in the United Kingdom.
There are frequently efforts reported in the international conference on Mid-Infrared
Optoelectronics: Materials and Devices (MIOMD).
The fundamental difficulty of mid-IR semiconductor devices arises from the small
energy (in the tenths of electron volts) of mid-IR photons, so the appropriate materials
for this energy range have a small bandgap. Consequently Auger recombination,
where electron-hole pairs recombine and transfer their energy and momentum to
another carrier instead of emitting a photon, becomes important. At room temperature
and above, Auger recombination becomes dominant in the mid-IR. Small bandgap
materials also often have large spin-orbit interactions. In certain materials (e.g., InAs
and InSb) energy splitting due to spin-orbit interactions becomes comparable to the
bandgap itself, resulting in heavy intervalence band absorption.
These fundamental limitations have prolonged mid-IR room temperature LED
development for decades. So far, mid-IR LEDs are still commercially very scarce.
In order to develop a mid-IR LED that works at room temperature, Telaire has
explored some semiconductor heterostructure LEDs. The problem with these LEDs
is a spectrum shift due to temperature variation. Instrumentation is thus extremely
difficult because of the lack of stability over the desired temperature range. We
therefore put effort into the resonant cavity (RC) LED. The RCLED was first
100 A. Nelson
introduced and developed by Dr. Schubert in 1990 (Schubert et al. 1992). There are
many advantages of an RCLED compared with conventional LEDs, such as high
spectral purity, narrow angular distribution, high brightness, and high efficiency.
The emission line width of an RCLED is determined by the cavity quality fac-
tor Q, not by the thermal energy kT. Therefore, one can design ideal emission line
widths with respect to a target gas spectrum. At the same time, the RCLED has
much better temperature stability compared with conventional LEDs because the
wavelength shift with temperature is determined by the thermal expansion coef-
ficient of the optical cavity and not by the energy gap of the active material.
The performance of the device is critically dependent on two factors. The first is
the spectral overlap between the spontaneous emission peak from the active layers
and the cavity resonance, and the second is the match between the backside mirror
reflection and the cavity resonance. Therefore, high-precision growth techniques
are required. Molecular beam epitaxy (MBE) of thin-film deposition easily
provides the necessary precision.
MBE technology was developed by A. Y. Cho and his colleagues at Bell Labs
in late 1970s1994. “The unique feature of MBE is the ability to prepare single
crystal layers with atomic dimensional precision” (Alfred ).
An MBE is an epitaxial process where source material, such as gallium, indium,
and arsenic, are evaporated onto a heated crystal substrate inside an ultrahigh
vacuum chamber with pressure less than 10-6 torr. The vapors from the heated sources
form beams that travel through the chamber and land on the substrate surface in a slow
deposition rate (typical around 1 µm/h for most III-V materials), arranging themselves
in a crystalline lattice based on that of the substrate. During the operation, reflection
high-energy electron diffraction (RHEED) is used for monitoring the growth of the
crystal layers. A computer controls shutters in front of each furnace, allowing precise
control of the thickness of each layer, down to a single layer of atoms.
So far most RCLED development has been in the visible and near IR region.
Mid-IR RCLED has certain challenges with its material system. However, for the
typical device with a few micron thicknesses, the cavity length fits the lowest orders
of the optical modes for mid-IR wavelengths. This promises a better overlap
between the resonant optical mode and the active region emission spectrum.
In 2004 Telaire developed a unique CO2 RCLED (Telaire RCLED) with an
emission spectrum that matches the CO2 absorption profile very well, as shown in
Fig. 5.6. Further narrowing of the LED spectral distribution can be achieved by
increasing the reflectivity of the elements of the resonant cavity.
Because of MBE’s ability for precise control, we have successfully grown our
RCLED structure1 to fit perfectly with the CO2 absorption spectral profile.
1
Hughes Research Laboratory performed the MBE growth of the RCLED structure.
5 Applications of Functional Thin Films and Nanostructures 101
1.8E-05
Ambient
N2 Purge
1.4E-05
Intensity (a.u.)
1.0E-05
6.0E-06
2.0E-06
3 3.5 4 4.5 5 5.5
Wavelength (um)
Summary
With the LED light source a variety of systems can be assembled. Table 5.5 shows
a comparison of the response time and power consumption of systems containing
an LED light source, as well as a typical system with an IR lamp for comparison.
In conclusion, thin-film technology has played a crucial role in NDIR systems.
Because of its fast development, one can imagine that many new systems will
emerge with faster and more reliable sources and detectors. As increasingly
102 A. Nelson
complex integrated circuits become available, sensors on the chip with smaller size
and better performance will soon become a possibility. Gas-sensing technology will
flourish with the many new NDIR devices.
References
Radislav A. Potyrailo
Abstract Chemical sensors for diverse applications for gas- and liquid-phase sensing
have their own design requirements. Thus, sensors typically have long timelines
from the concept through the evolution and cost reduction to commercial products.
For some applications, it is attractive to take advantage of previously developed,
optimized, and mass-produced physical transducers, optoelectronic, radiofrequency
identification, and other types of components and to rationally combine them with
sensing materials to produce new types of chemical sensors, more rapidly than it is
typically achieved. Widely deployed and accepted commodity consumer products
present a striking set of attractive capabilities applicable for advanced sensors.
This chapter presents several recent examples from our laboratory to demonstrate
developments in chemical sensors based on electrical, mechanical, and radiant signal-
transduction methodologies.
Introduction
Chemical sensors have found their niche among modern analytical instruments
when real-time determination of the concentration of specific sample constituents
is required. Development of sensors with new capabilities is driven by the ever-
expanding monitoring needs for determination of a wide variety of species in gases
and liquids. More sensitive sensors are required for analysis of ultratrace levels of
environmental pollutants, and pathogenic species in water. Sensors with faster
response are highly desired for monitoring of transient in vivo events and bedside
patients. More selective sensors are wanted for analysis of ambient urban or
battlefield air. For these and many other reasons, a wide variety of chemical sensors
and sensor systems is being developed.
R.A. Potyrailo
GE Global Research Center, 1 Research Circle, K1 3B39A, Niskayuna, NY 12309
e-mail: potyrailo@research.ge.com
A. Zribi and J. Fortin (eds.), Functional Thin Films and Nanostructures for Sensors, 103
Integrated Analytical Systems,
DOI: 10.1007/978-0-387-68609-7_6, © Springer Science + Business Media, LLC 2009
104 R.A. Potyrailo
Data processing
Transducer
Radiant energy
Sample Sensing Electrical energy Species identity
introduction material Mechanical energy Species concentration
Thermal energy
A B C D
Sample components
Radiant Polymeric film formulated Container-integrated pH, dissolved O2, pCO2 Fluorometrix Corp., Stow, MA, USA
with organic disposable patch www.fluorometrix.com
fluorescent dye
Radiant Metallo-porphyrin dye Colorimet-ric sensor Volatile organic compounds ChemSensing, Inc., Northbrook, IL,
on solid supports film array USA, www.chemsensing.com
Thermal Formulated catalytic Catalytic combustion Combustible hydrocarbons Delphian Corp., Northvale, NJ, USA
composition sensor www.delphian.com
Thermal Formulated catalytic Catalytic combustion Combustible higher Nemoto & Co, Milan, Italy
composition sensor hydrocarbons www.nemototech.com
105
106 R.A. Potyrailo
A wide variety of innovative ideas for sensors with new capabilities is being
originated by coupling new technological capabilities in micro- and nanofabrica-
tion of transducers (Hagleitner et al. 2002, 2003; Hierlemann and Baltes 2003;
Piner et al. 1999; Suzuki 2000) with new inorganic and polymeric sensing materials
(Adhikari and Majumdar 2004; Akmal and Usmani 1998; Frantzen et al.2004;
Janata and Josowicz 2002; Koinuma and Takeuchi 2004; McQuade et al.2000;
Potyrailo 2006; Scheidtmann et al.2005; Wolfbeis 2006) and with innovations in
sample manipulation (Erickson and Li 2004; Hansen et al.2002; Peterson 2005;
Tani et al.2004; Thorsen et al.2002; Vilkner et al.2004). Combinatorial and high-
throughput materials screening (Archibald et al. 2002a,b; Jandeleit et al.1999;
Koinuma and Takeuchi 2004; Maier et al.2002; Potyrailo and Amis 2003;
Potyrailo et al. 2004a; Potyrailo and Takeuchi 2005) and data-handling method-
ologies (Grate 2000; Jurs et al. 2000; Potyrailo et al.2004d) facilitate additional
improvements in sensor designs. Obviously, the design of a sensor for a particular
application will be dictated by the nature and requirements of that application.
Nevertheless, it is useful to set down the features that one would wish of an ideal
sensor for chemical species.
Ideally, the sensor would provide both adequate sensitivity (in some cases at the
single-molecule level) and a broad dynamic range. High selectivity towards the
species of interest and immunity to sample-matrix interferences should also exist.
In addition, an ideal sensor would be suitable for multicomponent measurements,
have fast and reversible response, and excellent long-term stability. Furthermore,
the ideal sensor must be robust, reliable, simple, economical to fabricate, of small
size, and with self-calibration capabilities. Of course, this “ultimate” sensor is far
from what is currently available. The most important respects in which existing
chemical sensors fall short are insufficient long-term stability, matrix interferences
(limited selectivity), and inadequate detection limits. Also, many modern analytical
sensing problems cannot be solved because of an absence of adequate detection
schemes for certain analytes and the practical challenges of distributed mapping of
analyte species over large remote areas or in vivo.
It is useful to note, that in real-world applications, the qualities of an ideal
sensor are often weighted differently according to application (Mitchell 1995;
Pickup and Alcock 1991). For example, reliability, long-term stability, and reso-
lution top the priority list for industrial sensor users whereas the size, cost, and
maturity of the technology are the least important factors. In contrast, medical
users focus on cost for disposable sensors (Mitchell 1995). The importance of
continuous monitoring also differs from application to application. For instance,
glucose sensing would still be acceptable if performed intermittently at a
frequency of 0.5–6 h whereas blood-gas sensors for use in intensive care should
be capable of continuous monitoring (Pickup and Alcock 1991). Specific
requirements for in vivo sensors include blood compatibility and sterilizability
(Meyerhoff 1993).
6 Chemical Sensors: New Ideas for the Mature Field 107
Table 6.3 Examples of Challenges and Sensor-Design Aspects Associated with Gas- and Liquid-Phase Chemical Sensing
Gas-Phase Chemical Sensing Liquid-Phase Chemical Sensing
Sub-systems Challenges Sensor-Design Aspects Challenges Sensor-Design Aspects
Sample – Particulate contamination – Preconcentration for detection limit – Corrosion – Protective coatings
introduction – Lack of representative improvement – Scale deposition – Replaceable sampling
sample for trace analysis – Lack of representa- components
– Time-domain modulation for
selectivity improvement – tive sample for trace
Integration into periodic self-clean- analysis
ing sensor system
Sensing material – Film poisoning – Temperature-stabilized operation – Reagent leaching – Covalent reagent attachment
– Water condensation (T > ambient) – Film poisoning to matrix
– Temperature, gas-flow modulation to – Film delamination – Cartridges of disposable sensors
facilitate more reversible and selective
response
Transducer – Corrosion – Transducer design for higher-order – Corrosion – Transducer design for higher-
– Signal-to-noise decrease with sensor response through hyphenated – Signal-to-noise order sensor response through
reduced transducer size transduction techniques and time-, decrease with reduced hyphenated transduction tech-
– Sophisticated readout with temperature-, sample- transducer size niques and time-, temperature-,
reduced transducer size flow-modulation methods for selec- – Sophisticated readout sample-flow-modulation methods
tivity, signal-to-noise, with reduced for selectivity, signal-to-noise,
stability improvement transducer size stability
improvement
Data – Reliable quantification of – Analysis of dynamic signatures for – Reliable quantification – Analysis of dynamic signatures
analysis four or less analytes in their selectivity and stability of four or less analytes for selectivity and stability
mixtures with sensor array improvement in their mixtures with improvement
of partially selective sensing – Analysis of multivariate signatures sensor array of partially – Analysis of mltivariate signa-
films and a single transduc- for selectivity, signal-to-noise, selective sensing films tures for selectivity, signal-to-
tion principle and stability improvement and a single transduc- noise, and stability
tion principle improvement
R.A. Potyrailo
6 Chemical Sensors: New Ideas for the Mature Field 109
Existing sensors typically have long timelines from the concept through evolution
and cost reduction to commercial products. For example, micromachined pressure
sensors first were contemplated in the mid-1950s and became common in automo-
tive applications only by the late 1980s. Similarly, micromachined accelerometers
were invented in the mid-1970s, however, it was not until the early 1990s when they
became common in automotive applications (Hagleitner et al. 2002). Likewise,
chemical sensors take many years from the concept to a commercially successful
product (Wohltjen 2006). Thus it has been recognized that it is attractive to take advan-
tage of previously developed, optimized, mass-produced, and thus cost-effective
physical transducers (Boussaad and Tao 2003; Ren et al. 2005), optoelectronic (Cho
and Bright 2001; Cho et al. 2002; Ivanisevic et al. 2001; Manzano et al. 2003;
Vo-Dinh et al. 1999), radiofrequency identification (Finkenzeller 2003; Potyrailo
2006; Potyrailo and Morris 2007), and other types of components and to combine
them rationally with sensing materials to produce new types of chemical sensors,
more rapidly than is typically achieved. As an attractive complementary approach,
widely deployed and accepted commodity consumer products present a striking set
of capabilities applicable for advanced sensors (see Fig. 6.2)
Radiofrequency identification (RFID) tags have been recognized as one of the
disruptive technologies (Christensen 1997; Christensen and Raynor 2003) and are
widely used ranging from detection of unauthorized opening of containers, to auto-
matic identification of animals, to tracking insects, and to tagging of garments,
mailing labels, and combinatorial chemistry reaction products (Finkenzeller 2003;
Nicolaou et al. 1995; Wang et al. 2006). Part of the attractiveness of RFID tags is a
function of their low cost. For example, the cost of passive RFID tags that operate
at 13.56 and 915 MHz has dropped from $0.5–0.7 in 2003, to $0.04–0.4 in 2005,
and to $0.02–0.05 in 2006 (Bachner 2005; Lawrence 2005)
However, it was recently demonstrated that a multianalyte chemical identification
and quantitation can be performed using conventional RFID tags as transducers of
electrical energy (see Fig. 6.2a). Unlike other approaches where a special tag is
designed at a much higher cost (Mascaro et al.2007; Nambi et al.2003; Want 2004),
this approach utilizes a conventional RFID tag that is coated with a chemically sensi-
tive film. In such an RFID chemical sensor, both the digital tag ID and the complex
impedance of the antenna are measured. The measured digital ID can provide infor-
mation about the sensor and the object onto which the sensor is attached.
Highly efficient transducers of mechanical energy are also available as mass-
produced components. For example, it was demonstrated that micromachined
wristwatch tuning forks can be used as efficient chemical sensors when combined
with nanofabricated polymer wires for detection of a variety of vapors at part per
billion concentrations (Boussaad and Tao 2003; Ren et al.2005) as shown in Fig.
6.2b. Unlike the commonly used cantilever techniques, the fork converts the force
directly to an electrical signal without involving intermediate optical displacement
signal readout. This eliminates extra sources of noise, simplifies the instrumenta-
tion, and, allows a compact design for device applications.
110 R.A. Potyrailo
2 3
polymer wire
illumination sequence 2
Fig. 6.2 Application of ubiquitous commodity electronic components and consumer products for
creation of chemical sensors. (a) Examples of 13.56 MHz passive RFID tags (with measured tag
type and tag ID) adapted for chemical sensing: (1) roll of RFID tags applicable for the roll-to-roll
deposition of chemically sensitive films, tag type = ISO 15693, ID of one of tags = E007 0000
0260 4AE8; (2) RFID tag on a plastic nonflexible support; tag type = I * CODE1, tag ID = 0900
000 3C8A 7AFA; (3) RFID tag on a plastic nonflexible support; tag type = I * CODE1, tag ID =
0900 000 457D 5E12. Scale bars are 10 mm. (b) Micromachined wristwatch tuning fork with an
attached nanofabricated polymer wire for chemical sensing. (c) Micromachined accelerometer
originally developed for automotive applications and coated with a sensing film for chemical sensing.
(d) Generated colorimetric patterns using a conventional computer screen for illumination of an
array of colorimetric sensing films. (e) Compact disk optical pickup head as a separate detection
unit outside an optical disk drive for analytical measurement. (f) Conventional optical disk drive
in a laptop computer for the quantitative chemical detection using chemical sensor films deposited
on conventional CD and DVD optical disks and extraction of analog signal from optical disk drive.
((a)–(f) are adapted from Lange et al. 2006, Manzano et al. 2003, Potyrailo et al. 2006, Potyrailo
and Morris 2007, Potyrailo et al. 2006, and Ren et al. 2005, respectively.)
consumer products that are widely employed in the office and home are attracting
significant attention for optical sensor applications due to their cost advantage over
analytical instruments produced only in small quantities, robustness in operation
due to the detailed manufacturability improvements, and ease of operation (see
Fig.s 6.2d–f).
Computer screens have been applied for the illumination of colorimetric assays
(Filippini et al.2006; Manzano et al. 2003). Flatbed scanners were first applied to
analytical chemistry in the 1980s (Hruschka et al.1983) and continue to receive
attention today for sensor applications (Nath and Chilkoti 2002; Rakow and Suslick
2000; Taton et al.2000; Zhang and Suslick 2005). Handheld digital color analyzers
have been employed to quantitation of colorimetric sensor films (Hirayama et al.
2000; Suzuki et al. 2002). Compact disk optical pickup heads as separate detection
units outside optical disk drives have been used for a variety of measurement appli-
cations including scanning optical microscopy (Benschop and Rosmalen 1991),
position sensing (Chu and Lin 2005; Quercioli et al. 1999), and biodetection (Lange
et al. 2006). Applications of modified computer optical disk drives have been reported
for transmission measurements (Gordon 1999). Error determination routines in an
optical disk drive have been used to detect the presence of biological molecules and
bacteria on the disk surface (Jones 2005; Jones and Thigpen 2005; La Clair and
Burkart 2003). Finally, an attractive application of conventional computer optical
disk drives for chemical sensing has been demonstrated with a straightforward
extraction of an analog signal from a drive for the quantitative detection of optical
changes of chemical sensor films deposited on conventional CD and DVD optical
disks (Potyrailo 2006; Potyrailo 2006).
These examples conclusively demonstrate the power of implementation of existing
microfabricated, optoelectronic, and other consumer components to accelerate the
development of sensors with previously unavailable capabilities, functionality, or
level of acceptance.
Case Studies
(a)
Sensing film
RFID sensor
RF
CF Antenna
Sensing film
Inert substrate
(b)
Antenna
RA
RFID Micro-
LA CA
Reader chip
RFID tag
(c)
Fp
Zp
Zre
F2
Zim
F1
Frequency
Fig. 6.3 Strategy for the application of conventional passive RFID tags for chemical sensing. (a)
Adaptation of a conventional RFID tag for chemical sensing by deposition of a sensing film onto
the antenna. (Inset, analyte-induced changes in the film material affect film resistance (RF) and
capacitance (CF) between the antenna turns). (b) Schematic of digital ID reading from the RFID
tag and the equivalent circuit of the antenna of the RFID tag. (c) Measured parameters from a
single RFID sensor for multicomponent chemical detection and quantitation. Arrows indicate
changes of respective measured parameters
In our developed RFID sensor, the antenna coil not only completes its normal
activities, but is also tasked to act as an analytical electrode. This tasking of the
RFID antenna as an analytical electrode is possible because the complex impedance
of the RF resonant circuit formed by the antenna is sensitive to the dielectric
constant of any materials on or within the environment around the antenna.
114 R.A. Potyrailo
Consequently, if the dielectric constant of the local environment around the antenna
is adjusted, it will affect the complex impedance of the antenna circuit through the
changes in film resistance (RF) and capacitance (CF) between the antenna turns (Fig.
6.3a, inset). To provide for selective changes in the local dielectric field, a sensing
film must be applied to the antenna. Resultantly, only analytes that are selected by
the film will be incorporated into the local electrical environment and produce a
useful signal from the RFID sensor.
It is critical to match the response mechanism of the sensing film with the trans-
duction principle of the sensor. For selective detection of polar vapors with the
RFID sensor, we selected a solid polymer electrolyte film such as Nafion. This
polymer is of a family of perfluorosulfonated polymers, and is a copolymer of
tetrafluoroethylene and sulfonyl fluoride vinyl ether (see Scheme 6.1). Nafion is a
glassy polymer at room temperature with the glass transition temperature of its acid
form at ~110°C and of its salts at ~220°C (Yeo and Eisenberg 1977). The ionic
conductivity of Nafion and the possibility to absorb polar vapors are provided by
hydrophilic ionic groups (–SO3–H+) whereas hydrophobic “backbone” groups
(–CF2–CF2–) are relatively chemically inert (Morris and Sun 1993). The precise
mechanism of ionic transport in Nafion is still under investigation and possibly
involves the inverted micelle and the percolation effects (Tailoka et al. 2003).
