Cantilever Array for Chemical Sensing: Development of a Moisture-Independent Sensor with Possible Application in Explosive Detection

1 – Introduction
Self-Assembled Monolayers (SAM), physically and chemically induced nanoscale motion
can be detected with high sensitivity and specificity.
This thesis is focused on cantilevers functionalized with self-assembled monolayers and
used for gas sensing. As regards the cantilever sensing modes, only experiments concerning
the static bending due to surface stress change are reported.
If we consider the title of the thesis, Cantilever Array for Chemical Sensing: develop-
ment of a moisture-independent sensor with possible application in explosive detection, it
has been explained, at this point, the role played by cantilevers as sensing platform. We
can deepen now the concept of chemical sensing and in particular gas sensing.
Chemical sensing spans a very board area and it includes detecting the presence or
absence of any one of the myriad chemical species. In the current work, chemical sensing
is used in the context of gas detection. Detection of gases and other volatile compounds
finds important applications in the areas of environment and air quality monitoring, food
processing industries, healthcare and in threat evaluation [4].
The last application comprises military and homeland security, which includes the de-
tection of explosive materials, biowarfare agents, and chemical warfare agents. Recent
events have made bio-chemical warfare agents detection technology become more impor-
tant than ever before to homeland security. It is reported here an effective sentence, found
in an IEEE paper [5], which help us in understanding why this field of research is driven
by considerable public money and private funding in the United States.
Terrorists have a huge economic advantage over law enforcement because it is,
many times, more expensive to detect terrorist threats than it is to deploy ter-
rorist threats. For example, a crude explosive device can destroy an airplane in
flight. On the other hand, even though sensitive detection of individual threats
may be currently possible, such techniques/sensor systems are bulky, expen-
sive, and require time-consuming procedures and constant operator attention.
[...]
Therefore an urgent need exists for the development of inexpensive, highly
selective, and extremely sensitive sensors to help combat terrorism. If such
sensors can be made miniature, they could be deployed in virtually any situ-
ation. Silicon-based microelectromechanical sensors represent an ideal sensor
platform for combating terrorism because these miniature sensors are inexpen-
sive and can be deployed almost anywhere.
From this quote, it is clear that sensor portability is an important factor which is
desirable in many applications, e.g. land-mine detection, subways, airports. In all these
situations, it proved to be much more simple to bring the instrument to the sample rather
than the sample to the instrument.
Even if portability is a big issue in chemical sensing, trying to address also the require-
ment of selectivity is an harder task for researchers. Current chemical sensor technology is
limited in terms of selectivity [6, 7, 8, 9]. In order to create a reliable chemical sensor, it is
necessary to minimize the occurrence of false positives, particularly when detecting a tar-
get molecule against a background of various interfering agents. Modern sensor coatings
2
approach this problem by utilizing arrays of polymers from which different fingerprints are
detected depending on the chemical, though this method requires information processing
intensive methods of pattern recognition [10, 11].
Unfortunately, the problem of selectivity still remains an issue, as the polymers used
have inadequate differences in affinity for various targets making them susceptible to non-
specific binding. This is detrimental to the sensor as these pattern recognition techniques
rely heavily on the quality of the data provided to have minimal uncertainty. The use of
off-the-shelf polymers and oxide films does not provide the diversity necessary for sensor
coatings which will be essential for advancement of sensor technology.
In the chemical sensing laboratory where I have worked, they have mimicked the
principles of molecular recognition in biology to achieve highly selective target binding
[12]. Utilizing the combinatorial screening power of phage display, they have developed a
recognition motif of amino acids capable of selectively binding the explosive TNT (trini-
trotoluene) while remaining inactive to DNT (dinitrotoluene), differing in a single nitro
group. With this selective molecular recognition element (receptor), they want to develop
a coating and a method to immobilize the coating onto the cantilever sensing platform.
In May 2007, when I joined the Majumdar’s group, the project was regarding the
binding of the receptor on the gold surface of the cantilevers through an hydrophilic linker.
The hypothesis was that if this hydrophilic coating had sat on both sides of the cantilver
and it had been able to adsorb the same amount of water on the two sides, it would have
saturated the output signal due to the change in external relative humidity for a certain
threshold.
