Advanced Biomaterials and Biodevicess

Part 2 INNOVATIVE BIODEVICES Ashutosh Tiwari and Anis N. Nordin (eds.) Advanced Biomaterials and Biodevices, (289–304) 2014 © Scrivener Publishing LLC

8 Label-Free Biochips Anis N. Nordin1,2 1Biosensors and Bioelectronics Centre, IFM, Linköping University, Linkoping Sweden 2Department of Electrical and Computer Engineering, International Islamic University Malaysia, Malaysia Abstract Artifact-free sensing or label-free biosensors are fast gaining popularity today due to their simple preparation methods, accuracy and noise-free results. This chapter focuses on the usage of label-free sensors for biological applications. There are various ways for non-tagged sensing, but the most common are using electro- chemical and acoustic wave sensors. A brief overview on the operating principles for both sensors is described in this chapter such that even non-chemists or biolo- gists can easily understand the device concept. State-of-the-art example devices for both electrochemical and acoustic wave sensors are also explained so that the reader can be updated with the current technology. Special emphasis is given to sensors that can be fabricated using microelectromechanical (MEMS) systems techniques, a fabrication process that is adapted from the commercial integrated circuits for interdisciplinary applications such as healthcare diagnostics or toxicity detections. MEMS biochips are getting increased attention from researchers due to its miniature size, portability and parallel processing capabilities. Keywords: Biosensors, MEMS, Bio-MEMS, label-free, electrochemical, bulk acoustic wave (BAW), surface acoustic wave (SAW), FBAR 8.1 Introduction The field of biosensors has been known for the past sixty years, with numer- ous available commercial devices available in the market. Amongst the *Corresponding author: [email protected]; [email protected] Ashutosh Tiwari and Anis N. Nordin (eds.) Advanced Biomaterials and Biodevices, (289–304) 2014 © Scrivener Publishing LLC 291

292 Advanced Biomaterials and Biodevices most widely used biosensor is the handheld amperometric glucose sensor (Turner, 2013). Progression of technology and an educated society have led to the popularity of personalized health monitoring devices. These devices allow users to be more in control of their health, leading towards better quality of lives and less frequent hospital visits. A driving force of the miniaturization of the biosensors is the tech- nological advancement of fabrication techniques in the semiconductor industry. There has been growing interest to develop miniature, portable and low-cost biosensors fabricated using MEMS technologies. MEMS technology has been adopted from the integrated circuit (IC) industry and applied to the miniaturization of a large range of sensing systems includ- ing biosensors. “Biological micro-electromechanical systems” (BioMEMS) is a subset of MEMS devices where biosensors are fabricated using either MEMS or IC fabrication techniques. Exciting advancement in technology has been reported in this field in the recent decades. Numerous biosen- sors have been fabricated to detect for a wide range of applications such as water toxicants (Arlett, Myers, & Roukes, 2011; Schneider, Marison, & von Stockar, 1996; Voiculescu et al., 2013) cytotoxicity studies, drug develop- ment, chemokinetic and chemotactic activity of cell and wound healing assays in vitro (Ayliffe, Frazier, & Rabbitt, 1999; Han, Yang, & Frazier, 2007; Keese & Giaever, 1994; Xiao & Luong, 2005). A niche area of focus in this field is as point of care (PoC) devices, which is has a wide range of applica- tions in the areas of genetics, diagnostics, drug discovery, environment, industrial monitoring and quality control (Satyanarayana, 2005). There are many reasons why scaling down in size is beneficial for biological analyses. Biological samples are small (nm) in size, making micrometer-sized tools for analysis better suited than conventional labora- tory beakers and test tubes (cm). Small lab-on-chips require only a small amount of biological samples, enabling more experiments to be run which leads to better result interpretation. Miniaturization also allows array of sensing elements, making it possible to have parallel and automated analy- ses when connected to the computer. 8.2 Label-Free Analysis Label-free methods are ideal for analyzing direct interactions between compounds and biological targets. Interactions can be in the form of chemical response or formation of molecular receptor-ligand binding. For bio-chemical sensors, additional chemistry other than the basic may be