The selectivity of response of Nafion to different vapors when Nafion is applied
onto the RFID sensor is provided by the differences in the resistance RF and capaci-
tance CF of the solid polymer electrolyte film on top of the RFID sensor antenna
structure upon exposure to different vapors (see Fig. 6.3a).
Scheme 6.1
10 (a)
−10
Delta Zp (Ohm)
−20
−30
H2O
−40
MeOH
−50 EtOH
ACN
−60
−70
0 0.5 1 1.5 2 2.5 3 3.5
Time (h)
1.5
(b)
MeOH
1 ACN 0.2
0.2EtOH 0.15
0.15 0.2
0.5 0.1
0.1
0.15 0.07 0.07 0.1
0.15
0 0.04 0.07
0.1 0.04
PC 2
0.04
0.02 0.02
−0.5 0.07 0.02
0.04
0.02
0
−1
H2O
−1.5 0.2
−2
−2 −1 0 1 2 3 4 5 6 7 8
PC 1
Fig. 6.4 RFID sensor response. (a) Measured Zp response for four analytes (H2O, EtOH, MeOH,
and ACN), six concentrations (0, 0.02, 0.04, 0.07, 0.10, 0.15, and 0.20 P/Po) and three replicates.
(b) Results of the PCA multivariate analysis of measured parameters F1, F2, Fp, and Zp of the
RFID sensor to the changes in H2O, EtOH, MeOH, and ACN at six concentrations each, and three
replicates per concentration. Numbers are P/Po values for individual vapors
6 Chemical Sensors: New Ideas for the Mature Field 117
the smallest measured vapor concentration (0.02 P/P0) to be 600 ppb. Upon a care-
ful selection of the sensing material for a specific application, such low detection
limits obtained with the RFID sensors were attractive for diverse applications in
environmental, homeland security, and other demanding applications.
Critical to the wide acceptance of the demonstrated RFID sensors is the
analyte-quantitation ability of these sensors without repositioning errors between
the RFID sensor and a pickup coil. We evaluated the capability for position-
independent (RFID sensor: pickup coil) analyte quantification using multivariate
analysis tools. For these evaluations, we assembled a test system that contained
an RFID sensor positioned in a low dead volume gas flow cell as shown in Fig.
6.5a. A pickup coil was positioned outside the flow cell on an X–Z translation
stage. Thus, the relative position of the RFID sensor and the pickup coil was
changed in a controlled fashion. The studied repositioning conditions are illus-
trated in Fig. 6.5b and included 5 mm step changes in the Z-direction
(0, 5, 10, 15, 20 mm, and back to 0 mm) and 5 mm step changes in the X-direction
(0, 5, 10, 15, 20, 15, 10, 5, 0 mm). To perform analyte-quantitation measure-
ments, the RFID sensor was coated with a composite poly(vinyl acetate)/carbon
black polymer film responsive to water vapor. For each position of the sensor,
two replicate exposures to water vapor were performed. Thus, during each step,
the sensor environment was switched twice from dry nitrogen to 45% relative
humidity (RH) nitrogen. The sensing film demonstrated a desired completely
reversible response (see Fig. 6.5c). Upon sensor repositioning, we were inter-
ested in three aspects of position-induced sensor response effects: (1) sensor
signal upon exposure to dry carrier gas (baseline signal), (2) sensor signal
change from the baseline signal upon exposure to 45% RH (analyte response),
and (3) sensor noise.
Results of the measured response of the RFID sensor to the changes in RH at
different distances from the pickup antenna are summarized in Fig. 6.6a–d. The
measured parameters included parameters from the real and imaginary portions
of the complex impedance, F1 shift, F2 shift, Fp shift, and Zp, as described
above. Unfortunately, as shown in Fig. 6.6a–d, none of the individual measured
parameters provided a position-independent analyte quantitation.
The position-induced affects can be summarized as the changes in the baseline
sensor signal and changes in analyte response magnitude upon exposure to 45%
RH. It was also observed that the sensor noise did not significantly change under
these testing conditions.
To address this problem in significant sensitivity of the sensor response to the posi-
tion of the pickup coil, we performed a principal components analysis of measured
parameters. The goal was to identify a possibility of rejecting the repositioning effects
in the multivariate domain. Results of the multivariate analysis of response of the RFID
sensor to the changes in RH at different distances from the pickup coil are summarized
in Fig. 6.6e. Thus, using multivariate analysis of RFID sensor response, we were able
to preserve position stability of the baseline sensor signal upon exposure to dry carrier
gas and the magnitude of the sensor signal change upon exposure to 45% RH.
118 R.A. Potyrailo
(a)
RFID sensor tag
Z
Network
analyzer
RF pickup antenna
(b)
25
Z Direction X Direction
20
Distance (mm)
15
10
0
1 2 3 4 5 6 7 8 9 10 11 12 13 14
Positions
(c)
RFID Sensor Signal
45%RH 45%RH
(Arb. Units)
5 min
Time
Fig. 6.5 Evaluation of the analyte-quantitation ability of RFID sensors in the presence of repositioning
errors between the RFID sensor and a pickup coil. (a) Assembled test system. (b) Different tested posi-
tions of the RFID sensor in the X- and Z-directions with respect to the pickup antenna. (c) Replicate
(n = 2) response of the RFID sensor to 45% RH and dry N2 for each X and Z position of the sensor
We also explored the applicability of the RFID sensors for chemical analysis in
liquids. Of course, the resonance of the sensor will be dumped in high-conductivity
solutions. However, measurements can be made in low-conductivity solutions, such
as deionized water. An isolating layer can be applied onto the sensor to make sensor
(a) (b)
150000 20000
(c) (d)
40000
Fp Shift (Hz) Zp
Frequency Shift (Hz)
150
20000
0
100
Zp
−20000
−40000 50
−60000
0 80 160 240 320 400 480 0 80 160 240 320 400 480
Time (min) Time (min)
(e)
5
3
Principal Component #2
−1
−2
−3
−4
−5
0 500 1000 1500 2000 2500 3000 3500 4000
Sample Number
Fig. 6.6 Multivariate analysis provides position-independent analyte quantification. (a) F1 fre-
quency shift of the maximum of the imaginary part of the complex impedance; (b) F2 frequency
shift of the minimum of the imaginary part of the complex impedance; (c) Fp frequency shift of
the maximum of the real part of the complex impedance; (d) Zp, the magnitude of the real part of
the complex impedance; (e) Results of principal components analysis of measured parameters
120 R.A. Potyrailo
(a)
100
0 Ion conc. (ppb)
99 4
8
98 12
Zp (Ohm)
100
97 20
98
Zp (Ohm)
96 96 28
95 94
92
94 0 10 20 30 40 50 44
NaCl conc. (ppb)
0.8
0.6
0.4
0.2
0
NaSO4 HCl NaCl KH2PO4
(c) Ionic Species
60000
50000
Freq. shift (Hz)
40000
30000
20000
10000
0
0 2 4 6 8 10
Acetone conc. %vol.
Fig. 6.7 Applicability of RFID sensors for monitoring in liquids. (a) Analysis of ion concentra-
tions in deionized water. Inset, linear calibration curve of detection of NaCl in deionized water
over ion concentration range from 0 to 40 ppb. (b) Response patterns of bare and polysulfone-
coated RFID sensors upon exposures to different ions in deionized water. (c) Quantitation of
residual organic solvents in water using a polysiloxane-coated RFID sensor
operation possible in solutions with high conductivity (Hofmann et al. 2005; Ong
et al. 2001). Analysis of concentrations of ions in water was performed with such a
sensor as shown in Fig. 6.7a. While the sensor was in contact with deionized water,
6 Chemical Sensors: New Ideas for the Mature Field 121
increasing concentrations of NaCl were added to the water. As expected, the sensor
response was affected by the increasing ion concentration. The inset of Fig. 6.6a
shows the linear calibration curve of detection of NaCl in deionized water over ion
concentration range from 0 to 40 ppb. The calculated detection limit (at S/N = 3) was
460 part per trillion (ppt). Because these measurements were performed with an
uncoated RFID sensor, we effectively measured water conductivity change related to
NaCl concentration. Of course, coating of the sensor with a chemically sensitive film
should provide a selective response to different ions in water. Fig. 6.7b illustrates the
change in the response pattern of the RFID sensor response upon coating of the sen-
sors with a polysulfone sensing film and analysis of different ionic species such as
NaSO4, HCl, NaCl, and KH2PO4 in water. Furthermore, another RFID sensor was
coated with a hydrophobic sensing film (polysiloxane) to demonstrate quantitation of
residual organic solvents in water. In particular, Fig. 6.7c illustrates a response of an
RFID sensor upon interactions with different concentrations of acetone in water.
In conclusion, we have shown that a careful design of the electrical energy trans-
ducer to measure several parameters at once provides a desired improvement in
sensor selectivity.
O O O O
N N N N Si-O Si O -Si
O O O O
O O O O y
x ~10
Scheme 6.2
122 R.A. Potyrailo
40 (a)
5.3
2.4
20
1.5
10
0.5
0
0 200 400 600 800 1000
Vapor concentration (ppb)
5.5
(b) y = 4.818−0.493x R2 = 0.999
5
4.5
log (K)
3.5
3
conventional polymers
2.5
−2 −1 0 1 2 3
Log (Vapor concentration, ppm)
(c) O C
Signal Intensity (counts)
Three-years old N Si
Fresh
Fig. 6.8 Performance of silicone polyimide as a vapor-sensing material upon deposition onto a
10 MHz TSM resonator: (a) concentration response curves for the determination of part-per-bil-
lion levels of TCE with 0.5–5.3 mm thick sensing films. (b). Partition coefficient as a function of
TCE concentration. (c) Long-term stability of new sensing material. Comparison of oxidation of
the freshly coated and three-year-old sensing films as determined by XPS
organic solvent vapors at trace levels (Potyrailo and Sivavec 2004). Typical calibration
curves for the determination of part-per-billion levels of TCE for several films of
different thickness deposited onto 10-MHz thickness-shear mode resonators are
6 Chemical Sensors: New Ideas for the Mature Field 123
presented in Fig. 6.8a. We observed that the sorption isotherm of new sensor poly-
mers to different chlorinated and other types of vapors was nonlinear.
The origin of this nonlinearity is likely to be from a combination of different vapor-
sorption mechanisms with contributions from both Langmuir- and non-Langmuir-
type noncovalent vapor–polymer interactions. When the frequency change was
related to the square root of the vapor concentration, linear responses were observed
with a slight upward non-linearity with the thickest tested film.
The magnitude of the response of conventional polymer materials at these concen-
trations was negligible, <1 Hz because partition coefficients of different conventional
polymer materials for TCE and other chlorinated volatile organic vapors of environ-
mental interest are typically (0.5–2) × 103 as reported in the literature (Finklea et al.
1998; Grate et al. 1997b, 1998; Jarrett and Finklea 1999; Patrash and Zellers 1993).
These partition coefficients are calculated assuming the sensor response is due to
mass-loading only (Grate et al. 1997b). Fig. 6.8b compares the partition coefficients
for TCE of new and conventional polymers. These data demonstrate a more than 100-
fold enhancement of the partition coefficient of our new material for detection at ppb
levels when compared to conventional currently used sensor materials such as phenyl-
methyl-polysiloxanes, poly(epichlorohydrin), poly(isobutylene), poly(ethylene
maleate), poly(ethylenimine), and many others (Bender et al. 2003; Grate et al.
1997a, 1998; Patrash and Zellers 1993, 1998; Rosler et al. 1998).
The concentration dependence of the partition coefficient as seen in Fig. 6.8b
originates from the nonlinear response of the materials to vapors. We believe that
this high partition coefficient may originate from the very favorable solubility of
organic vapors in the combination of the hard and soft blocks of this sensor mate-
rial. It is likely that in our silicone block polyimide, the generated microphase
domains with the approximate size of the polymer soft and hard segments create
sites for very favorable solubility interactions with the vapors.
Long-term stability of sensor materials is the critical parameter for practical
sensor operation. Oxidation of a sensor polymer is one of the main degradation
mechanisms that should be avoided. Oxidation may lead to the decrease of material
response to nonpolar organic vapors and to the increase of the response to polar
vapors, with water vapor being the main concern. We studied oxidation of freshly
coated and three-year-old sensor films using X-ray photoelectron spectroscopy
(XPS). No detectable differences in the amount and types of oxygen on the film
surface were found as shown in Fig. 6.8c. We also have found that the film thick-
ness was stable to within 5% as measured by the value of fundamental frequency
of coated crystals over time. This long-term stability of our new sensor materials is
provided by the nature of silicone polyimides. The flexible polysiloxane segments
enhance the stability of imide segments, and improve adhesion and membrane
properties of the film (Chang et al. 2001; McGrath et al. 1999). We also observed
that unlike conventional polyimides, water sensitivity of the new hybrid polyimides
was suppressed because of the silicone soft block.
The high sensitivity and long-term stability of these sensor materials make them
attractive for ultrasensitive sensors. These materials were implemented as thin films to
detect part-per-billion concentrations of TCE with 3 ppb limit of detection. Detection
limits for other chlorinated organic solvent vapors such as perchloroethylene (PCE),
124 R.A. Potyrailo
4 1
2
Scores on PC# 2
0
3
-2
4 5
-4
2
-6
-8
-10
0 10 20 30 40 50 60 70 80 90 100
Scores on PC# 1
Fig. 6.9 Plot of the first two principal components for the dataset obtained from the response of
the four-sensor array to five chlorinated organic vapors at different concentrations. Vapors: 1, cis-1,
2-DCE; 2, TCE, 3, trans-1, 2-DCE, 4, VC; 5, 1, 1-DCE. Each data point is the mean of three
measurements; squares represent one standard deviation
6 Chemical Sensors: New Ideas for the Mature Field 125
Fig. 6.10 Field-testing of a sensor system with several kinds of silicone polyimides developed
using high-throughput screening techniques
sensor’s response to carbon dioxide was linearly proportional to the polymer film
thickness. Polycarbonate polymer is a glassy polymer and has been employed
earlier for the reversible CO2 detection at room temperature (Ahuja et al. 1999).
The mechanism of the polymer response to CO2 is based on the diffusion of the gas
into the “microvoids” of the polymer film resulting in an increase of the film mass.
This polymer–CO2 interaction causes no detectable change in the mechanical prop-
erties of the film and minimal relaxation of the polymer matrix (Ahuja et al. 1999).
Upon CO2 sorption into polycarbonate, the mass increase of the film was detected
with the transducer. Other materials for CO2 determinations at room temperature
include tetramethylammonium fluoride tetrahydrate (Gomes et al. 1996), aminopro-
pylsiloxane–octadecylsiloxane copolymer (Oprea et al. 1999), polyethyleneimine
(Korsah et al. 1998), fluorinated polyimide (Hoyt et al. 1998), and tetrakis(hydroxy-
ethyl)ethylenedlamine (Fatibello-Filho et al. 1989).
Reproducibility and hysteresis-free sensor response are among the key figures
of merit of a chemical sensor. These metrics depend on the performance of both the
transducer and sensing film. Reproducibility of the sensor response was evaluated
by repetitively exposing the sensor to different CO2 concentrations in both increas-
ing and decreasing concentration sequences. We intentionally selected high CO2
concentrations (10–100% vol. CO2) to provide the worst-case scenario for the
sensor operation to study hysteresis effects. Quantification of sensor hysteresis was
performed over the wide concentration range as shown in Fig. 6.11a. We have
found that the response of the sensor to increasing and decreasing concentrations
of CO2 had no detectable hysteresis effect. The sensitivity of the CO2 sensor operat-
ing at an excitation voltage of 7.5 V was 57.50 ± 0.38 mV/1% vol. CO2. At these
testing conditions, the sensor demonstrated an excellent reproducibility. Repetitive
exposures to 100% vol. CO2 provided signal reproducibility of 0.17% RSD. From
126 R.A. Potyrailo
6400
5600
(a)
0
0 20 40 60 80 100
CO2 Concentration (%)
4000
(b)
3000
Sensor Response (mV)
2000
1000
0
−1000
−2000
−3000
0 2000 4000 6000 8000
Time (s)
250
(c)
Sensor Response (mV)
200
150
50
100
25
50
0
0 4000 8000
0
0 10000 20000 30000 40000
CO2 Concentration (ppm)
Fig. 6.11 Performance of polycarbonate sensing film upon deposition onto a micromachined
cantilever with an integrated piezoresistive readout. (a) Quantification of sensor hysteresis. (b)
Reproducibility of sensor response evaluated by repetitively exposing sensor to different (0%,
10%, 30%, 50%, 75%, and 100% vol.) CO2 concentrations. (c) Calibration results for the CO2
sensor for part-per-million CO2 concentrations
the analysis of the dynamic data collected during the reproducibility studies (see
Fig. 6.11b), we evaluated the response time of the sensor to a chemical stimulus.
We measured the time T90 required to achieve a 90% signal change upon a step
change of CO2 gas concentration to be about 45 s.
6 Chemical Sensors: New Ideas for the Mature Field 127
Optical sensors employ a sensing reagent that undergoes a change in its optical
property (e.g., elastic or inelastic scattering, absorption, luminescence intensity,
luminescence lifetime, or polarization state) upon interaction with the analyte species
(Potyrailo et al. 1998a, 2005d; Potyrailo and Hassib 2005; Potyrailo and Lemmon
2005). Organic reagents are widely employed in existing optical sensors (Bakker
et al. 1997; Wolfbeis 2004), however, such reagents often suffer from their intrinsic
limitations such as rapid photobleaching and short shelf-life. Application of engi-
neered nanostructures provides exciting opportunities for sensor materials with
previously unavailable capabilities (Convertino et al. 2003; Holtz and Asher 1997;
Jiang et al. 2005).
Nanomaterials promise to introduce a variety of new important properties into
sensor response such as improved selectivity, sensitivity, dynamic response, and
long-term stability (Potyrailo 2006). We recently introduced a new concept for
selective chemical sensing based on different size CdSe semiconductor nanocrys-
tals incorporated into a polymer matrix to overcome a photobleaching limitation of
traditional organic reagent-based sensor materials. We have found that when differ-
ent size CdSe nanocrystals made and passivated with tri-n-octylphosphine oxide
(TOPO) using known methods (Kovalevskij et al. 2004; Murray et al. 1993) were
128 R.A. Potyrailo
incorporated into a polymer film and photoactivated, each size of CdSe nanocrys-
tals unexpectedly demonstrated its own photoluminescence (PL) response pattern
upon exposure to polar and nonpolar vapors in air. When this composite response
was processed using multivariate analysis, a single film with different size CdSe
nanocrystals served as a selective sensor.
Our sensing films were produced by spin-casting of a solution of 2.8 and 5.6 nm
diameter nanocrystals and polymethylmethacrylate (PMMA) in toluene and were
photoactivated with a portable 407 nm diode laser. The fabricated sensor films were
exposed to methanol and toluene vapors at the vapor pressure of 46 and 11 torr,
respectively. This corresponded to concentrations of 61,000 and 14,000 ppm of
methanol and toluene, respectively. When a single wavelength was selected at the
peaks of PL spectra of each size of the nanocrystals, two kinetic responses were
obtained. Fig. 6.12a shows baseline-corrected response patterns of chemically
dependent PL of the two-size CdSe nanocrystals sensor film. Emission of the 2.8
nm nanocrystals was measured at 511 nm and emission of the 5.6 nm nanocrystals
was measured at 617 nm. We believe that the difference in the response patterns of
the nanocrystals is related to the combined effects of the dielectric medium sur-
rounding the nanocrystals, their size, and surface oxidation state. The sensitivity of
these sensors was defined as a signal change upon exposure to the known concen-
trations of vapors. The response and recovery kinetics of PL from the 2.8 nm
nanocrystals in PMMA upon exposure to methanol were very fast (<0.5 min).
However, 5.6 nm nanocrystals in the same sensor film exhibited much longer
response and recovery times upon interactions with methanol, 4 and 20 min,
respectively. The 5.6 nm nanocrystals had 4 min response and 0.5 min recovery
times upon interactions with toluene.
Computational studies of the effects of dielectric media on CdSe nanocrystal elec-
tronic properties show that the absolute value of the dipole moment of the nanocrystal
increases with elevated dielectric constant of the surrounding environment (Rabani
et al. 1999). The magnitude of the change in dipole moment increases with particle
size. The elevated diameter and surface area of larger particles means that even small
changes in the surface charge of the nanocrystals will result in significant changes in
the particle’s dipole moment. Both sizes of nanocrystals in our sensor films exhibited
a decrease in emission upon exposure to methanol (e = 33), which causes a large
increase in the local dielectric constant relative to the PMMA medium (e = 2.7–3.0).
Toluene (e = 2.2) has a slightly lower dielectric constant than PMMA and resulted in
an increase in the emission of the larger nanocrystals alone. The different size nano-
crystals also can have different coverage with the capping TOPO ligand (Leatherdale
and Bawendi 2001) and when immobilized into a polymer matrix (Kovalevskij et al.
2004), can contribute to the variable response pattern.