The ultimate goal of my experimental work was to optimize a coating able to link the
surface of the cantilever to the oligo-peptide receptor. This coating had to be optimized in
order to cancel, or at least, to reduce the effect of relative humidity change on the sensor
signal. Relative humidity is known to be one of the most important interfering agents
which affect chemical sensors. Avoiding the effect of water vapor is a challenging problem,
because water is always present in the environment in a concentration much higher than
the concentration of the target.
Since portable and cheap sensors are needed, it is unsuitable to condition the reference
air and odor headspace so that theirs humidity are held constant. It is much better to
exploit the natural presence of water in order to help the sensor working.
I would like to highlight that during my internship I have never worked with the TNT
receptor. I have taken part in the process of optimization of the cantilever coating material
and, in particular, I have performed the following operations:
• functionalization of the sensor surfaces with the coating;
• characterization of the coating through X-ray Photoemission Spectroscopy (XPS)
and Contact Angle (CA) measurements;
• measurements of the interaction between water vapor and coated cantilevers.
In the end, my work, concerning the optimization of a coating for chemical sensing,
has a value which is independent of explosives and TNT detection. In spite of that, the
3
1 – Introduction
concepts of chemical sensor and detection of explosives are extremely important in order to
contextualize my thesis and therefore I will deepen these concepts in the following chapter.
4
Chapter 2
Fundamentals of Chemical
Detection
2.1 Chemical Sensing: History and Definition
In instrumental chemical analysis, thanks to the advent of microelectronics, a general trend
towards miniaturization can be observed. In around 1950, many analytical instruments
were so big that they occupied entire laboratories. In the final decades of the last century, a
strong tendency towards automation appeared in chemical analysis. Big central laboratory
complexes were established. One reason for centralizing the resources is the high cost of
modern instruments.
However, at a certain point, it became obvious that not every problem could be solved
smoothly in this way. In many cases, it was difficult or impossible to transport samples
long distances without decomposition. Commonly, it proved to be much more simple to
bring the instrument to the sample rather than the sample to the instrument. Mobile
techniques of chemical analysis attracted increasing interest. Analysts started to look for
small, transportable analytical probes which could be stuck smoothly into a sample.
After the advent of Micro-Electro-Mechanical Systems (MEMS) another strong trend
towards miniaturization of measuring instruments in chemical analysis was observed. In
the most general form, MEMS consist of mechanical microstructures, microsensors, mi-
croactuators, and microelectronics, all integrated onto the same silicon chip.
There exist specific reasons to minimize the dimensions of analytical instruments. Very
small instruments are expected to work faster than big ones, since small-scale chemical
reactions generally happen faster and the volume of reagents involved is small. Moreover
it is much easier to carry out automation with smaller instruments and a higher degree
of automation improves the immunity of the instrument against interference, and can be
operated by less qualified personnel.
The term sensor started to gain currency during the 1970s [13] . This development was
caused by technological developments which are part of a technical revolution that contin-
ues to this day. Rapid adavances in microelecronics made available technical intelligence.
Machines became more “intelligent” and more autonomous. There arose a demand for
5
2 – Fundamentals of Chemical Detection
artificial sensing organs that would enable machines to orient themselves independently
in the environment. Robots, it was believed, should not execute a program blindly. They
should be able to detect barriers and adapt their actions to the existing environment.
In this respect, sensors first represented technical sensing organs , i.e. eyes, ears, and
tentacles of automatic machines. With our senses we can not only see, hear and feel, but
also smell and taste. The latter sensations are the results of some kind of chemical analysis
of our environment, either of the surrounding air or of liquids and solids in contact with
us. Consequently, chemical sensors can be considered artificial noses or artificial tongues.
Nowadays, it would be not be sufficient to see chemical sensors merely as some kind
of technical sensing organs. They can be used in many other fields besides just intelligent
machines. The official definition was given by IUPAC (International Union of Pure and
Applied Chemistry) in 1991 [14]:
A chemical sensor is a device that transforms chemical information, ranging
from concentration of a specific sample component to total composition anal-
ysis, into analytically useful signal.
Another good pragmatic desciption taken from the literature is the following [15]:
Chemical sensors are small-sized devices comprising a recognition element,
a transduction element, and a signal processor capable of continuously and
reversibly reporting a chemical concentration.