Label-Free Biochips 293 required to achieve electrical signals. Labeling or tagging is defined as the usage of additional chemicals other than the receptors to obtain an output signal at the transducer. A typical example of labeling is using dyes with nucleic acid probes to produce fluorescence. Unfortunately, the use of labels or dyes, also create artifacts in the output response. Removing the necessity of tagging allow researchers to conduct biological, physiological, and pharmacological analyses under conditions that more closely mimic those found in a living organism with in vitro assays. Elimination of reagents or cell engineering, also simplifies both assay development and sensor fabrication. In drug discovery research for example, label-free assays can developed for biological target validation, pathway deconvolution, receptor panning, agonist trafficking, as well as ligand-bias and orphan target assay development, among others. Label-free analysis has also gained popularity due to the success of pathway analysis- based drug discovery, which focuses on understanding interdependencies among biological pathways. The usage other methods such as electrical or chemical signals instead of dyeing, allows scientists to gain more under- standing on cell behavior and how they communicate with each other. 8.3 Electrochemical Biosensors Electrochemical biosensors convert biochemical events into electrical sig- nals. When a biological sample is placed on the sensor, the sensor should be able to produce an electrical signal, which corresponds to the concen- tration of the sample. Electrical signals are produced due to the transfer of electrons at the solid-liquid interface of the sensor. Depending on the experimental setup, different forms of electrical signals can be obtained such as voltage (potentiometry), current (amperometry) and impedance (impedance spectroscopy). In brief, potentiometry involves the measurement of voltage in the bio- sensor or electrochemical cell, when no current is applied. Measured volt- age should be proportional to the concentration of the analyte under test. The typical equation relating voltage or potential, E in the presence of an ion a, and an interfering species, b can be described using the Nernst equa- tion set out as Equation 8.1 below: RT( )E = E0 − Fx log aa + K a,b ab (8.1)

294 Advanced Biomaterials and Biodevices Solution Reference Electrode VG Gate Single strand Source Insulator DNA Induced channel Drain Substrate VG Figure 8.1 Simplified schematic of Potentiometric DNA Biosensor. where R is the gas constant, T is the temperature, F is the Faraday constant, x is the number of electrons involved in the electrochemical process, aa and ab are the activities of the two ions and Ka,b is the selectivity coefficient of the electrode. Due to its easy integration with electronics, a popular form of the poten- tiometric sensor is using the field-effect transistors (FETs). The FETs have three electrodes, drain, gate and source, placed on top of a silicon substrate as shown in Figure 8.1. The gate acts as the sensing or bioactive electrode and is electrically insulated from the drain and source by an oxide layer. A specific voltage, VD is applied to the drain electrode and the source elec- trode is connected to zero potential or ground. To eliminate background noise and to allow charge transfer, the reference electrode is biased at a volt- age VG. The reference electrode charges the ions in the probe at a specific potential. For DNA detection, single-stranded DNA is immobilized on the gate. When target DNA is present on the gate, hybridization occurs and the number of ionic charges on the surface of the gate changes. The change in surface charge created a electronic channel or pathway between the drain and the source. The concentration of the detected DNA is reflected in the amount of measured current flowing between the drain and the source. Other than making voltage measurements when no current is applied, the same experimental setup can be used instead by applying different voltages and measuring the resulting current. When different voltages or potential are applied, it disturbs the equilibrium condition of the solution and charge transfer and electron movement will occur on the surface of the