To obtain selective chemical response from individual response patterns of nanoc-
rystals of different size, we applied a principal components analysis (PCA) technique
(Beebe et al. 1998) to analyze the dynamic response data of the sensor film during its
repetitive exposures to methanol and toluene vapors. For PCA analysis, 482 PL spectra
(450–700 nm range) collected during the repetitive (n = 3) exposures of the sensor film
to methanol and toluene vapors were used without baseline drift correction. The capa-
6 Chemical Sensors: New Ideas for the Mature Field 129
(a) 1500
5.6 nm
1000
PL (counts)
500 2 2 2
1 1 1
2.8 nm
0
0 50 100 150 200
Time (min)
(b) 5
0
PC 3 ( 5.09%)
−5 1
2
−10
−15
−20
−15 −10 −5 0 5 10 15 20
PC 1 (53.96%)
(c) 1400 22
1200
230
1000 11
PL (counts)
220 11
PL (counts)
22
800 210
200
600
190
400 180
0 0.5 1 1.5 2 2.5 3 3.5
Time (h)
200
0
0 2 4 6 8 10 12 14 16
Time (h)
Fig. 6.12 Photoluminescence response of two-size (2.8 and 5.6 nm) CdSe nanocrystals in
poly(methyl methacrylate) polymer film upon repetitive exposures to methanol (1) and toluene (2)
vapors: (a) dynamic vapor response with emission of 2.8 nm nanocrystals at 511 nm and emission
of 5.6 nm nanocrystals at 617 nm. (b) Principal components scores plot demonstrating selective
film response to methanol and toluene. Unlabeled data points result from times when the film was
exposed to a blank (dry air). (c) Long-term stability of film response upon repetitive exposures to
methanol and toluene under a 16 h continuous 407 nm laser excitation
130 R.A. Potyrailo
bility for vapor discrimination using a single sensor film was evaluated from a PCA
model generated with autoscaled data that was built based on the full spectra dynamic
data. The relationship between the collected data was described by plotting scores of
relevant principal components of the PCA model versus each other as shown in Fig.
6.12b. Data analysis with the whole uncorrected spectra provided excellent vapor
discrimination capability due to the elimination of the PL drift-related artifacts. The
remaining response scatter in regions 1 and 2 was due to nonequilibrated sensor
responses during the kinetic experiments. The largest variance (captured by the first
principal component of the PCA model) exhibited the strongest response to both
vapors. The second PC captured the drift contributions in the response.
We further tested the stability of PL emission of new sensor films upon an
extended irradiation with the laser light. Such response stability is critical in continu-
ous monitoring applications. The parameters of interest in these evaluations were:
(1) the overall stability of the PL intensity and (2) the stability of the response pattern
to methanol and toluene vapors. Fig. 6.12c shows response patterns from 2.8 and 5.6
nm nanocrystals in PMMA film over 16 h of continuous exposure of the film to laser
radiation. During the laser exposure, the sensor film was periodically exposed to
methanol and toluene vapors. These data indicated that the response of new selective
sensor films was very stable upon extended exposure to laser excitation without
significant degradation of both emission intensity and response pattern.
Thus, we demonstrated selective detection of polar and nonpolar vapors by gen-
erating highly stable multivariate response patterns from different-size CdSe nanoc-
rystals incorporated in PMMA. Our studies are in progress to better understand the
effects of polymer matrices. Such work promises to complement existing solvato-
chromic organic dye sensors with more photostable and reliable sensor materials.
In our other application of radiant transduction, we have developed an approach
for using unmodified optical disk drives for chemical sensing of water and air
contamination (see Fig. 6.2f). In this sensor system, the analog signals were
acquired from conventional optical disk drives and these signals were used for
quantitative detection of optical changes of sensor films deposited on conventional
CD and DVD optical disks (Potyrailo et al. 2005a,b,c; Potyrailo et al. 2006).
Although the drives still perform their original function of reading and writing
digital content to optical media, they also provide analog signals for quantitative
sensor applications when sensor films are deposited onto optical disks. Because no
alteration of the manufacturing process of optical disks is required, any disk can be
employed for deposition and readout of sensor films. Such a sensor platform is
quite universal and can be applied for quantitative chemical and biological detec-
tion, as well as for the monitoring of changes of physical properties of regions
deposited onto a CD or DVD (e.g., during combinatorial screening of materials).
Fig. 6.13 depicts our concept for chemical and biological detection that employs
an analog signal from a conventional CD/DVD drive to quantify optical changes in
sensor films deposited on the read surface of CD or DVD disks. A conventional
optical disk drive for reading DVDs and CDs contains all needed components to
perform quantitative chemical and biological analysis. The drive has two lasers, 650
and 780 nm to read DVDs and CDs, respectively, a Si photodiode detector, and a
(a) Sensing films
Optical disk Reflective layer
Processor Photodetector
EIDE
interface Standard
optical disk drive
Control of
optical disk drive
A/D
LabVIEW
Data acquisition
and drive control Analog
Analog signal acquisition signal
and drive control
Reflective
layer
Sensor film
ity
ss
g
in
rin
an
iv
ne
ive
ct
te
rb
ick
le
ct
at
so
ef
Th
fra
Sc
Ab
Re
Fig. 6.13 Concept for quantitative chemical detection using a conventional optical disk drive and
a DVD or CD disk. (a) Schematic of a conventional optical disk drive and the methodology for
obtaining an analog signal from a photodiode detector and for controlling the optical disk drive. (b)
Double-pass interaction of a laser beam with the sensor film deposited onto the read side of an
optical disk. (c) Optical phenomena and parameters of sensor films involved in signal generation
132 R.A. Potyrailo
sophisticated laser tracking system to scan across the disk surface. In the developed
Lab-on-DVD sensor system, an analog signal from the photodiode is extracted
before it is digitized during the reading of the digital content from an optical disk and
brought into a data acquisition program as shown in Fig. 6.13a. This signal is used
for quantitative detection of changes of optical properties of chemical or biological
sensor films deposited on the read surface of the optical disk. In the Lab-on-DVD
sensor system, the optical disk drive is further controlled through the Enhanced
Integrated Disk Electronics (EIDE) interface. The controlled parameters of the opti-
cal disk drive include positioning of the laser pickup head at any specified radial
position, scanning the laser pickup head over a range of desired radii with a control-
led spatial resolution, and the linear rotation velocity of the optical disk.
In the absence of a sensor film, laser light is transmitted through the surface of
the optical disk, reflected from the disk’s reflective data layer and returned to the
photodiode detector. When a sensor film is applied onto the read side of the CD or
DVD disk, the laser light travels through the sensor film twice as shown in Fig.
6.13b. Upon interactions of the sensor film with chemical or biological species,
optical properties of the sensor film vary causing the change in the amount of light
detected by the photodiode detector of the laser pickup head and allowing for quan-
titation of sensor film response.
The optical system of a conventional CD/DVD drive focuses the laser light onto
a reflective layer inside the disk to a spot of about 1 mm and provides polarization
and phase control of the light that reaches the detector. In reading digital data from
a disk, these features are important for the rejection of ambient light and light
produced by scratches and other imperfections on the disk surface. In the current
application of optical disk drives, these features provide an opportunity for the
chemical and biological quantification based on a variety of optical phenomena that
can be produced in sensor films (see Fig. 6.13c).
Numerous possibilities exist for the use of conventional sensor materials as well
as nanomaterials for the realization of these optical phenomena. Color changes can
be produced by a wide variety of organic dyes developed in the twentieth century for
wet chemistry test methods and adapted for solid-film sensing (Capitán-Vallvey
et al. 2000; Chau and Porter 1990; Dybko et al. 1998). Plasmon resonance bands of
metal nanoparticles and other plasmonic nanostructures can be tailored to 650 and
780 nm laser wavelengths and band shifts can be produced by the aggregation or
deaggregation of these plasmonic nanostructures in sensing films (Elghanian et al.
1997; Nath and Chilkoti 2002). Photonic crystals can be easily assembled with their
diffraction peaks in the vicinity of the laser wavelengths with biochemical reactions
inducing peak shifts toward or from the laser wavelengths (Holtz and Asher 1997).
Variations in other optical phenomena such as scattering produced by phase and
morphology change in polymers (Potyrailo et al. 2002), metal film reflectivity
produced by film degradation (Bouten et al. 2002) or silver staining (Taton et al.
2000), and thickness and refractive index produced by polymer swelling (McCurley
and Seitz 1991; Miyata et al. 1999; Zhang et al. 2003) have much smaller wave-
length dependence and can be applied with any laser in the optical disk drive. Newly
introduced Blu-Ray® optical disk drives with their 405 nm lasers are also attractive
6 Chemical Sensors: New Ideas for the Mature Field 133
in expanding the range of optical sensing materials (e.g., freebase and metallo
porphyrins) for chemical and biological sensing using new-generation optical disks.
It is worth noting that we do not measure fluorescence or other types of emission
from the sensor films using the unmodified optical disk drives due to the optical
properties of the disk drive. However, certain types of new nanoparticle-based
colorimetric and other principles that can provide sensitivity comparable to or
greater than fluorescence can be employed (Steinberg et al. 1996; Taton et al. 2000;
Wetzl et al. 2004).
Performance of the Lab-on-DVD sensor system has been demonstrated for
quantitative chemical analysis in water and air. For example, for determination of
Ca2+, sensing films incorporated a Ca2+ sensitive dye Xylidyl Blue in a poly-
(2-hydroxyethy) methacrylate hydrogel matrix. For fabrication of films, the dye/
polymer/solvent (1-methoxy-2-propanol) solution was applied to the DVDs to
form 3 × 4 mm sensing films. After solvent evaporation, the sensing films
adhered to the DVD surface and were ready for testing for their response to Ca2+.
Water samples (20–40 mL volumes) with varied concentrations of Ca2+ were
applied to the sensing films and removed with pressurized house nitrogen gas
after 2 min of exposure (Potyrailo et al. 2006). After exposure, quantitative meas-
urements were performed in the optical disk drive. A typical calibration curve for
Ca2+ determinations is presented in Fig. 6.14a. The shape of the calibration curve
was typical of determinations of Ca2+ and other cations using organic chromog-
enic dyes immobilized in polymeric films (Seitz 1988; Wolfbeis 1991). The cal-
culated detection limit for Ca2+ determinations (at S/N = 3) was 5 ppm for the
Lab-on-DVD measurement system. This detection limit corresponded to a 0.023
absorbance unit resolution as obtained by measurement of the same colorimetric
films with the reference optical spectrometer. This detection limit in Ca2+ deter-
minations is applicable for both clinical and environmental applications.
In another application of Lab-on-DVD, for the determination of chlorine in
water, we have developed a indicator formulation that contained 1¢,1¢-diethyl-
4,4¢-carbocyanine iodide dye in a poly(2-hydroxyethymethacrylate) hydrogel
matrix. The indicator concentration in the film was optimized for detection of
chlorine at ppb concentrations. A calibration curve for chlorine determinations
with the Lab-on-DVD system is presented in Fig. 6.14b. Indicator film responses
were recorded after 2 min exposures to chlorine-containing water solutions (~30
mL volumes). Each data point was a result of measurements performed at three
radial positions separated by 0.5 mm. Typical noise in the determinations was
<10% RSD. The calculated detection limit for chlorine determinations (at S/N =
3) was 200 ppb.
We have also developed an approach for measurements of chemical species in
air using Lab-on-DVD. In this application, a vapor-introduction port has been made
into an optical disk drive and different concentrations of tested vapors were intro-
duced into the optical disk drive. A demonstration of this sensing approach has
been performed with quantitation of water vapor in air. A sensing film comprised
of a Nafion polymer formulated with rhodamine 800 dye has been applied to a
DVD. A signal change from the detector has been recorded in real-time upon
134 R.A. Potyrailo
(a)
100
60
40
20
0
1 10 100 1000
Calcium Conc. (ppm)
(b)
Normalized DVD Signal
100
75
50
25
76 (c)
Normalized DVD Signal
72
68
0 10 20 30 40 50 60 70
64 % RH
60
56 1 2 1 2
52
0 200 400 600 800 1000 1200
Time (s)
Fig. 6.14 Quantitation of chemical species in water and air using Lab-on-DVD sensor system. (a)
Ca2+ determinations in water. (b) Chlorine determinations in water. (c) Relative humidity determina-
tions in air. Each data point in (a) and (b) is the mean of three replicate measurements across different
radial positions of an indicator film; error bars are one standard deviation. Numbers in (c) are relative
humidity values: (1) 22% RH and (2) 67% RH. Baseline is response of the sensing film in dry air
exposure of the DVD with the sensing film to different relative humidity while the
DVD was reading in the optical disk drive. A reversible signal change has been
observed of the absorption of the sensing film as shown in Fig. 6.14c with the signal
change proportional to the relative humidity of air.
6 Chemical Sensors: New Ideas for the Mature Field 135
Summary
Acknowledgments This research has been inspired by the creative teammates at GE Global
Research, Nomadics, and Indiana University, Bloomington, IN who have coauthored original
contributions cited here: S. Boyette, M. D. Butts, J. R. Cournoyer, Z. Ding, K. Dovidenko,
W. P. Flanagan, S. K. Gamage, S. E. Genovese, L. Hassib, A. M. Leach, J. P. Lemmon, R. J. May,
W. G. Morris, E. Olson, J. J. Salvo, O. P. Siclovan, R. E. Shaffer, T. M. Sivavec, A. Vertiatchikh,
M. B. Wisnudel, and R. J. Wroczynski (GE), L. Salsman (Nomadics), R. C. Conrad, T. L.
Danielson, M. Johnson, and A. W. Szumlas (Indiana University), and H. Ghiradella (SUNY
Albany). Special thanks go to G. M. Hieftje (Indiana University), A. D. Ellington (while at Indiana
University),T. K. Leib, and A. Linsebigler (GE) for letting the creativity grow and expand.
References
Adhikari B, Majumdar S (2004) Polymers in sensor applications. Prog. Polym. Sci. 29:699–766.
Ahuja A, James DL, Narayan R (1999) Dynamic behavior of ultra-thin polymer films deposited
on surface acoustical wave devices. Sens. Actuators B 72:234–241.
Archibald B, Brümmer O, Devenney M, Giaquinta DM, Jandeleit B, Weinberg WH, Weskamp T
(2002a) Combinatorial aspects of materials science. In: Nicolaou KC, Hanko R, Hartwig W,
editors. Handbook of Combinatorial Chemistry. Drugs, Catalysts, Materials. Weinheim: Wiley,
chapter 34, pp. 1017–1062.
Archibald B, Brümmer O, Devenney M, Gorer S, Jandeleit B, Uno T, Weinberg WH, Weskamp T
(2002b) Combinatorial methods in catalysis. In: Nicolaou KC, Hanko R, Hartwig W, editors.
Handbook of Combinatorial Chemistry. Drugs, Catalysts, Materials. Weinheim: Wiley, chapter
32, pp. 885–990.
6 Chemical Sensors: New Ideas for the Mature Field 137
Artmann R (1999) Electronic identification systems: State of the art and their further development.
Comput. Electron. Agri. 24:5–26.
Bakker E, Bühlmann P, Pretsch E (1997) Carrier-based ion-selective electrodes and bulk optodes.
1. General characteristics. Chem. Rev. 97:3083–3132.
Ballantine DS, Jr., White RM, Martin SJ, Ricco AJ, Frye GC, Zellers ET, Wohltjen H (1997)
Acoustic Wave Sensors: Theory, Design, and Physico-chemical Applications. San Diego, CA:
Academic Press. 436 p.
Beebe KR, Pell RJ, Seasholtz MB (1998) Chemometrics: A Practical Guide. New York: Wiley.
Bender F, Barié N, Romoudis G, Voigt A, Rapp M (2003) Development of a preconcentration
unit for a SAW sensor micro array and its use for indoor air quality monitoring. Sens.
Actuators B 76:1–7.
Bennett MD, Leo DJ (2004) Ionic liquids as stable solvents for ionic polymer transducers. Sens.
Actuators A 115:79–90.
Benschop J, Rosmalen GV (1991) Confocal compact scanning optical microscope based on
compact disc technology. Appl. Opt. 30:1179–1184.
Booksh KS, Kowalski BR (1994) Theory of analytical chemistry. Anal. Chem. 66:782A–791A.
Boussaad S, Tao NJ (2003) Polymer wire chemical sensor using a microfabricated tuning fork.
Nano Lett. 3:1173–1176.
Capitán-Vallvey LF, de Cienfuegos-Gálvez PA, Fernández Ramos MD, Avidad-Castañeda R (2000)
Determination of calcium by a single-use optical sensor. Sens. Actuators B 71:140–146.
Carrano JC, Jeys T, Cousins D, Eversole J, Gillespie J, Healy D, Licata N, Loerop W, O’Keefe M,
Samuels A, Schultz J, Walter M, Wong N, Billotte B, Munley M, Reich E, Roos J (2004)
Chemical and Biological Sensor Standards Study (CBS3). In: Carrano JC, Zukauskas A, edi-
tors. Optically Based Biological and Chemical Sensing for Defence. Bellingham, WA: SPIE
– The International Society for Optical Engineering. pp. xi–xiii.
Chang TC, Wang GP, Hong YS, Chen HB (2001) Characterization and degradation of hydrogen-
bonded acidic polyamideimides linked by disiloxanes. Polym. Degrad. Stab. 73:301–308.
Chau LK, Porter MD (1990) Optical sensor for calcium: Performance, structure, and reactivity of
calcichrome immobilized at an anionic polymer film. Anal. Chem. 62:1964–1971.
Cho EJ, Bright FV (2001) Optical sensor array and integrated light source. Anal. Chem.
73:3289–3293.
Cho EJ, Tao Z, Tehan EC, Bright FV (2002) Multianalyte pin-printed biosensor arrays based on
protein-doped xerogels. Anal. Chem. 74:6177–6184.
Christensen CM (1997) The Innovator’s Dilemma: When New Technologies Cause Great Firms
to Fail. Boston, MA: Harvard Business School Press.
Christensen CM, Raynor ME (2003) The Innovator’s Solution: Creating and Sustaining Successful
Growth. Boston, MA: Harvard Business School Press.
Chu C-L, Lin C-H (2005) Development of an optical accelerometer with a DVD pick-up head.
Meas. Sci. Technol. 16:2498–2502.
Convertino A, Capobianchi A, Valentini A, Cirillo ENM (2003) A new approach to organic sol-
vent detection: High-reflectivity Bragg reflectors based on a gold nanoparticle/Teflon-like
composite material. Adv. Mater. 15:1103–1105.
DeLongchamp DM, Hammond PT (2003) Fast ion conduction in layer-by-layer polymer films.
Chem. Mater. 15:1165–1173.
Dutta P, Tipple CA, Lavrik NV, Datskos PG, Hofstetter H, Hofstetter O, Sepaniak MJ (2003)
Enantioselective sensors based on antibody-mediated nanomechanics. Anal. Chem.
75:2342–2348.
Dybko A, Wróblewski W, Rozniecka E, Pozniakb K, Maciejewski J, Romaniuk R, Brzózka Z
(1998) Assessment of water quality based on multiparameter fiber optic probe. Sens. Actuators
B 51:208–213.
Elghanian R, Storhoff JJ, Mucic RC, Letsinger RL, Mirkin CA (1997) Selective colorimetric
detection of polynucleotides based on the distance-dependent optical properties of gold nano-
particles. Science 277:1078–1081.
Erickson D, Li D (2004) Integrated microfluidic devices. Anal. Chim. Acta 507:11–26.
138 R.A. Potyrailo
Fatibello-Filho O, de Andrade JF, Suleiman AA, Guilbault GG (1989) Piezoelectric crystal moni-
tor for carbon dioxide in fermentation processes. Anal. Chem. 61:746–748.
Feng C-D, Sun S-L, Wang H, Segre CU, Stetter JR (1997) Humidity sensing properties of nation
and sol-gel derived SiO2/Nafion composite thin films. Sens. Actuators B 40:217–222.
Filippini D, Alimelli A, Di Natale C, Paolesse R, D’Amico A, Lundström I (2006) Chemical sens-
ing with familiar devices. Angew. Chem. Int. Ed. 45:3800–3803.
Finkenzeller K (2003) RFID Handbook. Fundamentals and Applications in Contactless Smart
Cards and Identification. Hoboken, NJ: Wiley.
Finklea HO, Phillippi MA, Lompert E, Grate JW (1998) Highly sorbent films derived from
Ni(SCN)2(4-picoline)4 for the detection of chlorinated and aromatic hydrocarbons with quartz
crystal microbalance sensors. Anal. Chem. 70:1268–1276.
Franke ME, Koplin TJ, Simon U (2006) Metal and metal oxide nanoparticles in chemiresistors:
Does the nanoscale matter? Small 2:36–50.
Frantzen A, Scheidtmann J, Frenzer G, Maier WF, Jockel J, Brinz T, Sanders D, Simon U (2004)
High-throughput method for the impedance spectroscopic characterization of resistive gas
sensors. Angew. Chem. Int. Ed. 43:752–754.
Furuki M, Pu LS (1992) Hybrid gas detector of squarylium dye Langmuir-Blodgett film deposited
on a quartz oscillator. Thin Solid Films 210/211:471–473.