2.2 Elements of Chemical Sensors
Accodingly to IUPAC [16], chemical sensors usually contain two basic components con-
nected in series: a chemical (molecular) recognition system (receptor) and a physicochem-
ical transducer. In other documents, additional elements are considered to be necessary,
in particular units for signal amplification and for signal conditioning (see Figure 2.1).
Figure 2.2 shows the working principle of a general chemical sensor. When the molec-
ular analytes like explosives, biochemical warfare agents, and hazardous gases are intro-
duced into the sensor chamber by a chemical delivery system, those analytes react with
the receptor coating layer, and an electrical, optical, or mechanical response signal due to
the molecular interaction is generated by the transducer.
2.2.1 Classification of Sensors
Today, signals are processed nearly exclusively by means of electrical instrumentation.
Accordingly, every sensor should include a transducing function, i.e. the actual concentra-
tion value, a non-electric quantity, must be transformed into an electric quantity voltage,
current or resistance.
Classification of sensors is accomplished in different ways. Prevalent is a classification
following the principles of signal transduction [14] . The following sensor groups result.
• Optical sensors, following absorbance, reflectance, luminescence, fluorescence, re-
fractive index, optothermal effect and light scattering.
6
2.2 – Elements of Chemical Sensors
Figure 2.1: Scheme of a typical chemical sensor system [13]
Figure 2.2: General structure of a chemical sensor [17]
• Electrochemical sensors, among them voltammetric and potentiometric devices, chem-
ically sensitized field effect transistor (CHEMFET) and potentiometric solid elec-
trolyte gas sensors.
7
2 – Fundamentals of Chemical Detection
• Electrical sensors including those with metal oxide and organic semiconductors as
well as electrolytic conductivity sensors.
• Mass sensitive sensors, i.e. piezoelectric devices and those based on surface acoustic
waves.
• Magnetic sensors (mainly for oxygen) based on paramagnetic gas properties.
• Thermometric sensors based on the measurement of the heat effect of a specific
chemical reaction or adsorption which involves the analyte.
• Surface stress sensors.
• Other sensors, mainly based on emission or adsorption of radiation.
2.2.2 The Receptor
The receptor function is fulfilled in many cases by a thin layer which is able to interact with
analyte molecules, catalyse a reaction selectiveley, or partecipate in a chemical equilibrium
together with the analyte. Receptor layers can respond selectively to particular substances
or to a group of substances. The term molecular recognition is used to describe this
behaviour.
The interaction between the analyte and the sensor material can be reversible or irre-
versible. In the first case the analyte molecules dissociate from the sensor material when
the external concentration is removed, and overall they undergo no net change. Examples
of this type include the adsorption of gases into polymer films and the interaction of gases
with conducting polymers. In these cases the interaction between the sensor material and
the analyte are determined by the intermolecular forces between the two and are the results
of hydrogen bonding, dipole/dipole, dipole/induced dipole, dispersion, and hydrophobic
interactions. These interactions are dependent on the shapes and charge distributions
within the analyte molecules and the sensor material, and are similar to the interactions
operative in the biological system between the odorants and the receptor proteins.
In the irreversible case, the analyte undergoes a chemical reaction at the sensor surface
catalyzed by the sensor material. Here the analyte is consumed in the sensing process,
although the numbers of molecules reacting are often a small proportion of the total
number within the sample.
Chemical sensor materials can also be divided into three categories according to the
type of receptor material used.
First, there is a range of inorganic crystalline or polycrystalline materials which have
been used as the chemically sensitive materials in gas sensors. These include semiconduc-
tors, as in MOSFET structures, metal oxides, zeolite absorbent materials, and metallic
catalysts. In general, these materials are robust and are often operated at elevated tem-
peratures where they function as catalytic materials, in irreversible, or chemically reactive,
sensors.
Second, there is an increasing range of organic materials and polymers which are finding
application in chemical sensors and gas sensors. These materials have the advantage of
8
2.3 – Characterization of Chemical Sensors
being more flexible in design and more readily modified chemically to develop arrays of
materials with different properties. In general, these materials are used at, or close to,
room temperature and operate as reversible sensor materials.
Third, there are biologically derived materials such as proteins, enzymes, and antibod-
ies. These materials often offer considerable selectivity and have been much investigated
as components of biosensors.