Gomes MTSR, Rocha TA, Duarte AC, Oliveira JABP (1996) The performance of a tetramethyl-
ammonium fluoride tetrahydrate coated piezoelectric crystal for carbon dioxide detection.
Anal. Chim. Acta 335:235–238.
Göpel W (1996) Ultimate limits in the miniaturization of chemical sensors. Sens. Actuators A
56:83–102.
Grate JW (2000) Acoustic wave microsensor arrays for vapor sensing. Chem. Rev. 100:2627–2648.
Grate JW, Rose-Pehrsson SL, Venezky DL, Klusty M, Wohltjen H (1993) Smart sensor system for
trace organophosphorus and organosulfur vapor detection employing a temperature-controlled
array of surface acoustic wave sensors, automated sample preconcentration, and pattern recog-
nition. Anal. Chem. 65:1868–1881.
Groves WAGrate JW, Abraham H, McGill RA (1997a) Sorbent polymer materials for chemical
sensors and arrays. In: Kress-Rogers E, editor. Handbook of Biosensors and Electronic Noses.
Medicine, Food, and the Environment. Boca Raton, FL: CRC. pp. 593–612.
Grate JW, Kaganove SN, Bhethanabotla VR (1997b) Examination of mass and modulus contribu-
tions to thickness shear mode and surface acoustic wave vapour sensor responses using parti-
tion coefficients. Faraday Discuss. 107:259–283.
Grate JW, Kaganove SN, Bhethanabotla VR (1998) Comparisons of polymer/gas partition coef-
ficients calculated from responses of thickness shear mode and surface acoustic wave vapor
sensors. Anal. Chem. 70:199–203.
Groves WA, Zellers ET, Frye GC (1998) Analyzing organic vapors in exhaled breath using a
surface acoustic wave sensor array with preconcentration: Selection and characterization of the
preconcentrator adsorbent. Anal. Chim. Acta 371:131–143.
Hagleitner C, Hierlemann A, Lange D, Kummer A, Kerness N, Brand O, Baltes H (2001) Smart
single-chip gas sensor microsystem. Nature 414:293–296.
Hansen CL, Skordalakes E, Berger JM, Quake SR (2002) A robust and scalable microfluidic
metering method that allows protein crystal growth by free interface diffusion. Proc. Natl.
Acad. Sci. USA 99:16531–16536.
Hansen KM, Ji H-F, Wu G, Datar R, Cote R, Majumdar A, Thundat T (2001) Cantilever-based
optical deflection assay for discrimination of DNA single-nucleotide mismatches. Anal.
Chem. 73:1567–1571.
Harsanyi G (1995) Polymer Films in Sensor Applications. Lancaster, PA: Technomic.
Hierlemann A, Baltes H (2003) CMOS-based chemical microsensors. Analyst 128:15–28.
Hirayama E, Sugiyama T, Hisamoto H, Suzuki K (2000) Visual and colorimetric lithium ion sens-
ing based on digital color analysis. Anal. Chem. 72:465–474.
Hirschfeld T (1985) Instrumentation in the next decade. Science 230:286–291.
6 Chemical Sensors: New Ideas for the Mature Field 139
Hirschfeld T, Callis JB, Kowalski BR (1984) Chemical sensing in process analysis. Science
226:312–318.
Hofmann MC, Kensy F, Buechs J, Mokwa W, Schnakenberg U (2005) Transponder-based sensor for
monitoring electrical properties of biological cell solutions. J. Biosci. Bioeng. 100:172–177.
Holtz JH, Asher SA (1997) Polymerized colloidal crystal hydrogel films as intelligent chemical
sensing materials. Nature 389:829–832.
Hoyt AE, Ricco AJ, Bartholomew JW, Osbourn GC (1998) SAW sensors for the room-tempera-
ture measurement of CO2 and relative humidity. Anal. Chem. 70:2137–2145.
Hruschka WR, Massie DR, Anderson JD (1983) Computerized analysis of two-dimensional elec-
trophoretograms. Anal. Chem. 55:2345–2348.
Hsieh M-D, Zellers ET (2004) Limits of recognition for simple vapor mixtures determined with a
microsensor array. Anal. Chem. 76:1885–1895.
Hughes RC, Yelton WG, Pfeifer KB, Patel SV (2001) Characteristics and mechanisms in ion-
conducting polymer films as chemical sensors. Polyethyleneoxide J. Electrochem. Soc.
148(4):H37–H44.
Ivanisevic A, Yeh J-Y, Mawst L, Kuech TF, Ellis AB (2001) Light-emitting diodes as chemical
sensors. Nature 409:476.
Janata J (1989) Principles of Chemical Sensors. New York: Plenum.
Janata J, Josowicz M (2002) Conducting polymers in electronic chemical sensors. Nat. Mater.
2:19–24.
Jandeleit B, Schaefer DJ, Powers TS, Turner HW, Weinberg WH (1999) Combinatorial materials
science and catalysis. Angew. Chem. Int. Ed. 38:2494–2532.
Jarrett MR, Finklea HO (1999) Detection of nonpolar vapors on quartz crystal microbalances with
Ni(SCN)2(4-picoline)4 coatings. Anal. Chem. 71:353–357.
Jiang P, Smith DWJ, Ballato JM, Foujger SH (2005) Multicolor pattern generation in photonic
bandgap composites. Adv. Mater. 17:179–184.
Jones CL (2005) Cryptographic hash functions and CD-based optical biosensors. Prob. Nonlinear
Anal. Eng. Syst. 2(23):17–36.
Jurs PC, Bakken GA, McClelland HE (2000) Computational methods for the analysis of chemical
sensor array data from volatile analytes. Chem. Rev. 100:2649–2678.
Kindlund A, Sundgren H, Lundstrom I (1984) Quartz crystal gas monitor with gas concentrating
stage. Sens. Actuators 6:1–17.
King WH, Jr. (1964) Piezoelectric sorption detector. Anal. Chem. 36:1735–1739.
Koinuma H, Takeuchi I (2004) Combinatorial solid state chemistry of inorganic materials. Nat.
Mater. 3:429–438.
Korsah K, Ma CL, Dress B (1998) Harmonic frequency analysis of SAW resonator chemical sen-
sors: Application to the detection of carbon dioxide and humidity. Sens. Actuators B
50:110–116.
Kovalevskij V, Gulbinas V, Piskarskas A, Hines MA, Scholes GD (2004) Surface passivation in
CdSe nanocrystal-polymer films revealed by ultrafast excitation relaxation dynamics. Phys.
Stat. Sol. B 241(8):1986–1993.
Kuban P, Berg JM, Dasgupta PK (2004) Durable microfabricated high-speed humidity sensors.
Anal. Chem. 76:2561–2567.
La Clair JJ, Burkart MD (2003) Molecular screening on a compact disc. Org. Biomol. Chem.
1:3244–3249.
Lange SA, Roth G, Wittemann S, Lacoste T, Vetter A, Grässle J, Kopta S, Kolleck M, Breitinger
B, Wick M, Hörber JKH, Dübel S, Bernard A (2006) Measuring biomolecular binding events
with a compact disc player device. Angew. Chem. Int. Ed. 45:270–273.
Lavrik NV, Sepaniak MJ, Datskos PG (2004) Cantilever transducers as a platform for chemical
and biological sensors. Rev. Sci. Instrum. 75:2229–2253.
Leatherdale CA, Bawendi MG (2001) Observation of solvatochromism in CdSe colloidal quantum
dots. Phys. Rev. B 63:165315 1–6.
Mackay RS, Jaconson B (1957) Endoradiosonde. Nature 179:1239–1240.
140 R.A. Potyrailo
Potyrailo RA (2006) Polymeric sensor materials: Toward an alliance of combinatorial and rational
design tools? Angew. Chem. Int. Ed. 45:702–723.
Potyrailo RA, Hassib L (2005) Analytical instrumentation infrastructure for combinatorial and
high-throughput development of formulated discrete and gradient polymeric sensor materials
arrays. Rev. Sci. Instrum. 76:062225.
Potyrailo RA, Lemmon JP (2005) Time-modulated combinatorially developed optical sensors for
determination of non-volatile analytes in complex samples. QSAR Comb. Sci. 24:7–17.
Potyrailo RA, May RJ (2002) Dynamic high throughput screening of chemical libraries using
acoustic-wave sensor system. Rev. Sci. Instrum. 73:1277–1283.
Potyrailo RA, Morris WG (2007) Multianalyte chemical identification and quantitation using a
single radio frequency identification sensor. Anal. Chem. 79:45–51.
Potyrailo RA, Sivavec TM (2004) Boosting sensitivity of organic vapor detection with silicone
block polyimide polymers. Anal. Chem. 76:7023–7027.
Potyrailo RA, Sivavec TM (2005) Dual-response resonant chemical sensors for multianalyte
analysis. Sens. Actuators B 106:249–252.
Potyrailo RA, Takeuchi I, editors. (2005) Special feature on combinatorial and high-throughput
materials research. Meas. Sci. Technol. 316: 1–4
Potyrailo RA, Chisholm BJ, Olson DR, Brennan MJ, Molaison CA (2002) Development of com-
binatorial chemistry methods for coatings: High-throughput screening of abrasion resistance
of coatings libraries. Anal. Chem. 74:5105–5111.
Potyrailo RA, Conrad RC, Ellington AD, Hieftje GM (1998a) Adapting selected nucleic acid
ligands (aptamers) to biosensors. Anal. Chem. 70:3419–3425.
Potyrailo RA, Ding Z, Butts MD, Genovese SE (2008) Selective chemical sensing using structur-
ally colored core-shell colloidal crystal films, 8:815–22.
Potyrailo RA, Ghiradella H, Vertiatchikh A, Dovidenko K, Cournoyer JR, Olson E (2007b)
Morpho butterfly wing scales demonstrate highly selective vapour response. Nat. Photon.
1:123–128 (Cover Story).
Potyrailo RA, Hobbs SE, Hieftje GM (1998b) Optical waveguide sensors in analytical chemistry:
Today’s instrumentation, applications and future development trends. Fresenius’ J. Anal.
Chem. 362:349–373.
Potyrailo RA, Leach AM, Morris WG, Gamage SK (2006a) Chemical sensors based on
micromachined transducers with integrated piezoresistive readout. Anal. Chem. 78:
5633–5638.
Potyrailo RA, May RJ, Sivavec TM (2004b) Recognition and quantification of perchloroethylene,
trichloroethylene, vinyl chloride, and three isomers of dichloroethylene using acoustic-wave
sensor array. Sens. Lett. 2:31–36.
Potyrailo RA, Morris WG, Leach AM, Sivavec TM, Wisnudel MB, Boyette S (2006b) Analog
signal acquisition from computer optical drives for quantitative chemical sensing. Anal. Chem.
78:5893–5899.
Potyrailo RA, Morris WG, Wroczynski RJ (2004c) Multifunctional sensor system for high-
throughput primary, secondary, and tertiary screening of combinatorially developed materials.
Rev. Sci. Instrum. 75:2177–2186.
Potyrailo RA, Szumlas AW, Danielson TL, Johnson M, Hieftje GM (2005d) A dual-parameter
optical sensor fabricated by gradient axial doping of an optical fibre. Meas. Sci. Technol.
16:235–241.
Quercioli F, Tiribilli B, Ascoli C, Baschieri P, Frediani C (1999) Monitoring of an atomic force
microscope cantilever with a compact disk pickup. Rev. Sci. Instrum. 70:3620–3624.
Rabani E, Hetényi B, Berne BJ, Brus LE (1999) Electronic properties of CdSe nanocrystals in the
absence and presence of a dielectric medium. J. Chem. Phys. 110:5355–5369.
Rakow NA, Suslick KS (2000) A colorimetric sensor array for odour visualization. Nature
406:710–713.
Ren M, Forzani ES, Tao N (2005) Chemical sensor based on microfabricated wristwatch tuning
forks. Anal. Chem. 77:2700–2707.
142 R.A. Potyrailo
Rosler S, Lucklum R, Borngraber R, Hartmann J, Hauptmann P (1998) Sensor system for the
detection of organic pollutants in water by thickness shear mode resonators. Sens. Actuators
B 48:415–424.
Savran CA, Knudsen SM, Ellington AD, Manalis SR (2004) Micromechanical detection of pro-
teins using aptamer-based receptor molecules. Anal. Chem. 76:3194–3198.
Scheidtmann J, Frantzen A, Frenzer G, Maier WF (2005) A combinatorial technique for the search
of solid state gas sensor materials. Meas. Sci. Technol. 16:119–127.
Seitz WR (1988) Chemical sensors based on immobilized indicators and fiber optics. CRC Crit.
Rev. Anal. Chem. 19(2):135–173.
Sepaniak MJ, Datskos PG, Lavrik NV, Tipple C (2002) Microcantilever transducers: A new
approach in sensor technology. Anal. Chem. 74:568A–575A.
Sheehan PE, Whitman LJ (2005) Detection limits for nanoscale biosensors. Nano Lett.
5:803–807.
Singh K, Shahi VK (1998) Electrochemical studies on nation membrane. J. Memb. Sci.
140:51–56.
Smit MA, Ocampo AL, Espinosa-Medina MA, Sebastián PJ (2003) A modified Nafion membrane
with in situ polymerized polypyrrole for the direct methanol fuel cell. J. Power Sources
124:59–64.
Steinberg TH, Jones LJ, Haugland RP, Singer VL (1996) SYPRO orange and SYPRO red protein
gel stains: One-step fluorescent staining of denaturing gels for detection of nanogram levels of
protein. Anal. Biochem. 239:223–237.
Su M, Li S, Dravid VP (2003) Microcantilever resonance-based DNA detection with nanoparticle
probes. Appl. Phys. Lett. 82:3562–3564.
Su P-G, Sun Y-L, Lin C-C (2006) A low humidity sensor made of quartz crystal microbalance
coated with multi-walled carbon nanotubes/Nafion composite material films. Sens. Actuators
B 115:338–343.
Suzuki H (2000) Advances in the microfabrication of electrochemical sensors and systems.
Electroanalysis 12:703–715.
Suzuki K, Hirayama E, Sugiyama T, Yasuda K, Okabe H, Citterio D (2002) Ionophore-based
lithium ion film optode realizing multiple color variations utilizing digital color analysis. Anal.
Chem. 74:5766–5773.
Tailoka F, Fray DJ, Kumar RV (2003) Application of Nafion electrolytes for the detection of
humidity in a corrosive atmosphere. Solid State Ionics 161:267–277.
Tani H, Maehana K, Kamidate T (2004) Chip-based bioassay using bacterial sensor strains immo-
bilized in three-dimensional microfluidic network. Anal. Chem. 76:6693–6697.
Taton TA, Mirkin CA, Letsinger RL (2000) Scanometric DNA array detection with nanoparticle
probes. Science 289:1757–1760.
Thompson M, Stone DC (1997) Surface-Launched Acoustic Wave Sensors: Chemical Sensing and
Thin-Film Characterization. New York: Wiley. 196 p.
Thorsen T, Maerkl SJ, Quake SR (2002) Microfluidic large-scale integration. Science
298:580–584.
Tricoli V, Nannetti F (2003) Zeolite-Nafion composites as ion conducting membrane materials.
Electrochim. Acta 48:2625–2633.
Vikalo H, Hassibi B, Hassibi A (2006) A statistical model for microarrays, optimal estimation
algorithms, and limits of performance. IEEE Trans. Signal Proc. 54:2444–2455.
Vilkner T, Janasek D, Manz A (2004) Micro total analysis systems. Recent developments. Anal.
Chem. 76:3373–3386.
Vo-Dinh T, Alarie JP, Isola N, Landis D, Wintenberg AL, Ericson MN (1999) DNA biochip using
a phototransistor integrated circuit. Anal. Chem. 71:358–363.
Wang H, Feng C-D, Sun S-L, Segre CU, Stetter JR (1997) Comparison of conductometric humid-
ity-sensing polymers. Sens. Actuators B 40:211–216.
Wang J (2002) Electrochemical detection for microscale analytical systems: A review. Talanta
56:223–231.
6 Chemical Sensors: New Ideas for the Mature Field 143
Wang J, Musameh M, Lin Y (2003) Solubilization of carbon nanotubes by Nafion toward the
preparation of amperometric biosensors. J. Am. Chem. Soc. 125:2408–2409.
Wang N, Zhang N, Wang M (2006) Wireless sensors in agriculture and food industry – Recent
development and future perspective. Comput. Electron. Agri. 50:1–14.
Want R. (2004) Enabling ubiquitous sensing with RFID. Computer 37(4):84–86.
Ward M, Buttry DA (1990) In situ interfacial mass detection with piezoelectric transducers.
Science 249:1000–1007.
Wetzl BK, Yarmoluk SM, Craig DB, Wolfbeis OS (2004) Chameleon labels for staining and
quantifying proteins. Angew. Chem. Int. Ed. 43:5400–5402.
Wolfbeis OS (2004) Fiber-optic chemical sensors and biosensors. Anal. Chem. 76:3269–3284.
Wolfbeis OS (2006) Fiber-optic chemical sensors and biosensors. Anal. Chem. 78:3859–3874.
Wu R-J, Sun Y-L, Lin C-C, Chen H-W, Chavali M (2006) Composite of TiO2 nanowires and
Nafion as humidity sensor material. Sens. Actuators B 115:198–204.
Yeo SC, Eisenberg A (1977) Physical properties and supermolecular structure of perfluorinated
ion-containing (Nafion) polymers. J. Appl. Polym. Sci. 21:875–898.
Zahedi S, Navid R, Hassibi A (2004) Statistical modeling of biochemical detection systems. Pro-
ceedings of the 26th Annual International Conference of IEEE Engineering in Medicine and
Biology Society, EMBS; San Francisco, CA, USA, pp. 208–211.
Zemel JN (1990) Microfabricated nonoptical chemical sensors. Rev. Sci. Instrum.
61:1579–1606.
Zen J-M, Kumar AS (2001) A mimicking enzyme analogue for chemical sensors. Acc. Chem.
Res. 34:772–780.
Zhang C, Suslick KS (2005) Colorimetric sensor array for organics in water. J. Am. Chem. Soc.
127:11548–11549.
Zhang Y, Ji H-F, Brown GM, Thundat T (2003) Detection of CrO42− using a hydrogel swelling
microcantilever sensor. Anal. Chem. 75:4773–4777.
Chapter 7
Applications of Functional Thin Films
for Mechanical Sensing
Chang Liu
Mechanical sensing plays very important roles in today’s world, with applications
such as crash sensors in automotive airbag deployment systems, flow sensors for
monitoring and controlling the flow in semiconductor equipment, blood pressure
sensors and heartbeat monitors in pacemakers, and acceleration sensors for motion-
sensitive electronics gaming controllers. With the growing needs of autonomous
systems and interactive communication and entertainment systems, it is foreseeable
that the needs for high-performance and low-cost mechanical sensors will continue
to grow.
Mechanical sensing is a vast field encompassing many variables, including:
position, displacement, force, stress, strain (deformation), contact pressure, speed,
acceleration, vibration, rotational velocity, fluid pressure, fluid velocity, fluid shear
stress, touch, roughness, texture, and softness or hardness.
C. Liu
Department of Mechanical Engineering, Northwestern University, 2145 Sheridan Road, Rm.,
Evanston, IL60208-3111, USA
e-mail: changliu@northwestern.edu
A. Zribi and J. Fortin (eds.), Functional Thin Films and Nanostructures for Sensors, 145
Integrated Analytical Systems,
DOI: 10.1007/978-0-387-68609-7_7, © Springer Science + Business Media, LLC 2009
146 C. Liu
Principle of Transduction
Functional films may occasionally be used for displacement. For example, func-
tional thermal electric-based temperature sensors may be used for measuring tem-
perature changes. Displacement may also alter optical reflectance, transmittance, or
148 C. Liu
the refractive index of the device, and the optical signal can be picked up by
functional thin films.
Functional films are widely used for monitoring changes of mechanical strain. This
is the focus of this chapter. Mechanical strain can be detected by using piezoelectric
films or piezoresistive films. As mentioned in Chapter 2, a piezoelectric film exhib-
its the behavior of direct piezoelectric effect, charge accumulation due to applied
stress/strain. A piezoresistive film may be made of an elemental metal, a metal
alloy, doped semiconductor, or organic polymers. Their resistance changes as a
function of applied stress/strain.
Functional films have other applications for mechanical sensors beyond displace-
ment and stress sensing. These are outlined in this section.
Interestingly, mechanical sensors often involve actuators. Actuators may be used
to actively tune the properties of mechanical elements in order to change the char-
acteristics of the mechanical sensor on the fly, with the purpose of increasing the
sensitivity, increasing the dynamic range, or reducing the noise floor.
Actuators connected with beams or membranes have been used in the past. They
may be based on functional films such as piezoelectric films. A piezoelectric film
also exhibits a reverse piezoelectric effect alongside the direct piezoelectric effect
(for sensing). According to the reverse piezoelectric effect, an applied electric field
changes the stress/strain profile in the piezoelectric film and thereby changes the
position or stiffness of mechanical elements.
Active tuning of sensor characteristics can also be achieved using electrostatic
actuators, thermal bimetallic actuators, or magnetic actuators, to name a few.
Other functional thin films, such as hydrogels, electroactive polymers, and con-
ductive polymers, have also been used for actuation.
used in mechanical sensors are reviewed here and their methods of preparation
discussed. A detailed discussion of material preparation methods for piezoelectric
and piezoresistive materials can be found in Liu (2005) and Madou (2002).
Piezoelectric Films
easily (Luginbuhl et al. 1997; Wang et al. 2003), using single- or multiple-layer
deposition.
Using a processing technique called screen printing, even thicker PZT films
can be reached in a single pass (Bernstein et al. 1997; Chen et al. 1994; Walter
et al. 2002), with the highest piezoelectric coupling coefficient being 50 pC/N,
significantly lower than what is achievable in bulk PZT. The screen printing ink
consists of submicron PZT powders obtained commercially, and lithium car-
bonate and bismuth oxide as the bonding agent. After screen printing, the
deposited materials are dried and then fired at high temperature for densifica-
tion. The sol–gel deposition process is constantly being advanced. Pinhole-free
PZT films up to 12 mm thick have been realized with d33 in the 140–240 pC/N
range (Bernstein et al. 1997), although single-layer deposition thickness of 0.1
mm is more common.
ZnO material can be grown using a number of methods, including rf or dc sput-
tering, ion plating, and chemical vapor deposition. In the MEMS field, ZnO is most
commonly deposited by magnetron sputtering (Wenzel and White 1988; Yamamoto
et al. 1980) on various materials, with the c-axis (or Z-axis) close to the normal of
a substrate. For ZnO, the c-axis is spontaneously formed without poling.
Strategies for reducing the intrinsic stress of ZnO have been explored in order to
realize large-area thicker films (Zesch et al. 1991). As-deposited ZnO films have
significant compressive stress, ranging from 1 GPa to 135 MPa (DeVoe and Pisano
2001). The stress can be reduced using thermal annealing (e.g., at 500°C for 5 min)
to the 100 to 80 MPa range.
A popular electrode material on top of the ZnO thin film is aluminum, which can
be etched using a solution of KOH, K3Fe(CN)6, and water (1 g:10 g:100 ml). ZnO
itself can be etched using wet etchants such as CH3COOH:H3PO4:water (1:1:80 ml)
at fast rate (Niu and Kim 2003).
Piezoresistive Films
The most commonly used thin films that exhibit piezoresistive properties are doped
silicon. A schematic diagram of a representative process used for doping selective
regions of silicon with dopant atoms is shown in Fig. 7.1. The desired shape of the
resistor is shown in the topmost Fig.. The resistor feature should be moderately
doped (with concentration ranging from 1015 to 1018 cm−3). The two ends of the
resistor should have higher doping concentration, on the order of 1019–1020 cm−3, in
order to form ohmic contacts with metal leads.
Strain gauges made of thin-film metals do not compare favorably with semi-
conductor strain gauges in terms of piezoresistive gauge factors. However, metal
thin films provide sufficient performance for many applications. Using metal
instead of a semiconductor eliminates the needs of doping and lengthy process
steps. Also metal resistors can be deposited and processed under temperatures
much lower than what would be needed for doping semiconductors. Metal can
7 Applications of Functional Thin Films for Mechanical Sensing 151
contact region
resistor (moderately doped)
with metal leads
(heavily doped)
A A’
generally sustain much greater elongation before fracture. As such, metal resistors
can be placed on polymer materials for polymer MEMS devices (e.g., tactile
sensors; Engel et al. 2003) and provide improved mechanical robustness compared
with silicon counterparts.
Nanocomposite polymer is a new class of promising piezoresistive material. By
doping nonconductive elastomers with conducting particles such as metal powders,
carbon black, and carbon nanotubes, one can turn a nonconducting organic material
into a conducting one through the networks of conducting particles dispersed
uniformly within. A typical resistivity curve as a function of doped conducting
particles is illustrated in (Fig. 7.2). At a percolation threshold, the resistivity of the
matrix drastically decreases.
The goal of designing a successful blend of nanocomposite conductive material
lies in the ability to achieve resistivity transition at a low doping level (or loading
level), thereby avoiding mechanically stiffening the polymer matrix.
152 C. Liu
10
Log resistivity (Ω cm)
2
0.0 0.2 0.4 0.6 0.8 1.0
Fig. 7.2 The resistivity of a nanocomposite elastomer decreases with increasing loading concen-
tration. At the percolation threshold, the resistivity reduces drastically due to formation of the
cooperative conducting network
It has been demonstrated that by doping the polymer matrix with high aspect
ratio carbon nanotubes (multiwall), the percolation limit for resistivity reduction is
on the order of 0.5% by volume. This represents a 20× decrease from the material
loaded with carbon black particles, which have aspect ratios of roughly 1:1.
The increased aspect ratio of CNTS (on the order of 100:1 or higher) significantly
reduces the percolation threshold.
The nanocomposite elastomer is exciting because it is a multifunctional mate-
rial. As shown in Fig. 7.3, the resistance of a resistor formed by the nanocomposite
material may change when a mechanical strain is placed on it. This is because the
mechanical strain changes the concentration and conductive pathways. We have
shown that the effective gauge factor resulting from this piezoresistive behavior can
be as large as 10.
Because many physical sensors rely on the measurement of displacement and stress
caused by an external event, it is important to analyze the displacement and stress.
The analysis of beams and membranes under complex loading conditions is suffi-
ciently covered in textbooks such as Gere and Timoshenko (1997). In this chapter
we simply provide the most commonly used formula for analyzing stress and
7 Applications of Functional Thin Films for Mechanical Sensing 153
Fig. 7.3 (Top) Scanning electron micrograph of cross-section of PDMS with exposed CNT tubes.
(Bottom) Schematic diagram of piezoresistivity behavior
2. The guided boundary conditions allow two linear DOFs but restrict the rota-
tional DOF.
3. The free boundary conditions provide for both linear DOFS and rotation. At a
free end, a point on a beam may translate and rotate. A representative example
is the free end of a diving board.
These three distinct types of boundary conditions are graphically represented in
Table 7.2.
The general method for calculating the curvature of a beam under small displace-
ment is to solve a second-order differential equation of a beam:
d2 y
EI = M ( x ), (7.1)
dx 2
where M(x) represents the bending moment at the cross-section at location x and y
the displacement at location x. The x-axis runs along the longitudinal direction of
the cantilever. This second-order differential equation can be solved by the two
boundary conditions at the two ends of a beam. However, due to space limitations,
we do not discuss the solution in detail.
Rather, Table 7.3 summarizes the most commonly encountered loading condi-
tions and the maximum vertical and angular displacement, whereas F is the point
loading force (N, Newton), l the length of beam (m), E the Young’s modulus of
beam material (N/m2), d the vertical displacement (m), and q the angular displace-
ment (arc angle).
7 Applications of Functional Thin Films for Mechanical Sensing 155
Table 7.3 Most Commonly Encountered Beam Loading Conditions and Displacement
End constraints and loading conditions Maximum values of stress and
displacement
Maximum q occurs at the free end
Fl 2
Max value of θ =
2EI
Maximum vertical displacement occurs at the
free end
Fixed-free beam under point loading at the free end. Fl 3
Max. value d=
3EI
q at the free end equals zero due to guided
boundary condition.
Fl 3
Max. value d=
Fixed-guided beam under point loading at the free 12EI
end.
Wl 3
Max. value d=
192EI
A fixed-fixed beam with point loading applied at
the center.
Maximum q occurs at the end of the beam
Ml
Max q =
EI
Maximum vertical displacement occurs at the
free end
Fixed-free beam under a torque loading. The
position of the torque along the length is not Ml 2
Max d=
relevant. 2EI
The analysis of intrinsic stress in a beam under flexural loading can be very com-
plicated when the loading condition is complex (e.g., multiple loading, rotational
loading, etc.). Here in this section, we focus on the stress versus loading analysis
for the most common beam, the cantilever, a fixed-free beam under point loading
at the free end.
156 C. Liu
When a beam is loaded by force or couples, stresses and strains are created
throughout the interior of the beam. Loads may be applied at a concentrated loca-
tion (concentrated load), or distributed over a length or region (distributed load).
To determine the magnitude of these stresses and strains, one first must find the
internal forces and internal couples that act on cross-sections of the beam.
The distribution of longitudinal stress is first described qualitatively (Fig. 7.4).
Under a transverse loading of a concentrated force at the free end, the torque distribu-
tion through the length of the beam is nonuniform; it is zero at the free end and
reaches a maxim at the fixed end. At any cross-section, the signs of longitudinal
7 Applications of Functional Thin Films for Mechanical Sensing 157
stresses change across the neural axis. The magnitude of stresses at any point on the
cross-section is linearly proportional with respect to the distance to the neutral axis.
The magnitude of the maximum stress associated with individual cross-sections
changes linearly with respect to the distance to the free end, reaching a sectionwide
maximum at the top and bottom surfaces. These are the reasons why piezoresistors
are commonly found on the surface of a cantilever and near the fixed end.
The maximum strain for the entire cantilever occurs at the fixed end, where
x = L. In fact, in many routine design tasks, the sole interest is to find the magnitude
of the maximum stress/strain at the fixed end. The maximum strain is expressed as
a function of the total torque M(x):
M ( x )t FLt (7.2)
e max = = .
2 EI 2 EI
Membranes are often used in microsensors. In this section, we review one of the
simplest cases of membrane loading: by a uniformly distributed pressure on one
side.
The governing equation for membrane displacement under a uniform pressure
loading p is:
∂4 w ∂4 w ∂4 w p
+ 2 + = , (7.3)
∂x 4 ∂x 2 ∂y 2 ∂y 4 D
Et 3
D= . (7.4)
12(1 − n 2 )
In the case of a square membrane with fixed boundaries, the two-dimensional distribution
of membrane displacement and the magnitude of longitudinal stress along the x-axis are
illustrated in Fig. 7.5. Several important qualitative observations can be made:
1. The maximum displacement occurs at the center of the diaphragm.
2. The maximum stress occurs at the center points of two opposite edges and in the
center of the membrane. The stress along the edge and the center have different
signs. These locations with high stress value are preferred for placement of
piezoresistive sensors for detecting membrane deformation.
In many application cases, only the maximum displacement and the maximum
stress are of interest. These can be calculated using an empirical formula. The
158 C. Liu
Fig. 7.5 Normalized displacement (left) and stress in the x-axis (right)
σmax
p
a
b wcenter
σcenter
a
apb 4
wcenter = , (7.5)
Et 3
with the value of the proportional constant a determined by the ratio of a to b.
The value of a can be found by using the look-up table in Fig. 7.6. The maximum
stress (at the center point of the long edge) and the stress in the center of the plate
are:
b1 pb 2
s mas = , (7.6)
t2
b 2 pb 2 (7.7)
s center = ,
t2
with the values of b1 and b2 listed in the table as well.
7 Applications of Functional Thin Films for Mechanical Sensing 159
There are many published works that cover mechanical sensors based on functional
thin films. Instead of citing and reviewing each distinct work, we provide one
example of a mechanical sensor that is inspired by the biological haircell. This
example is meant to serve a few purposes:
1. It will outline the typical steps for designing and developing a mechanical
sensor.
2. It will illustrate the fabrication process and material preparation of functional
thin films.
3. It will illustrate the multitude of functions that mechanical sensors can achieve.
My group is developing artificial haircell sensors that mimic the haircell sensor,
widely found in many animals, that performs a large variety of functions. The bio-
logical haircell, a common neuronal mechanoreceptor, is responsible for a wide
variety of sensing in different animal species. A haircell consists of a cilium attached
to a neuronal cell. When the cilium is displaced mechanically, the neuron mem-
brane is placed under stress and the neuron fires electrical pulses, analogous to
digital signals sent through computer bus lines.
Although the structure is exceedingly simple, haircells in biology cover a wide
range of mechanical sensing functions (illustrated in Fig. 7.7). Haircells are respon-
sible for hearing (human cochlea), flow sensing (insects, spiders, and fish), vibra-
tion sensing (insects), equilibrium sensing (human inner ear), and joint angle
sensing (insect), to name a few examples. Since 1998, our group has been developing
artificial haircell sensors as modular building blocks of sensors for flow, vibration,
touch, and acoustic vibration.
Microfabrication and miniaturization technology have been growing rapidly in
the past ten years with the advancement of microelectromechanical systems
(MEMS) technology. MEMS technology offers potentials for realizing high-
performance sensors that contain integrated electronics for on-chip, low noise signal
amplification and efficient signal processing. Sensor performance is uniform as the
photolithography technique ensures high degrees of precision. The technology is
potentially low cost because of the batch processing similar to integrated circuits.
However, the argument for the low-cost aspect of the MEMS technology with
relevance to robotics sensors must be made carefully. The low-cost promise can be
realized if the demand for a particular sensor is high, and hence reduces the costs
of one-time setup and design. Examples include the accelerometers made by Analog
Devices and Motorola. Because the demand for the automotive air-bag is high, the
sensors can be made commercially in such a manner that early research and devel-
opment costs can be compensated.
This scenario is not necessarily true for all sensors that are important for robotics
use. Many sensors, such as flow sensors, tactile sensors, and gyros/balance sensors,
are associated with small commercial demand and unique performance character-
istics. The lack of the economy of scale therefore can prevent applications of advanced
sensors in robotics systems.
160 C. Liu
Haircell
Acoustic sensor
Vibration sensor
Flow sensor
Contact sensor
Fig. 7.8 (a) Scanning electron micrograph of an artificial haircell sensor; (b) a schematic
diagram; (c) optical micrograph of an array
where w and t are the width and thickness of the horizontal cantilever beam. The
term I denotes the moment of inertia of the horizontal support beam. The stress
distribution has been simulated using ANSYS finite element software. The result-
ant diagram of stress before and after lateral loading is shown in Fig. 7.9.
The length of the horizontal cantilever beam is on the order of 200–600 mm.
The length of the vertical cilium is approximately the same as the length of the
horizontal beam.
The scanning electron micrograph and optical micrograph of the finished artificial
haircell are shown in Fig. 7.8b,c.
The fabrication process uses a silicon-on-insulator (SOI) wafer that consists of
three layers: a top layer thin epitaxial, single crystal silicon film (approximately 10 mm
thick), a sandwiched silicon dioxide layer (insulator), and a supporting substrate
(500 mm thick) (Fig. 7.10). First, the top epitaxial layer is selected doped into the
shape of a piezoresistor. The doping is completed by using ion implantation of
boron atoms into a silicon lattice. The designed stopping range of the boron atom
is half the thickness of the epitaxial layer.
An oxide insulator is grown over the top surface of the SOI wafer (step b).
The oxide layer is selectively patterned by first depositing a photoresist layer on
top, photolithographically patterning the photoresist, and etching the exposed
silicon oxide through the photoresist mask. The photoresist is then removed by
using organic solvents.
162 C. Liu
A metal conductor layer is then deposited and patterned. It works wire leads that
connect from the piezoresistors to the periphery of the chip. The wafer is covered
again with a photoresist thin film which is then photolithographically patterned to
form the cantilever. The photoresist serves as a mask to etch the epitaxial silicon to
form the cantilever (d). The wafer’s back side is covered with photoresist and pat-
terned to form deep, through wafer trenches that expose the underlying silicon
oxide layer. The wafer is turned around again. This time the front side is covered
with a thick layer of photodefinable epoxy film called SU-8. The thickness of the
SU-8 layer corresponds to the height of the hair in the end. The SU-8 epoxy is
photolithographically patterned to form the artificial cilium. Finally, the oxide is
removed to allow the cantilever to be free-standing.
7 Applications of Functional Thin Films for Mechanical Sensing 163
This process is carefully designed such that every step that involves etching
achieves maximum selectivity. In step e, the deep trench etching of silicon has very
low etch rate on oxide, therefore the process is self-terminating. In step g, the oxide
etch has minimum effect on the hair and the substrate.
The artificial haircell sensor can be used to measure vibration, touch, and flow
velocity. For flow velocity measurement, moving fluid past the cilium introduces
frictional force and drag force on the cilium, which in turn generates momentum
and stress distribution at the base of the hair. The haircell sensors have been pack-
aged and tested in water tunnels at various flow speeds. The response of the output
as a function of the input velocity is illustrated in Fig. 7.11. The minimum detection
limit for the sensor is 0.5 mm/s in water.
Conclusions
This section outlines the basic issues, challenges, and opportunities of applying
functional films for mechanical sensors. The types of mechanical sensors were
outlined first. Then the principles that are often used for measuring displacement
and stress within beams and membranes were discussed. Subsequently, the methods
commonly used for preparing functional thin films for integrated sensors were
164 C. Liu
0.8
Sensor Output (V)
0.6
0.4
0.2
0
0.0 0.1 0.2 0.3 0.4
Inflow velocity (m/s)
References
Bernstein JJ, Finberg SL, Houston K, Niles LC, Chen HD, Cross LE, Li KK, Udayakumar K
(1997) Micromachined high frequency ferroelectric sonar transducers, IEEE Transactions on
Ultrasonics, Ferroelectrics and Frequency Control, 44:960–969
Chen HD, Udayakumar KR, Cross LE, Bernstein JJ, Niles LC (1994) Development and electrical
Characterization of lead zirconate titanate thick films on silicon substrates, Proceedings of the
Ninth IEEE International Symposium on Applications of Ferroelectrics, University Park,
Pennsylvania, August 7–10, 1994
DeVoe DL, Pisano AP (2001) Surface micromachined piezoelectric accelerometers (PiXLs),
Journal of Microelectromechanical Systems, 10:180–186
Engel J, Chen J, Liu C (2003) Surface micromachined piezoelectric accelerometers (PiXLs),
Journal of Microelectromechanical Systems, 10: 180–186
Gere JM, Timoshenko SP (1997) Mechanics of Materials, 4th edn. PWS, New York
Liu C (2005) Foundations of MEMS. Prentice-Hall, Englewood Cliffs, NJ
Lu J, Chu J, Huang W, Ping Z (2003) Microstructure and electrical properties of Pb(Zr, Ti)O3 thick
film prepared by electrostatic spray deposition, Sensors and Actuators A: Physical, 108:2–6
Luginbuhl P, Collins SD, Racine GA, Gretillat MA, De Rooij NF, Brooks KG, Setter N (1997)
Microfabricated lamb wave device based on PZT sol-gel thin film for mechanical transport of
solid particles and liquids, Journal of Microelectromechanical Systems, 6:337–346
Madou MJ (2002), Fundamentals of Microfabrication: The Science of Miniaturization, 2nd edn.
CRC Press, Boca Raton, FL
7 Applications of Functional Thin Films for Mechanical Sensing 165
Niu MN, Kim ES (2003) Piezoelectric bimorph microphone built on micromachined parylene
diaphragm, Journal of Microelectromechanical Systems, 12:892–898
Walter V, Delobelle P, Moal PL, Joseph E, Collet M (2002) A piezo-mechanical characterization
of PZT thick films screen-printed on alumina substrate, Sensors and Actuators A: Physical,
96:157–166
Wang LP, Wolf RA, Wang Y, Deng KK, Zou L, Davis RJ, Trolier-McKinstry S (2003) Design,
fabrication, and measurement of high-sensitivity piezoelectric microelectromechanical sys-
tems accelerometers, Journal of Microelectromechanical Systems, 12:433–439
Wenzel SW, White RM (1988) A multisensor employing an ultrasonic Lamb-wave oscillator,
IEEE Transactions on Electron Devices, 35:735–743
Yamamoto T, Shiosaki T, Kawabata A (1980) Characterization of ZnO piezoelectric films pre-
pared by RF planar magnetron sputtering, Journal of Applied Physics, 51:3113–3120
Zesch JC, Hadimioglu B, Khuri-Yakub BT, Lim M, Lujan R, Ho J, Akamine S, Steinmetz D,
Quate CF, Rawson EG (1991) Deposition of highly oriented low-stress ZnO films, Proceedings
of IEEE Ultrasonics Symposium, 1:445–448
Chapter 8
Sensing Infrared and Terahertz Regions
by Functional Films
Abstract Designing functional films and nanostructures has a key role in the
performance of the infrared (IR) sensing and terahertz (THz) sensing that are
based particularly on quantum well, wire, and dot structures. Sensing the electro-
magnetic (EM) spectra is an extremely important issue for various fields, from
understanding the universe, living cells, and elementary particles to numerous
applications. To give a glimpse of the field in connection to functional films and
nanostructures as sensing elements, in this chapter we briefly discuss infrared
(IR) sensing and terahertz (THz) sensing. For IR sensing we limit ourselves to
low-dimensional semiconductor functional films. For THz sensing we discuss:
(a) how strain in thin films influences THz absorption from impurities, (b) plasma
effects in two-dimensional electron gas (2DEG), and (c) ultrasensitive bolometers
based on metal films.