2.3 Characterization of Chemical Sensors
2.3.1 Parameters of Chemical Sensors
The performance of chemical sensors should be expressed in the form of numbers. The fol-
lowing list contains static as well as dynamic parameters which can be used to characterize
the performance of chemical sensors [13].
• Sensitivity : change in the measurement signal per concentration unit of the analyte,
i.e. the slope of the calibration curve.
• Detection limit : the lowest concentration value which can be detected by the sensors
in question, under definite conditions. Whether or not the analyte can be quantified
at the detection limit is not determined. Procedures for evaluation of the detection
limit depend on the kind of sensor considered.
• Dynamic range: the concentration range between the detection limit and the upper
limiting concentration.
• Selectivity : an expression of whether a sensor responds selectively to a group of an-
alytes or even specifically to a single analyte. Quantitative expressions of selectivity
exist for different types of sensors.
• Linearity : the relative deviation of an experimentally determined calibration graph
from an ideal straight line. Usually values for linearity are specified for a definite
concentration range.
• Resolution: the lowest concentration difference which can be distinguished when the
composition is varied continuosly. This parameter is important chiefly for detectors
in flowing streams.
• Response time: the time for a sensor to respond from zero concentration to a step
change in concentration. Usually specified as the time to rise to a definite ratio of
the final value. Thus, e.g. the value of t99 represents the time necessary to reach
99% of the full-scale output. The time which has elapsed until 63% of the final value
is reached is called the time constant.
• Hysteresis: the maximum difference in output when the value is approached with an
increasing and a decreasing analyte concentration range. It is given as a percentage
of full-scale output.
9
2 – Fundamentals of Chemical Detection
• Stability : the ability of the sensor to maintain its performance for a certain period
of time. As a measure of stability, drift values are used, e.g. the signal variation for
zero concentration.
• Life cycle: the length of time over which the sensor will operate. The maximum
storage time (shelf time) must be distinguished from the maximum operating life.
The latter can be specified either for continuous operation or for repeated on-off
cycles.
2.3.2 ROC Curve
There are three figures of merit related to gas sensing technology: reversibility, sensitivity,
and selectivity [18].
First, reversibility refers to the possibility of multiple usage of the sensor. If the
chemical reaction between target and probe molecules is too strong, the sensor can not
recover after it is exposed to a target, making it unavailable for further usage. As we said
before, to get a reversible sensor, weak interaction mechanisms, such as van der Waals
force, hydrogen binding, etc., should be used as the molecular interaction mechanism.
Second, sensitivity determines the level of response of a sensor to a target. Highly
sensitive devices produce a large measurable response for a given target level. The basic
approach to increase the sensitivity is to increase the sensor surface area to volume ra-
tio. Sensitivity is related to the smallest amount of target that can be detected, which
is the detection limit, and for most explosives lies in the range from ppb to ppt (see
Section 2.5). Commercially available gas sensors like metal oxide, conductive polymer,
and surface acoustic wave (SAW) sensors have reasonable detection limits (several part
per million) for various gases but they can not detect ultra low concentration of chemical
warfare agents. Recently, silicon nitride or silicon microcantilevers coated with a thiol self
assembled monolayer or polymers have been studied for gas phase explosive detection at
the 10–30 part per trillion (ppt) level of concentration, as reported in Section 3.4.
Third, selectivity refers to the degree of specificity to a particular target. Most gas
sensors have partial selectiveness, so they respond to interfering agents as well as the
specific target agent. Water vapor or humidity is known to be a major interfering agent
as it is explained in the next section. In general, when we try to increase the sensor
sensitivity, the fraction of false positive signal tends to increase as well. In other words, the
selectivity of a sensor decreases. Therefore, increasing sensitivity alone does not improve
system performance. One needs to improve selectivity as well, thereby creating improved
ROC curves (see Figure 2.3). Therefore, we need to develop highly selective sensor coating
materials on top of highly sensitive sensors. The questions that need to be addressed are:
what kind of coating material is promising and how to test lots of potentially promising
sensor coating materials.
Figure 2.3 shows a typical Receiver Operating Characteristic (ROC) curve (true pos-
itive rate or sensor sensitivity versus false positive rate) that is widely used. The best
condition is described by the red trend: true positive rate next to 1 (it means that every-
time the target is present in the sensor chamber, the sensor responds, i.e. high sensitivity),
10