Introduction
A. Zribi and J. Fortin (eds.), Functional Thin Films and Nanostructures for Sensors, 167
Integrated Analytical Systems,
DOI: 10.1007/978-0-387-68609-7_8, © Springer Science + Business Media, LLC 2009
168 M. Willander et al.
Most QWIPs, QRIPs, and QDIPs are based on vertical heterostructures consisting
of one or several QWs or two-dimensional arrays of QRs or QDs separated by the
barrier layers. The QW, QR, or QD structures serving as the photodetector active
region, where IR radiation is absorbed, are sandwiched between heavily doped
emitter and collector contact layers. The active region can be either doped (with
dopants of the same type as the contact layers) or undoped. Usually the photodetec-
tors in question are made of n+-N-n-N-n+—or N+-N-n-N-N+-heterostructures with
n+—or N+-contact layers, respectively, N-type barrier layers, and n-type QWs, QRs,
170 M. Willander et al.
Fig. 8.1 Schematic view of (a) QWIP, (b) QRIP, and (c) QDIP structures (Ryzhii et al. 2004 a-b)
or QDs. Schematic views of vertical QWIP, QRIP, and QDIP device structures are
shown in Fig. 8.1.
The absorption of IR is associated with the electron intersubband transitions
from bound states in QWs, QRs, or QDs into continuum states above the barriers
or into excited quasi-bound states near the barrier top. The bound-to-continuum
transitions or bound-to-quasi-bound transitions followed by fast escape into the
continuum result in the photoionization of QWs, QRs, or QDs and the appearance
of mobile electrons. Bound electrons accumulated in QWs, QRs, or QDs can create
a significant space charge in the active region. In photodetectors made of N+-N-n-
N-N+ heterostructures with the same material of the contact and barrier layers, the
electrons are injected from the emitter to the active region overcoming a potential
barrier in the latter formed solely by the space charge. Hence, the electron injection
in such photodetectors is of thermionic origin.
Due to a conduction band offset at the n+ -N heterointerface in n+-N-n-N-n+
photodetectors, the pertinent heterobarrier prevents the penetration of electrons
from the emitter contact to the active region. However, a sufficiently strong electric
field at this heterointerface, caused by the external electric field and the space
charge, can result in a marked electron tunneling to the active region providing the
electron tunneling injection. The QWIPs fabricated, studied, and used in applica-
tions are primarily based on heterostructures with tunneling injection, for example,
on heterostructures with n+–GaAs contact layers and QWs separated by N-AlGaAs
barrier layers. However, the majority of QDIPs are made of QD structures with the
same material (e.g., GaAs) as the contact and barrier layers, although QDIPs with
more exotic structures were investigated.
Under a bias voltage applied between the emitter and collector contacts the cur-
rent across the active region depends on the applied voltage and the injection condi-
tions (which are dictated by properties of the emitter contact). In normal operation
mode the current is limited by the space charge formed by electrons captured in
QWs, QRs, or QDs. Under dark conditions, the space charge in the active region is
8 Sensing Infrared and Terahertz Regions by Functional Films 171
(a) (b)
Emitter
Emitter
QDs (QRs) QDs (QRs)
Collector Collector
Fig. 8.2 Schematic view of the conduction band profiles in single QR array QRIP or single QD
array QDIP with thermionic (a) and tunneling (b) injection. Arrows indicate electron trajectories
in the active regions (Ryzhii et al. 2004 a-b)
determined by the balance between the processes of the electron capture into QWs
(QRs or QDs) and the processes of the electron thermoemission (or tunneling) from
them. The space charge modifies the electric field distribution and affects the bar-
rier at the emitter edge of the active region (in photodetectors with tunneling injec-
tion) or leads to the formation of a controlled potential barrier inside this region.
Under IR illumination, the photoionization of QWs, QRs, or QDs shifts the
balance between the electron capture and emission resulting in the redistribution of
the electric field in the active region which, in turn, gives rise to a change in the
injected current. The total current across the photodetectors includes two compo-
nents: the current caused by the electrons emitted from QWs, QRs, or QDs and the
injected current. Because the capture of mobile electrons is usually a rather slow
process, the main portion of the dark current and photocurrent is due to the injec-
tion. The conduction band profiles in a QRIP and QDIP (with a single array of QRs
and QDs, respectively) and the electron trajectories in their active regions are
shown schematically in Fig. 8.2.
Thus the operation of QWIPs, QRIPs, and QDIPs is associated with the current
across the photodetector active region limited by the bound space charge, which is
controlled by the incident IR radiation. Despite similarities in the QWIP, QRIP, and
QDIP principles of operation, there are some distinctions (Liu et al. 2004; Ryzhii
et al. 2004 a-b):
1. Different degree of the discreteness of the energy spectrum of bound electrons
and, therefore, different statistics of these electrons, capture probability, and
selection rules for intersubband transitions
2. Different spatial distributions of the electric potential in the active region, par-
ticularly in the lateral direction (virtually uniform in QWIPs and strongly non-
uniform in QRIPs and QDIPs with low-density QR and QD arrays)
3. Different dependences of the electron capture probability on the concentration
(number) of bound electrons
172 M. Willander et al.
The dark current and photocurrent in the photodetectors in question are determined,
first of all, by the rates of thermionic emission from QWs (QR or QD arrays) and
their photoemission by IR. In QWIPs with multiple QWs as well as in multiarray
QDIPs and QRIPs, the contribution of different QWs or arrays can be slightly dif-
ferent. However, for a qualitative study, one can disregard this difference.
The interaction between the electron gas in each QW and the gas of mobile
electrons propagating over continuum states above the inter-QW barriers is rather
weak. Due to this, the energy distributions of electrons in QWs are given by the
Fermi distribution function with the temperature coinciding with the lattice tem-
perature T. At the same time, the electron sheet concentrations in QWs can be far
different from those obtained when the whole electron system (which includes
electrons in QWs, mobile electrons, and electrons in the contacts) is in equilibrium.
In this case, the rate of thermoemission from QWs and QR (or QD) array is deter-
mined by the activation energy ea = ei − eF. Here ei and eF are, respectively, the ioni-
zation energy of the QWs (QRs or QDs) and their Fermi energy (see Fig. 8.3).
The electron gas in QWs can be considered as a two-dimensional system, so
the Fermi energy of electrons in QWs with respect to the bottom of the lowest
subband equals
⎡ p 2 Σ ⎤ p 2 Σ
e F(QW ) = k BTIn ⎢exp( ) − 1⎥ ≈ , (8.1)
⎢ mk BT ⎥ m
⎣ ⎦
where Σ is the average electron sheet concentration in QWs (or, in the follow-
ing, in QR or QD arrays), ¯h and κB are the reduced Planck constant and the
Boltzmann constant, respectively, and m is the electron mass. The last term in
the right-hand side of Equation (8.1) corresponds to the situation where
k BT < p 2 Σ / m .
8 Sensing Infrared and Terahertz Regions by Functional Films 173
(a) (b)
εa
εa
εi εi
QR subband QD shells
bottoms
εa εa
εi εi
Fig. 8.3 Energy spectra of (a) QRs and (b) QDs (Ryzhii et al. 2004 a-b)
e (QW )
⎡
⎢
= k BT In ⎢exp(
p 2 Σ ⎤
⎥
) − 1⎥ ≈
p 2 Σ
mk BT a 2 ∑ QR ⎥ m a ∑ QR
F 2
⎢ (8.2)
⎣ ⎦
−1/ 2
where a and ∑ QR are the width of QRs and the lateral distance between QRs
(lateral period of the QR array), respectively. In the case of QRIPs with rather nar-
row QRs having only one energy level of lateral quantization, the electron system
in each QR constitutes a one-dimensional gas. In this case, one obtains:
Σ
2
p 2
e F(QR ) ≈ . (8.3)
8m ∑ QR
Relatively large QDs (in the lateral directions) can have several quantum shells and
be able to accept a rather large number of electrons. In most experiments, QDIPs
with relatively large QDs were studied. A QD array with QDs of this type can be
considered as a disintegrated QW. Because many quantum shells can be occupied
by electrons in these QDs, one can neglect the discreteness of their electron levels.
174 M. Willander et al.
In this case, a similar formula to that used for QWs can be used to compute the
activation energy. However, it is necessary to take into account that real sheet
electron density inside the QD equals Σ / a 2 ∑ QD . In the previous expression,
a and ∑ QD are the QD lateral size and the sheet density of QDs in each array,
respectively. The QD density is limited by the condition of a weak overlap of the
wave functions of neighboring QDs (Liu et al. 2004). Otherwise, the probability of
the electron photoemission by normal incident IR can become fairly small; see, for
example, Helm et al. (1994). This means that the product a ∑ QD should not be
2
⎡ p 2 ∑ ⎤ p 2 ∑
e (QD ) ⎢
= k BTIn exp( ⎥
) −1 ≈ . (8.4)
⎢ mk BTa 2 ∑ QD ⎥ ma 2 ∑
F
⎣ ⎦ QD
Both simplified and rather detailed device models of QWIPs with tunneling and
thermionic injection (Liu 1992; Ryzhii 1997; Ryzhii and Liu 1999; Ryzhii et al.
2002 a-b) lead to the following relationship which provides an estimate of the ther-
mal dark current,
eGth
j th ≈ , (8.5)
p
where e is the electron charge, Gth is the rate of thermoemission (per unit area of
a QW, a QR, or QD array), and 〈 p〉 is the average probability of the capture of a
mobile electron passing across a QW.
Considering Equations (8.1), (8.2), (8.4), and (8.5) and introducing ∑ T = mk BT / p ,
2
⎡ ⎤
j th(QWIP ) α
1 ⎢exp( ∑ ) − 1⎥ exp(− ei )
p
(QW )
⎢
⎣
∑T ⎥
⎦
k BT
(8.6)
1 ∑ ei
≈ exp( )exp(− )
p
(QW )
∑ T
k BT
a 2 ∑ QR ⎡ ∑ ⎤
e
j (L − QRIP )
α ⎢exp( ) − 1⎥ exp(− i )
⎢ ⎥
∑T a 2 ∑ QR
th (QR )
p k BT
⎢⎣ ⎥⎦
a 2 ∑ QR ∑ e
(8.7)
≈ exp( )exp(− i )
∑ a 2 ∑ QR
(QR )
p k BT
T
8 Sensing Infrared and Terahertz Regions by Functional Films 175
a 2 ∑ QD ⎡ ∑ ⎤
j (L − QDIP )
α ⎢exp( ) − 1⎥ exp(− ei )
⎢ ∑ T a ∑ QD ⎥
th (QD ) 2
p k BT
⎣ ⎦
(8.8)
a 2 ∑ QR ∑ ei
≈ exp( )exp(− ).
p
(QR )
∑ T
a 2
∑ QD
k BT
2
a 2 ∑ QR p ∑ ei
( S − QRIP )
j α exp( exp( − ).
8∑ T ∑ QR
th ( QR )
p k BT (8.9)
In the case of relatively small QDs having a single quantum shell (with maxi-
mum two electrons), considering that e a = e i (see Fig. 8.3), for the thermoemission
rate one obtains:
∑ ei
jth( S −QDIP ) α exp( − ) . (8.10)
∑
( QD )
2 p k BT
QD
Capture Probability
Ryzhii 1999; Ryzhii et al. 1999; Kochman et al. 2003) with the average kinetic
energy of mobile electrons e >> k BT , the average capture probability exhibits a
steep roll-off with increasing electric field. This, according to Equation (8.10),
results in a steeply increasing dark current–voltage characteristic.
The processes of the electron capture in QRIPs and QDIPs, as already pointed
out, have distinctive features. First of all, the quasi-discreteness and discreteness
of the electron spectra in QRs in QDs can substantially affect the electron
capture. This effect can be particularly pronounced in QDIPs when the quantum-
level separation exceeds the energy of polar optical phonons leading to the
phonon “bottleneck” in the electron capture. Second, due to a limited number of
quantum states in each QD, these states can be fully occupied preventing the
electron capture (because of the Pauli exclusion principle) under certain condi-
tions including excessive doping of the active region and/or large applied voltage.
Third, in QRIPs and QDIPs with low-density QR and QD arrays, the negative
potential of QRs and QDs charged by the captured electrons can result in an
effective repulsion of mobile electrons. This can be a substantial factor limiting
their capture (Ryzhii et al. 2000 a-b, 2001 a-c; Ryzhii 2001; Sergeev et al. 2002).
In particular, in QDIPs, the capture probability can be presented in the following
form (Ryzhii et al. 2001 a-c; Ryzhii 2001),
⎡max N
∑ QD − ∑ ⎤⎦ e ∑
2
∝ (a 2 ∑ QD ) ⎣
( QD )
p exp( − ) (8.11)
max N ∑ QD e CQD ∑ QD
where CQD is the efficient QD capacitance that depends on the QD size as well as
the spacing between QDs. This formula describes the variation of the capture prob-
ability with changing average QD occupancy N = ∑ / max N ∑ QD, where max N
is the maximum number of electrons that can be accepted by a QD. It shows that
p
( QD )
tends to zero when ∑ approaches the maximum value allowed by the
Pauli principle, that is, to max N ∑ QD . Equation (8.11) also shows that p
( QD )
as
a function of ∑ contains an exponential factor associated with the effect of the
repulsion of mobile electrons. Assuming that e corresponds to the temperature T
= 80 K and setting ∑ = ∑ QD , C ≈ 2 ae r ( e r = 12 is the dielectric constant),
QD
p 3/ 2
and a = 15 nm, for the exponential factor in Equation (8.11) one obtains ~ .04. One
can show that the p (QR ) also steeply decreases with increasing ∑ . Equations
(8.6)–(8.9) indicate that the thermal dark current in L-QRIPs and L-QDIPs are fairly
sensitive functions of ∑ and they are steeper than this current in QWIPs. Strongly
∑ and p – ∑ dependences result in an even more
( QD ) ( QR )
decreasing p –
dramatic rise of the thermal dark current in L-QRIPs and L-QDIPs with increasing
∑ and its much higher values compared to those in QWIPs.
8 Sensing Infrared and Terahertz Regions by Functional Films 177
T = 80 K
1.5 c = 10
5
1
Ratio of dark currents
1.0
0.5
0.0
0 2 4 6 8 10
Average electron concentration, 1011 cm−2
Fig. 8.4 Ratio of dark current in a S-QDIP and in a QWIP as a function of average electron sheet
concentration at different ratios of capture probability and T = 80 K
c=5 T = 120 K
1.5
80 K
40 K
Ratio of dark currents
1.0
0.5
0.0
0 2 4 6 8 10
Average electron concentration, 1011 cm−2
Fig. 8.5 The same as in Fig. 8.4 but for different temperatures
In contrast to L-QDIP, S-QDIPs with rather dense QD arrays can exhibit lower
( S − QDIP )
dark current than QWIPs do. Figs. 8.4 and 8.5 show the ratio jth / jth(QWIP ) as
a function of the average electron sheet concentration calculated using Equa-tions
( QW ) ( QD )
(8.6) and (8.10) for different ratios p / p and different temperatures. The
QD density is chosen to be ∑ QD = 5 × 10 cm with the maximum possible number
11 −2
178 M. Willander et al.
Similarly to Equation (8.10), the density of the photocurrent can be related to the
rate of photoemission of a QW (QR or QD array) by the following equation.
eG ph es ∑ I
j ph ≈ = . (8.12)
p p
es (QW ) ∑
R ( QWIP )
= ( QW ) (8.13)
Ù p
where hW is the photon energy. Similar formulae can be used for QRIPs and
QDIPs, if s(QW) and (p)(QW) are replaced by s(QR) and (p)(QR), and s(QD) and (p)(QD). One
can see that the rate of photoemission in all the photodetectors under considera-
tion is a rather weak (linear, near linear if the photoemission cross-section depends
somehow on the electron concentration) function of the average electron concentration.
8 Sensing Infrared and Terahertz Regions by Functional Films 179
It is instructive that both the dark current and the responsivity do not explicitly
depend on the number of QWs (QR and QD arrays) K in the photodetector.
Under normal operation conditions (p) is rather small. This corresponds to the
situation when the number of electrons emitted from one QW (one array) is much
smaller than the number of the electrons injected by the emitter. In such a case, the
current gain (or photoelectric gain) can be large. This quantity is defined as the ratio
of the total flux of the injected electrons jtk/e to the total rate of thermoemission
from all the QWs Htk = KGtk (under dark conditions) or as the ratio of the total flux
of the injected electrons jpk/e to the total rate of photoemission from all the QWs
Hpk = KGpk (under sufficiently strong illumination), hence, g = jtk/eKGtk or g =
jpk/eKGpk. Using Equation (8.5) or (8.12), the current (photoelectric) gain in QWIPs
can be estimated by
1
g (QWIP ) ≈ ( QW ) (8.14)
K p
For QRIPs and QDIPs one obtains similar estimates. Because the electron capture
processes in QDIPs can be attenuated due to such reasons as the phonon bottle-
neck effect, the Pauli principle, and the formation of repulsive potential, the
responsivity of QDIPs can be substantially higher than the responsivity of
QWIPs. Due to the above-mentioned reasons, the QDIP responsivity can be as
large as several A/W (Raghavan et al. 2002). Relatively large values of the
responsivity are also achieved in QDIPs with lateral structure (Miesner et al.
2001; Lee et al. 1999), in which a small capture probability is primarily due to
repulsion of mobile electrons by charged QD arrays. However, a small capture
probability in QDIPs results in not only high values of the responsivity, but in
large dark current as well.
As shown theoretically in Ryzhii (1997), Ryzhii and Liu (1999), and Ryzhii
et al. (2002 a-b), the electron sheet concentration in QWIPs with different mecha-
nisms of the electron injection from the emitter contact into the active region (tun-
neling or thermionic) is determined by the donor sheet concentration (per one QW)
in this region ∑ D , the number of QWs K, and the applied voltage V. Generally, the
electron sheet concentrations in different QWs in the QWIP can be different. In
QWIPs with the tunneling injection, the electron sheet concentrations in QWs adja-
cent to the emitter can be either smaller or larger than in the QWIP active region
bulk. This is confirmed by numerical modeling of QWIPs (Ershov et al. 1995; Sa’ar
et al. 1998). To estimate the average electron sheet concentration ∑ , one can use
the following simple formula,
2C
∑ − ∑D ≈
e
(V − VC ) (8.15)
contact into the QWIP active region is of thermionic origin (Ryzhii et al. 2002 a-b),
VC < 0. Equation (8.15) explicitly indicates that the total charge of the QWIP active
region changes with varying applied voltage. This is because the electric field
induces extra electrons in QWs (Luryi 1985; Ryzhii and Ershov 1995). Similar
calculations (Ryzhii et al. 2000 a-b, 2001 a-c) show that Equation (8.15) can also
be used for QDIPs. Equation (8.15) is in qualitative agreement with experimen-
tal results (Duboz et al. 2003).
Equations (8.6)–(8.9) show that the rate of thermoemission from QWs, QRs, and
QDs exponentially increases with ∑ , which, according to Equation (8.15), is a
function of the applied voltage. As a result, the dark current–voltage characteristics
of QWIPs, QRIPs, and QDIPs are very much in agreement with more detailed
calculations and experimental data. Because the photoemission rate is a much
smoother function of ∑ (what is meant by smoother, lower sensitivity) and,
therefore, V, the responsivity increases with the bias voltage more slowly than the
dark current.
Detectivity
H ph
D* = (8.16)
2 ÙI H th
The QWIP detectivity can be expressed in terms of the average electron concentra-
tion ∑ by
s QW K ∑ ei
D*(QWIP ) α exp( )
2k BT
∑
exp( ) −1
∑ T (8.17)
∑ ei
≈ s QW K ∑ exp( −
2∑ T
) exp(
2k BT
).
The detectivities of L-QRIPs and S-QRIPs are given by the following equations,
respectively.
8 Sensing Infrared and Terahertz Regions by Functional Films 181
ei
s QR K ∑ exp(
2k BT
)
D*( L −QRIP ) α
(a 2 ∑ QR )1/ 4
∑
exp( ) −1
∑ T
a 2 ∑ QR
(8.18)
s QR K ∑ ∑ ei
≈ exp( ) exp( )
(a 2
∑ QR
1/ 4
) 2∑ T a 2
∑ QR
2k BT
2
s QR K ∑ p2 ∑ ei
D*( S −QRIP ) α exp( ) exp( ).
(a 2
∑ QR
1/ 4
) 16∑ T ∑ QR 2k BT (8.19)
s QD K ∑ ∑ ei
≈ exp( ) exp( )
(a 2
∑ QD ) 1/ 4
2∑ T a 2
∑ QD
2k BT
2
s QD K ∑ p2 ∑ ei
D*( S −QDIP ) α exp( ) exp( ). (8.21)
(a 2
∑ QD
1/ 4
) 16∑ T ∑ QD 2k BT
ei
max D*(QWIP ) α 2 K s QW ∑ T exp( − 1) (8.22)
2k BT
ei
max D*( L −QRIP ) α 2 K s QR (a 2 ∑ QR )1/ 4 ∑ T exp( − 1) (8.23)
2k BT
182 M. Willander et al.
8∑ T ∑ QR ei 1
max D*( S −QRIP ) ∝ s QR K exp( − )
p (a 2 ∑ QR )1/ 4 2k BT 2 (8.24)
ei
max D*( L −QDIP ) ∝ 2s QD K ∑ T a 2 ∑ QD exp( − 1).
2k BT
(8.25)
ei
max D*( S −QDIP ) ∝ 2s QD K ∑ T ∑ QD exp( ).
2k BT (8.26)
Relationships (8.22), (8.23), and (8.25) yield the following expressions for the
ratios of the detectivities.
and
T = 80 K (a)
4 S-QDIP
Normalized detectivity, a.u.
0
T = 40 K QWIP (b)
6
L-QDIP
4
0
0 2 4 6 8
Average electron concentration, 1011 cm−2
Fig. 8.6 Normalized detectivities of QWIP, L-QDIP, and S-QDIP versus average electron sheet
concentration at (a) T = 80 K and (b) T = 40 K
same for both QDIPs). Fig. 8.6 shows that even if the QD density is markedly
smaller than the maximum possible density (see below), the detectivity of S-QDIPs
can significantly exceed that of QWIPs (Ryzhii et al. 2004 a-b). The superiority of
S-QDIPs over QWIPs in detectivity can be particularly pronounced at low
temperatures.
Let us estimate max ∑ QD / ∑ T which determines the ratio of the S-QDIP and
QWIP responsivities. The QD density in QDIPs is limited by the requirement of a
weak overlap of the wave functions of neighboring QDs (Liu et al. 2004). Due to
such an overlap, a narrow miniband can arise. Its width in QDIPs is estimated as
1 2 me i
D ≈ e i exp( − ).
2 ∑ QD
(8.31)
Broadening of the ground states in QDs into the miniband does not affect the
activation energy if D << 2k BT . Taking into account this inequality and using
Equation (8.31), we obtain the following condition.
184 M. Willander et al.
2 me i
∑ QD
<
e
. (8.32)
2 ( In( i ))2
2k BT
∑ QD e
< 2p ( i ) 2
1
. (8.33)
∑ T
k BT In (e i / 2k BT )
Using Equation (8.30) and condition (8.33), one can obtain the following
inequality,
s QD
in which the right-hand side can be fairly large at not too small ratio because
s QW
G >> 1 . Indeed, choosing ei = 100 meV and T = 40 – 80 K, one can obtain T ~
390–425. Taking into account relationship (8.34), one may conclude that S-QDIPs
with QD densities close to the maximum value can really exhibit much larger detec-
tivity than QWIPs if the photoemission cross-section sQD for normal IR incidence
and sQW for inclined incidence are close to each other or, at least, the former is
not very small. However, one needs to note that the ratio of the photoionization
cross-sections can strongly depend on the structure of QR and QD arrays (Vasanelli
et al. 2001; Li et al. 2003).
Conclusions
We considered the operation principles of QWIPs, QRIPs, and QDIPs and features
of their characteristics. The comparison of these infrared photodetectors showed:
• QRIPs and QDIPs can exhibit much higher responsivity than QWIPs due to
lower capture probability and, therefore, larger photoelectric gain. Higher
responsivity is inevitably accompanied by higher dark current, because it is
amplified with the same gain.
• QRIPs and QDIPs based on low-density arrays of relatively large QRs and QDs
(L-QRIPs and L-QDIPs, in our terms) should definitely be inferior to QWIPs in
detectivity.
• QRIPs and, particularly, QDIPs based on extremely dense arrays of narrow
QRs and small QDs, in which the bound electrons are really one-dimensional
and zero-dimensional, respectively, can significantly surpass QWIPs in
detectivity.
8 Sensing Infrared and Terahertz Regions by Functional Films 185
The shallow donors and acceptors in semiconductors are the most basic impuri-
ties, which control the electrical properties of materials. Understanding the elec-
tronic structures of such impurities is very important in semiconductor physics
and in applications of semiconductor devices. The impurities in bulk semicon-
ductor materials have been investigated for a long time and their properties are
well established. Due to the recent development of advanced growth techniques
such as molecular beam epitaxy and metal–organic chemical vapor deposition, it
is possible to fabricate ultrasharp interface semiconductor heterostructures, such
as quantum well and superlattice structures. Such low-dimensional semiconductor
186 M. Willander et al.
Confined Acceptors
L>>a0 L~a0
Fig. 8.7 Schematic drawing of a quantum well potential. The circles represent the Bohr radius of
the acceptor wavefunction for two different cases
8 Sensing Infrared and Terahertz Regions by Functional Films 187
confinement potential will compress the acceptor wavefunction along the growth
direction, resulting in the modification of the impurity energy levels and hence the
modification of the Bohr radius of the 1S wavefunction. The detailed theoretical
treatment concerning acceptors is presented later. Here we briefly discuss some
characteristics of acceptors confined in QW structures.
Near the G point in the Brillouin zone, the upper valence band consists of three
p-like states and is consequently sixfold degenerate, if the spin is taken into
account. Due to the spin–orbit interaction, these states are separated into two
groups, which are characterized by the total angular momentum quantum number
J = 1/2 and J = 3/2, with the J = 3/2 band being lowest in energy. The corresponding
acceptor ground state is denoted 1S3/2 (G). This acceptor level is fourfold degenerate
in the bulk case, but splits into two Kramers doublets, with the heavy hole and light
hole characters in the quantum well case due to lower symmetry. The electronic
structure of the acceptor bound exciton (BE) is illustrated in Fig. 8.8. Depending
on whether the electron–hole interaction or the cubic crystal field is the strongest
effect, the acceptor BE states can be denoted according to the j–j coupling theory;
that is, the BE states are j = 1/2, 3/2, and 5/2, or the symmetry from the crystal-field
scheme, that is, G6, G8, and G7 + G8. In the following discussion, we denote the
acceptor BE as j = 1/2 (G6), j = 3/2 (G8), and j = 5/2 (G7,8), respectively. The transi-
tion A1 (B1) of the acceptor BE corresponds to the transitions between the j = 5/2
(G7,8) (j = 3/2 (G8)) BE state and the heavy hole (hh) acceptor state (1S3/2(G6)),
whereas A2 is related to the j = 5/2 (G7,8) BE and the light hole (lh) acceptor state
(1S3/2(G7)).
e-h coupling
h-h (Crystal-field
Acceptor BE QW potential
coupling scheme)
−
J=0,S=1/2 J=1/2 (Γ6)
+
+ − J=2,S=1/2
J=3/2 (Γ8)
J=5/2 (Γ7,8)
A1 A2 B1 B2
1S3/2 (Γ7)
Fig. 8.8 Schematic picture of the acceptor bound exciton (BE) and the acceptor states at different
perturbation conditions. The possible acceptor BE transitions confined in QW structures are also
indicated in the figure
188 M. Willander et al.
As mentioned above, the calculations of the acceptor electronic structures are more
complicated than the donor case, due to the multiple-valence band nature and
strong Coulomb coupling between the different subbands. The QW potential lifts
the degeneracy of the valence band at G8 into two doublets with symmetry G6 and
G7, due to lower symmetry from Td in bulk material until D2v in symmetric QW
structures. When acceptors are introduced at the center regions of the well layer, the
symmetry is unchanged, so the ground states of the acceptors also split into two
doublets. However, if doped acceptors are located away from the center region, the
symmetry from the D2v in the center doping case reduces to C2v symmetry.
The theoretical calculation of acceptors confined in QW structures are mainly
based on either the effective mass theory (Masselink et al. 1983, 1985; Pasquarello
et al. 1989; Loehr et al. 1990; Fraizzoli and Pasquarello 1990, 1991) or the tight-
bindinglike model (Einevoll and Chang 1990) in strain-free QW systems such as
GaAs/AlGaAs structures. The calculations (Masselink et al. 1983, 1985; Pasquarello
et al. 1989; Einevoll and Chang 1990) provide the energies of the acceptor ground
and the excited states. It is found that the binding energy of the acceptor ground
state G6 depends on the well width and the location of acceptors in the well layer.
At a later stage, the effective mass calculations of acceptor energies, based on the
work by Pasquarello et al. (1989), were expanded to include magnetic field effects
and stress effects or built-in strain in a lattice-mismatched QW system by QX (Zhao
et al. 1994a, b1998; Zhao and Willander, , 1999, 2000; Zhao et al. 2001). Based on
these calculations, the detailed energy levels of acceptors and oscillator strengths of
the transitions between the acceptor ground and different excited states in QW
structures in the presence of magnetic field or stress perturbations were obtained
and can be compared with experimental data.
The acceptor Hamiltonian expressed in electron energy is given by the following
operator,
(
H = − H kin + H c + H QW ) (8.35)
⎡P + Q L M 0 ⎤
⎢ L+ P −Q 0 M ⎥⎥
H kin =⎢ + (8.36)
⎢ M 0 P −Q −L ⎥
⎢ + + ⎥
⎣ 0 M −L P + Q⎦
where
g 12 2
P= k
2 m0
g 2 2 2
Q= (k x + k y2 − 2 kz2 )
2 m0
8 Sensing Infrared and Terahertz Regions by Functional Films 189
g 3 2
L = −i 3 (k x − ik y )kz
2 m0
2
M= 3 (g 2 + g 3 )(k x − ik y )2 .
4 m0
The Hc and HQW are the Coulomb potentials due to the acceptors and the quantum
well potential, respectively. The QW potential, HQW, must properly include the
effects of the deformation potential and bandgap offset between well and barrier
materials. That is, HQW will contain a square-well potential, DEQWhh,lh, for the hh and
the lh, and includes a potential difference (Vp) between the hh and lh band edges in
the well due to built-in strain, thus:
H QW = DEhh
QW
,lh + Vp .
Once all effects are properly included in the acceptor Hamiltonian, the Hamiltonian
given in Equation (8.35) acts on a four-component function:
The energy levels of the shallow acceptor states and corresponding wave functions
are derived. The s component of an acceptor envelope function of definite angular
momentum m can be expanded into a set of basis functions, separable in the coor-
dinates r and z:
By using these wavefunctions, most of the integrals that appear in the matrix ele-
ments of the Hamiltonian can be computed analytically. Particularly, in the calcula-
tion of the matrix elements of the Coulomb potential Hc, an auxiliary integral that
decouples r and z coordinates is introduced using the well-known transformation:
1
= ∫ e − z q J 0 ( rq )dq.
(r − z )
2 2
25.00
2S-1S
20.00 2P-1S
Energy (meV)
15.00
Δ (lh-hh)
10.00 Exciton
5.00
2P-2S
Δ(1S)
0.00
−0.002 0.000 0.002 0.004 0.006 0.008 0.010
In concentration
Fig. 8.9 The acceptor 1S-2S and 1S-2P transitions versus In-concentration in Be-acceptor doped
InxGa1-xAs/Al0.3Ga0.7As QW structures. Dots are experimental data and lines are the theoretical
calculated results according to the model discussed in text. ΔExciton(lh–hh) represents the energy
separation between the light and heavy hole-free exciton transition. Δ(1S) is the energy separation
between the acceptor 1S3/2(G6) and 1S3/2(G7) states
8 Sensing Infrared and Terahertz Regions by Functional Films 191
calculated results. We would like to point out that according to the theoretical
calculations, the change of the acceptor electronic structures, in the range of
In-concentrations used here, is mainly due to the variation of the biaxial potential.
Based on the previous examples, we can confidently conclude that the used theo-
retical model is very reliable in calculating the energy levels of acceptors confined
in QW structures. Concerning the possible stimulated far-infrared emissions, the
research has touched on many different ways (Brazis 1995; Andronov 1987; Faist
et al. 1994; Sirmain et al. 1997). Particularly, a far-infrared stimulated emission
from uniaxially stressed p-Ge (Odnoblyudov et al. 1998, 1999; Altukhov et al.
1992) reported recently has become very interesting. On the other hand, the pos-
sible THz emission can also be used for the reversed process, that is, THz detection.
According to estimated critical layer thickness (Matthews and Blakeslee 1974;
Tsao et al. 1987; People and Bean 1985; Ekenstedt et al. 1993; Wang et al. 1991;
Fritz 1987), the possible THz transitions have been calculated in the strained QW
systems such as InGaAs/GaAs and SiGe/Si QW structures (Zhao and Willander
2000). Therefore, the THz sensing devices based on impurity states in QW struc-
tures can be designed according to the above theoretical calculations.
W
+ + + + + + + + + +
– – – – – – – – – –
2DEG
channel
Lc Lg Lc
ΔLg
Ld
2DEG
channel Gate
Gate
Fig. 8.10 Schematic view of HEMT structure (a) at moderate drain voltage and (b) in saturation
regime (high drain voltage); (c) band diagram of HEMT doped gate layer; and (d) band diagram
of HEMT with a resonant-tunneling structure between the channel and the gate instead of the gate
layer. Arrows show electron tunneling from the channel into the gate
∂∑ ∂∑ J
+ = . (8.38)
∂t ∂x e
Here, ∑ ( x, t ) are the hydrodynamic electron velocity and the electron sheet
concentration, respectively, y(z,x,t) is the self-consistent electric potential in the
area surrounding the 2DEG channel, u is the collision frequency of electrons with
impurities and phonons (not the electron–electron collision frequency), e and m are
the absolute value of electron charge and the electron effective mass, J = J(x,t) is
the leakage electron current from the 2DEG channel or the electron current injected
into the channel, the x-axis is directed in the 2DEG channel plane, and the z-axis is
directed perpendicular to this plane. Equation (8.37) can be generalized by intro-
ducing the terms associated with the electron pressure and the 2DEG viscosity.
However, these terms are not particularly essential (Dyakonov and Shur 1996;
Rudin and Samsonidze 1998; Rudin et al. 1999). Equations (8.37) and (8.38)
should be supplemented by the Poisson equation for the self-consistent electric
potential (Dyakonov and Shur 1996; Shur and Ryzhii 2003; Nakayama 1974):
∂ 2y ∂ 2y 4pe
+ = (∑ d − ∑ ) · d ( z ), (8.39)
∂z 2 ∂x 2 k
194 M. Willander et al.
where k is the dielectric constant, ∑ is the donor sheet concentration, and d(Z)
d
is the Dirac delta function. In the case of a 2DEG channel with a highly con-
ducting plane (gate) placed parallel to the channel at a sufficiently short dis-
tance W (in HEMT-like structures, W is the thickness of the gate structure
between the channel and the gate contact), the above 2D Poisson equation can
be replaced by its simplified consequence (the so-called gradual channel
approximation; Shur (1990)):
j 4pe
= (∑ − ∑ d ). (8.40)
W k
Here, f=f(x,t) is the electric potential in the 2DEG channel; that is, j = j z = 0 .
Equations (8.37)–(8.40) are widely used for theoretical analysis of plasma effects
in 2DEG systems.
To study plasma wave propagation and the response of the 2DEG system to
external perturbations (incoming signals) a small-signal analysis based on the
linearized versions of Equations (8.37)–(8.40) is used. In this approach, the ac
components of the electron hydrodynamic velocity uw ( x ) exp( −iwt ), the electron
sheet concentration, and the self-consistent potential are presented as
uw = uw ( x ) exp( −iwt ), ∑ w = ∑ w ( x ) exp( −iwt ), and jw = jw ( x ) exp( −iwt ), respec-
tively, where w is the signal frequency. In particular, these linearized equations
obtained for a uniform 2DEG channel with the dc electron sheet concentration
∑ 0 result in the following dispersion equations for the plasma waves with
the frequency w and the wavenumber q propagating in the x-direction (so that
jw α exp(i(qx − wt )) ),
2pe2 ∑ 0
w (w + in ) = q (8.41)
km
in the ungated 2DEG channels, and
4pe2 ∑ 0 W
w (w + in ) = q2 (8.42)
km
in the gated 2DEG channels with W < < Lg (Lg is the length of the gated section, i.e.,
the gate length). Equations (8.41) and (8.42) applied to 2DEG channels with
relatively small collision frequency ( n << w ) result in Rew α ∑ 0 q (Stern 1967)
and Rew α ∑ 0 Wq (Chaplik 1972; Nakayama 1974), respectively. The spec-
tra of plasma waves in both ungated and gated 2DEG channels are strongly different
from the spectrum of plasma waves in 3DEG, where Re w is virtually independent
of q. The damping of plasma waves in channels of both types is determined mainly
by the electron collision frequency: w ≈ n / 2 . The electron leakage from (injection
into) the 2DEG channel due to tunneling or resonant tunneling (RT) through the
gate layer or more complex structure (see Fig. 8.12) can lead to an additional damp-
ing or, vice versa, to its suppression. In some cases, the latter processes can result
in a negative damping, that is, in plasma wave instability.
8 Sensing Infrared and Terahertz Regions by Functional Films 195
The 2DEG channels usually have limited length and are supplied with highly
conducting contacts (e.g., the source and drain contacts in the case of a field-effect
transistor). The edges of the 2DEG channel result in the reflection of the propagating
plasma waves and the formation of the standing plasma waves (plasma oscillations).
Due to the plasma wave the spectrum is quantized. If the potential at the highly
conducting contacts separated by the spacing L is fixed (so that, e.g.,
jw x = 0 = jw x = L = 0 ), a simplified quantization rule yields q = qn = pn / L , where n
= 1, 2, 3, … is the index of the oscillations mode. In particular, Equation (8.42) for
a HEMT structure with the gate length Lg » L (i.e., Lc< < Lg) yields w » nWg,
where:
4pe2 ∑ 0 W
Wg =
kmL2g (8.43)
channels depends on the gate voltage Vg, the plasma frequencies in such channels
can be tuned by this voltage. Fig. 8.11 shows the variation of the fundamental plasma
frequency Wg / 2p and the quality factor Q = 4Wg / pn with changing gate voltage
Vg calculated for HEMT structures with relatively short ungated gate-source and
drain-gate regions (Lc< < Lg).
196 M. Willander et al.
1.25 120
ν = 2x1011s−1 Lg = 0.5 μm
Plasma frequency, THz 1.00 = 0.75 μm
= 1.0 μm 90
Quality factor
0.75
60
0.50
30
0.25
0.00 0
0 0.2 0.4 0.6 0.8 1
Gate voltage swing, V
Fig. 8.11 Fundamental plasma frequency and quality factor versus applied gate voltage calcu-
lated for HEMT structures with different gate length (Lg> Lc)
Plasma effects in HEMT structures such as that shown in Fig. 8.10 can be used for
resonant detection of THz radiation. As proposed previously (Dyakonov and Shur
1996), the excitation of plasma oscillations by incoming THz radiation in the
HEMT 2DEG channel results in variations of the dc current in the channel or of the
dc voltage between the source and the drain. This is due to a rectification effect
associated with hydrodynamic nonlinearity of Equations (8.1) and (8.2) governing
the plasma oscillations. The observation of the resonant detection of THz radiation
in HEMTs was reported in several publications (Lu et al. 1998; Lu and Shur 2001;
Deng et al. 2002; Teppe et al. 2005). THz detectors based on HEMT-like structures
utilizing the excitation of plasma oscillations associated with different mechanisms
of nonlinearity were proposed in Satou et al. (2003), Khmyrova and Ryzhii (2000),
and Ryzhii et al. (2000 a-b).
In these devices, the rectified component of the dc current through the gate
structure with a nonlinear dependence of the leakage channel-gate current is used
as the measurand. The described nonlinearity can be associated with strong
8 Sensing Infrared and Terahertz Regions by Functional Films 197
Rw 1
≈ (8.44)
R0 ⎡⎣sinh (pn / 4Wg ) + cos2 (pw / 2Wg )⎤⎦
2
+ + + + + + + + + + + + +
− − − − − − − − − − − − −
+ + + + + + + + + + + + +
− − − − − − − − − − − − −
L l
(c)
eVbi
eV0
Fig. 8.12 Structures of THz detectors based on lateral Schottky junction with (a) an ungated, (b)
gated channel, and (c) device band diagram under forward bias (V0 < 0)
198 M. Willander et al.
μ = 30x104cm2 /Vs
= 12x104cm2 /Vs
1000 = 6x104cm2 /Vs
10
1
0.5 1 1.5 2 2.5
Frequency, THz
Fig. 8.13 Normalized responsivity as function of THz radiation frequency calculated for detec-
tors based on lateral Schottky junction with different electron mobility in the 2DEG channel
The same ratio but for a resonant detector with an ungated channel (see Fig.
8.12a) is given
Rw 1
≈ (8.45)
R0 ⎡⎣sinh (pnw / 4Wu ) + cos2 (pw 2 / 2Wu2 )⎤⎦
2 2
p 2 e2 ∑ 0
Wu = (8.46)
kmL
Comments
The concepts discussed above can result in the development of novel THz heter-
ostructure devices such as detectors and frequency multipliers of THz radiation.
However, the device proposals considered do not exhaust all interesting new ideas;
8 Sensing Infrared and Terahertz Regions by Functional Films 199
see, for example, Govorov et al. (1998, 1999), Ryzhii et al. (2002 a-b), and
Hanabe et al. (2005). The specific features of plasma waves in the 2DEG channel
can be used not only in the devices like those discussed above. The linearity of
plasma wave dispersion ( Re w α q ), relatively large plasma wave phase and group
velocity (S » 108cm/s), and the possibility to control the plasma wave propagation
and the interaction plasma waves between each other by locally applied voltage
open up wide prospects to create new systems for processing of THz signals (in
particular, delay lines, interferometers, etc.).
Cosmology experiments in the last few years (BOOMERanG, WMAP) have dis-
covered that the universe consists of 73% dark energy, 23% dark matter, and only
4% ordinary matter. The most shocking news is the acceleration of the universe by
unknown forces (Breakthrough of the Year: Illuminating the Dark Universe 2003).
Experiments to resolve the nature of these mysterious dark components will require
a new generation of ultrasensitive detectors to get a more detailed picture of the
cosmic microwave background radiation (Breakthrough of the Year: Illuminating
the Dark Universe 2003).
A new principle to realize an ultrasensitive THz detector was proposed by Kuzmin
et al. (Kuzmin 2000; Kuzmin et al. 1998; Kuzmin and Golubev 2002). A novel con-
cept of the cold-electron bolometer (CEB) is based on strong direct electron cooling
of the absorber removing all incoming power from the supersensitive absorber to the
readout system with considerably higher dynamic range. This concept is purposed to
overcome the main contradiction of supersensitive detectors: overheating by back-
ground power load due to high sensitivity of the detector. Moreover, additional artifi-
cial dc heating of the TES (transition-edge sensor) for electrothermal feedback will
be replaced by effective electron cooling (see Fig. 8.14). This could entail a
0 0
100 mk 230 mk Te 100 mk 230 mk Te
Tph Tph
Fig. 8.14 Comparison of the CEB and TES concepts (Kuzmin 2004)
200 M. Willander et al.
Fig. 8.15 Capacitively coupled CEB with SIN tunnel junctions for temperature measurements
and electron cooling (strong electrothermal feedback) [Kuzmin 2004]. An AFM picture of the
right part of the bolometer shows SIN junction and a part of the large antenna
Fig. 8.16 Record electron cooling achieved at Chalmers University in real bolometer configuration
due to improved quasiparticle trapping: (a) Au trap just near the junctions; (b) improved shape of
superconducting electrode; (c) usual cross geometry [Kuzmin et al. 2004, Agulo et al. 2004]
conductor (Masi et al. 2004; Kuzmin and Mauskopf 2005). The achieved cooling
results give a good basis for realization of high-performance CEBs working at real
conditions of background power load.
Model: Here we assume that the SIN tunnel junctions are voltage-biased, and the
current is measured by SQUID. The sensitivity of the device is then characterized
by the current responsivity SI, which is the ratio of the current change detected by
the SQUID and the change in the power load of the bolometer caused by a detected
signal:
∂I ∂I∂T
SI = w = .
∂Pw ∂P (8.47)
−i w cv L + 5SLTe +4
∂T
Here cn = gTe is the specific heat capacity of the normal metal; 5 S L Te4 is the
thermal conductance between electron to the phonon subsystems in the normal
metal, ∑ is a material constant, ∧ is a volume of the absorber, Te and Tph are the
electron and phonon temperatures of the absorber; ∂P / ∂T is the thermal conduct-
ance of the NIS junction, and P(t) is the incoming radio frequency power.
The noise is captured by the noise equivalent power, which is the sum of three
different contributions, and is defined as
dI 2
2
NEPtotal = NEPe2− ph + NEPSIN
2
+ . (8.48)
SI2
Here,
NEPe2− ph = 10 kB SL (Te6 + Tph6 ) (8.49)
−18
8x10
NEP
Goal: NEP=10−18 W
6 Po= 0, R=6 k Ω
10−13 W, 6 k Ω
NEP (W/Hz1/2 )
10−13 W, 0,5 k Ω
4 Po- microwave background load
1/2
2 δ ISQUID= 50 fA/Hz
0
0.75 0.80 0.85 0.90 0.95 1.00 1.05
V/Δ
Fig. 8.17 NEP in presence of various background power loads and various efficiencies of direct
electronic cooling for bath temperature 100 mK (Kuzmin and Golubev 2002)
It is necessary to take into account the effect of the electron cooling of the metallic
strip by the NIS tunnel junctions.
Effect of background power load: Our analysis of the effect of background
power load on noise performance for different configurations of CEB bolometers
shows that the optimal configuration of the bolometer is a CEB with voltage-biased
SIN tunnel junctions and a SQUID readout (Kuzmin et al. 2004). The volume of
the absorber is equal to 0.05 mm3, which is typical for our experiments.
The current noise of SQUID is equal to 50 fA/Hz1/2 in our simulations. The
results are shown in Fig. 8.17 for two levels of microwave background power: P0=
0 and 0.1 pW. The latter figure is a realistic background power load P0 for a band-
width of 10% at frequencies in the range of 300–1000 GHz for background tem-
perature Tbg= 3 K. The first curve without background load (P0= 0) produces NEP
= 2 × 10−19 W for typical junction resistance (R) equal to 6 kW.
A considerable increase of the NEP to 8 × 10−18 W/Hz1/2 is obtained for P0= 0.1
pW. The electron temperature also increases from 100 to 230 mK. Decreasing R to
0.5 kW improves the efficiency of the electronic cooling and returns the NEP to the
acceptable level of 8 × 10−19 W/Hz1/2 and Te to the level of 100 mK. The NEP goal
for the future projects is 10−18 W/Hz1/2 (Kuzmin 2000; Kuzmin et al. 1998; Kuzmin
and Golubev 2002) and can be achieved with these system parameters.
Concept of an optimal bolometer: We have analyzed the optimal CEB in the
presence of the final background power load (P0 = 0.1 pW) for fixed parameters of
the SQUID-amplifier (10 fA/Hz1/2) at T = 300 mK (Agulo et al. 2004). The optimal
204 M. Willander et al.
regime can be realized when thermal “cooling conductance” through the tunnel
junctions dominates the “fundamental” electron–phonon conductance. In these
circumstances, an NEP level of 10−18 W/Hz1/2 at 300 mK can be achieved.
The dependences of the NEP on a volume of the absorber show that there is no
optimal value of NEP for the volume of absorber L. The reason for the flattening
for small volumes is that we have achieved full transference of P0 to the amplifier,
so that the NEPe-ph constitutes less than 50% of the total NEP. The critical point of
the optimal regime is a point of equality of NEPSIN and NEPe-ph at L = 0.003 mm3.
The dependences of the NEP on the resistance of the SIN tunnel junctions R gives
the optimum value R around 1.5 kW. For higher values of R, the electron cooling is
not as effective and responsivity is decreased, increasing noise of the SIN junction
(4) and SQUID (2). Decrease of R lower than the optimal point increases the shot
noise (reverse proportional to R) without any increase in responsivity because of
saturation in transferring power.
Expected results: Improvement of noise equivalent power of the bolometer
receiver due to realization of the optimal concept to the level of NEP ~ 10−18
W/Hz1/2 at 300 mK is expected.
This question has arisen in relation to the highest requirements on NEP for future
NASA missions. The question is how realistic are these requirements on NEP =
10−20 W/Hz1/2.
The ultimate performance of CEB and other concepts has been analyzed
(Kuzmin 2004). The NEP is determined by shot noise due to power load. Other
sources of noise are neglected due to small values. For the level of P0 = 10 fW this
limit can be achieved using relatively low temperatures (~100 mK) and small volume
of the absorber (∧ ≤ 0.002mm3) when we can neglect the electron–phonon noise
component.
A general ultimate NEP formula has been derived (Kuzmin 2004):
The lowest NEP can be achieved for CEB with the lowest level of quantization.
However, even these extreme parameters of P0 and Equant show that it is rather unre-
alistic to achieve NEP = 10−20 W/Hz1/2 as announced in NASA requirements for
future missions.
This formula (8.51) is used for estimation of ultimate parameters of CEB and
other bolometers for given parameters of detector systems. The work on analysis of
ultimate parameters of CEB will be prolonged.
Preliminary results: Record attowatt sensitivity of the cold-electron bolometer.
The Chalmers group has made the measurements of the cold-electron bolometer
in a current-biased mode. They have measured the voltage response to applied
low-frequency modulation of the heating current on the normal metal absorber.
The detector responsivity was 1.5 × 1010 V/W at 35 Hz at 100 mK. The frequency
dependence of responsivity was tested and was found to decrease with increasing
frequency of the modulated signal owing to the inclusion of the signal attenuation
due to cryogenic filters. The corresponding noise equivalent power of the bolo-
meter was obtained, using an operational amplifier at 300 K from the noise of the
bolometer divided by the detector responsivity (Fig. 8.18).
The record noise equivalent power for the CEB was found to be better than 10−18
W/Hz1/2 at 100 mK and at frequencies higher than 100 Hz for a background power
load of 2.6 fW. The next step is measurement of NEP in voltage-biased mode.
For the next breakthrough in our knowledge about dark matter and dark energy,
we need a new generation of detectors. Priority of this topic can be determined as
96% (dark universe) to 4% (ordinary matter).
A cold-electron bolometer is a good candidate to become the leading concept in
this development.
6
NEP Total
5 NEP Bolo
NEP Amp
NEP (10−18 w/Hz1/2)
100 1000
frequency (Hz)
Fig. 8.18 The dependence of the total, bolometer, and amplifier NEP of the cold-electron bolom-
eter to the modulation frequency. We have stepped to the 19th power of CEB sensitivity for fre-
quencies higher than 100 Hz. The dashed line represents the value of the NEPs as predicted by the
nonequilibrium theory of the CEB
206 M. Willander et al.
Realization of the cold-electron bolometer would be the turning point from arti-
ficial heating (TES) to effective electron cooling lower than phonon temperature
which should bring clear benefits for supersensitive detectors.
The famous contradiction between supersensitivity and saturation can be over-
come by strong electrothermal feedback removing power to the next stage with
higher dynamic range.
The CEB concept could be implemented for new balloon telescopes OLIMPO,
CLOVER, and PILOT.
Summary
We have discussed sensing of two parts of the EM spectra, namely the IR region
and the THz region. For the IR region we have particularly analyzed QWIP, QRIP,
and QDIP and found that QRIP and QDIP should be superior to QWIP for IR detection.
For the THz region we first analyzed how influence of strain on impurities can be
used for THz detection (and generation). Then we analyzed in detail plasma effects
in two-dimensional electron systems for sensing (and generation). Finally we gave
an example of an ultrasensitive THz detector for space application which has shown
excellent experimental figures of merit.
References
Masi S, de Bernardis P, et al. (2006) In: Background Microwave Radiation and Intracluster
Cosmology eds, Melchiorri F and Rephaeli Y (New IOS press publication)
Masselink WT, Change Y-C, Morkoc H (1983) Phys. Rev. B, 28:7373
Masselink WT, Change Y-C, Morkoc H (1985) Phys. Rev. B, 32:5190
Matov OR, Meskov OF, Popov VV (1998) J. Exp. Theor. Phys., 86:538–544
Matthews JW, Blakeslee AE (1974) J. Crystal Growth, 27:118
Miesner C, Brunner K, Abstreiter G (2001) Phys. Stat. Sol. (a), 224:605–607
Nakayama M (1974) J. Phys. Soc. Jpn., 36:393–398
Odnoblyudov MA, Chistyakov VM, Yassievich IN, Kagan MS (1998) Phys. Stat. Sol. (b),
210:873
Odnoblyudov MA, Yassievich IN, Kagan MS, Galperin YUM, Chao KA (1999) Phys. Rev. Lett.,
83:644
Pan D, Towe E, Kennerly S (1998) Appl. Phys. Lett., 73:1937–1939
Pan D, Towe E, Kennerly S (1999) Appl. Phys. Lett., 75:2719–2721
Pasquarello A, Andreani LC, Buczko R (1989) Phys. Rev. B, 40:5602
People R, Bean JC (1985) Phys. Appl. Lett., 47:322
Phillips J, Kamath K, Bhattacharya B (1998) Appl. Phys. Lett., 72:2020–2022
Phillips J, Bhattacharya P, Kennerly SW, Beekman DW, Dutta M (1999) IEEE J. Quant. Electron.,
35:936–940
Raghavan S, Rotella P, Stinz A, Fuchs B, Krishna S, Morath C, Cardimona DA, Kennerly SW
(2002) Appl. Phys. Lett., 81:1369–1371
Rappaport N, Finkman E, Brunhes T, Boucaud P, Sauvage S, Yam N, Thanh V Le, Boucier D
(2000) Appl. Phys. Lett., 77:3224–3226
Rokhinson LP,Chen CJ, Tsui DC, Vawter GA, Choi KK (1999) Appl. Phys. Lett., 74:759–761
Rosencher E, Vinter B, Luc F, Thibaudeau L, Bois P, Nagle J (1994) IEEE Trans. Quant. Electron.,
30:2875–2882
Rudin S, Samsonidze G (1998) Phys. Rev. B, 58:16369–16373
Rudin S, Samsonidze G, Crowne F (1999) J. Appl. Phys., 86:2083–2088
Ryzhii M, Ryzhii V (1999) Jpn. J. Appl. Phys., 38:5922–5927
Ryzhii M, Ryzhii V, Willander M (1999) Jpn. J. Appl. Phys., 38:6650–6653
Ryzhii V (1996) Semicond. Sci. Technol., 11:759–765
Ryzhii V (1997) Appl. Phys. Lett., 70:2532–2534 and J. Appl. Phys., 81:6442–6448
Ryzhii V (1998) Jpn. J. Appl. Phys., 37:5937–5944
Ryzhii V (2001) Jpn. J. Appl. Phys., 40:L148–L150 and Appl. Phys. Lett., 78:3346–3348
Ryzhii V (ed.) (2003) Intersubband Infrared Photodetectors. Singapore: World Scientific
Ryzhii V, Ershov M (1995) Semicond. Sci. Technol., 10:687–690
Ryzhii VI, Fedirko VA (1983) Sov. Phys. Semicond. 17:850–851
Ryzhii V, Liu HC (1999) Jpn. J. Appl. Phys., 38:5815–5822
Ryzhii VI, Bannov NA, Fedirko VA (1984) Sov. Phys. Semicond., 18:481–493
Ryzhii V, Khmyrova I, Ryzhii M, Ershov M (1996) J. Phys. IV, 6:C3-157–159
Ryzhii V, Khmyrova I, Shur M (2000a) J. Appl. Phys., 88:2868–2871
Ryzhii V, Khmyrova I, Mitin V, Stroscio M, Willander M (2001a) Appl. Phys. Lett.,
78:3523–3525
Ryzhii V, Khmyrova I, Pipa V, Mitin V, Willander M (2001b) Semicond. Sci. Technol.,
16:331–338
Ryzhii V, Khmyrova I, Ryzhii M, Mitin V (2004a) Semicond. Sci Technol., 19:8–16
Ryzhii V, Khmyrova I, Ryzhii M, Pipa V, Mitin V, Willander M (2001c ) Proc. SPIE,
4288:396–406
Ryzhii V, Khmyrova I, Shur M (2002a), J. Appl. Phys., 91:1875–1881
Ryzhii V, Pipa V, Khmyrova I, Mitin V, Willander M (2000b) Jpn. J. Appl. Phys.,
39:L1283–L1285
Ryzhii V, Ryzhii M, Liu HC (2002b) J. Appl. Phys., 92:207–213
Ryzhii V, Satou A, Shur MS (2003) J. Appl. Phys., 93:10041–10045
Ryzhii V, Satou A, Shur MS (2005) Phys. Stat. Sol. (a), 202:R113–R115
8 Sensing Infrared and Terahertz Regions by Functional Films 209
A D
Absorption Deposition 42, 45, 53, 54, 55, 58, 59, 60, 61,
Optical 14, 15 62, 63, 70, 76, 102, 110, 127, 128, 132
Actuator 2 Glancing-angle 32, 36, 48, 63, 64, 77, 78
Accuracy 2, 5, 8, 23 Oblique angle 46, 47, 48, 50, 63
Aging 3, 8 Material 2
Architecture Process 21, 31
Sensor 3, 4 Pulsed-laser 36
Attributes Vapor 31, 32, 33, 35, 36, 37, 45, 77,
Performance 4, 5 78, 149, 152, 187
Electrochemical 32, 41, 63, 69, 77, 78
Thin film 32, 38, 46, 70, 76, 93, 100,
B 102, 112, 115
Behavior Nanostructure 77, 78
plastic 12, 13 Sol-gel 151, 152
elastic 12 Detector
Photo 94, 133, 170, 171, 172, 173, 174,
181, 183, 186, 208, 209, 210
C Dielectric 10, 15, 16
Calibration 3, 6, 8, 9, 99, 106, 107, 120, Diffusion 7, 8, 21, 100, 127, 140, 153
121, 122, 126, 133, 164 Drift 5, 8, 28, 94, 101, 130, 132, 138
Ceramic 11, 14, 39, 70, 81, 149
Conductivity
Thermal 11, 18, 93 E
Electrical 18, 82, 111 Electronics 3, 5, 7, 9, 29, 80, 81, 82, 84, 96,
Ionic 114, 118, 120, 121, 168 97, 100, 109, 134, 147, 161, 178
Photo 190, 195 Energy 3, 6, 10, 11, 12, 13, 14, 15, 17, 18,
Confinement 21, 27, 28, 36, 39, 68, 70, 71, 82, 88,
Geometrical 3, 11, 12, 14, 16 89, 90, 96, 101, 102, 106, 107, 111,
Quantum 169, 186, 187 112, 113, 114, 123, 124, 129, 137, 170,
Curve 171, 173, 174, 176, 178, 185, 188, 189,
Calibration 3, 120, 121, 122, 133 190, 191, 192, 193, 201, 206, 207
Response 5, 23, 95, 96, 97, 122, 151 Surface 11, 13
211
212 Index
Etching 2, 31, 32, 41, 42, 45, 70, 71, 72, 76, MEMS 2, 16, 19, 21, 22, 24, 29, 68, 69,
78, 80, 83, 84, 165 74, 75, 76, 79, 81, 82, 84, 87, 88,
Error 106, 129, 148, 149, 151, 152, 153,
Systematic 8 161, 167
Random 8 Metal 11, 12, 14, 15, 16, 21, 22,
36, 39, 44
Miniaturization 2, 106, 137, 138, 140, 142,
F 161, 167
Fabrication 1, 2, 9, 15, 20, 24, 31, 32, 33, 34, Modulus
36, 38, 39, 44, 51, 55, 58, 59, 62, 63, Elastic/Young’s 11, 13, 16, 140, 151, 156,
64, 65, 66, 69, 70, 71, 73, 74, 75, 77, 159
79, 80, 81, 82, 83, 84, 85, 87, 108, 114, Model 9, 12, 13, 14, 15, 16, 19, 39, 63, 132,
135, 137, 142, 144, 147, 151, 161, 162, 137, 138, 142, 144, 145, 176, 181, 190,
164, 165, 167, 171 191, 192, 193, 195, 204
Force Multilayers 2, 98
Dispersion 11, 123
London 11
Van Der Waals 11 N
Frequency Nano-
Resonance 2, 6 Structure 1, 2, 3, 7, 12, 13, 15, 16, 17, 27,
31, 32, 37, 38, 41, 44, 49, 50, 51, 52,
53, 55, 59, 60, 62, 63, 64, 72, 73, 74,
G 76, 77, 78, 79, 81, 83, 87, 129, 134,
Grains 12 137, 169, 170, 187
Technology 16, 43, 63, 64, 81, 82,
83, 85, 208
H Scale 11, 12, 13, 14, 15, 60, 61, 62, 63,
Hysteresis 5, 8, 127, 128 80, 81, 82, 85, 137, 140, 144
Homogeneity 13, 14 Materials 11, 12, 13, 14, 63, 81, 114, 129,
134
Morphology 7
I Fluid 13, 14, 16
Index Instrument 10
Refractive 11, 16, 67, 99, 133, 134, Particle 11, 15, 16, 27, 31, 32, 41, 44,
137, 150 45, 62, 76, 77, 78, 80, 81, 107, 134,
Inhomogeneity 13, 14 135, 139, 140, 144
Interactions Regime 11, 13
Ionic 11 Noise 3, 6, 7, 8, 10, 18, 24, 94, 95, 96, 110,
Interface 12, 61, 95, 133, 134, 137, 140, 111, 119, 129, 135, 137, 138, 150, 161,
172, 187 187, 202, 204, 205, 206, 207
L P
Liquid 11, 12, 13, 14, 28, 32, 34, 38, 39, Point
41, 44, 67, 75, 89, 95, 105, 109, 110, Melting 11, 12, 16
116, 120, 122, 123, 139, 167 Polymer 14, 16, 21, 22, 23, 28, 31, 39, 40,
41, 42, 43, 62, 63, 66, 70, 72, 73, 76,
77, 78, 79, 80, 82, 83, 107, 108, 111,
M 112, 114, 116, 119, 123, 124, 125,
Materials 126, 127, 129, 130, 131, 132, 134,
Bulk 11, 15 135, 138, 139, 140, 141, 142, 143,
Selection 1, 9, 16, 147 144, 145, 150, 153, 154
Measurand 3, 4, 5, 6, 7, 8, 9, 10, 17, 18, 19, Precision 2, 5, 8, 24, 29, 100, 102,
21, 22, 23, 24, 25, 26, 27, 28, 74, 198 137, 161
Index 213