Label-Free Biochips 295 bioactive electrode. This technique is known as amperometry. Different variations of this technique are employed in experiments such as: i. Chronoamperometry, which applies a square-wave potential and monitors the change in steady state current with respect to time. ii. Cyclic voltammetry, applies a triangular-wave potential and monitors the change in current with respect to applied volt- age (cyclic voltammogram). Both amperometry and potentiometry techniques use direct current (DC) or fixed potentials during their experiments. In contrast, electro- chemical impedance spectroscopy (EIS) technique uses sinusoidally vary- ing voltages, applied at different angular frequencies. This slightly more complex technique measures the output angular current and computes the real and imaginary components of electrical impedance using Ohm’s law as set out in Equation 8.2 below: Zˆ( ) = Eˆ( ) (8.2) Iˆ( ) where w is the angular frequency, Z is the impedance, E is the potential and I is current. ^ indicate the complex value of the impedance, current and voltage respectively. As can be seen, all these three techniques impedance, amperometry and potentiometry basically use the same setup, with minor modifications in the applied voltage and measured output. Due to this, the basic transis- tor can be used for all three techniques. Figure 8.2 illustrates a compact biochip, which incorporates all three different sensors on the same silicon chip (Nakazato, 2013). These sensors were fabricated using conventional complementary oxide semiconductor (CMOS) process, which is a mature technology for fabrication of integrated circuits (IC). This biochip lever- ages on the low-cost IC fabrication process to produce highly accurate, miniature, compact, real-time biosensors. The sensitivity of the FET biosensors can be dramatically improved by placing nanostructures on top of the active surface. Figure 8.3 is such an example where silicon nanowire FETs are used to increase the surface area of the bioactive layer. In this way, the FET can detect even small variation of charges that occur on the wire’s surface. The variation of charges is reflected

296 Advanced Biomaterials and Biodevices Amperometric Potentiometric (pH) Potentiometric Impedimetric Figure 8.2 Silicon-based amperometric, potentiometric and impedimetric sensor all integrated on a single chip (Nakazato, 2013). A 15–20 m B S D Ca2+ Nu ~ 2nn DAQ Hypoxic buffer Lock-in PC12 cell amplifier Dopamines C SD 3m MPC aptamer/SiNW-FET Aptamer MBS 5‘ 3‘ APTMS PTMS ~ 1nn SiNW SiNW SiNW SiNW Figure 8.3 (A) Illustration of the experimental setup of a DNA-aptamer- modified MPC SiNW-FET device for detecting exocytotic DA under hypoxic stimulation from living PC12 cells. (B) Optical microscopy image of an MPC SiNW-FET device. S = source; D = drain. (C) Procedure for immobilization of the DNA-aptamer on an MPC SiNW- FET. Abbreviations: APTMS, (3-aminopropyl)trimethoxysilane; PTMS, propyltrimethoxysilane; MBS, 3-maleimidobenzoic acid N-hydroxysuc- cinimide ester (Li et al., 2013).

Label-Free Biochips 297 as the change in measured current. These silicon nanowire FETs are used to detect very low concentrations (~100 pM) of dopamine. DNA-aptamers were functionalized on the biosensors. The sensor is not only very sensitive, it is also has high specificity and is able to detect dopamine from ascorbic acid, catechol, phenethylamine, tyrosine, epinephrine, and norepinephrine. This sensor was also applied to monitor DA release under hypoxic stimu- lation from living PC12 cells. The real-time recording of the exocytotic DA induced by hypoxia reveals that the increase in intracellular Ca2+ that is required to trigger DA secretion is dominated by an extracellular Ca2+ influx, rather than the release of intracellular Ca2+ stores (Li et al., 2013). 8.4 Acoustic Wave-based Mass Sensors Another totally different label-free measuring technique that is commonly used in biosensors is acoustic-wave based mass sensing. This device employs piezoelectric acoustic waves, or traveling waves in a piezoelectric layer or crystal. Piezoelectricity is a phenomenon where mechanical strain in the material is generated by an applied electrical field or vice versa. The applied electric field can be in the form of oscillating sinusoidal electrical signals, and in such case, the induced mechanical strains in the material will be in the form of standing or travelling waves. The acoustic waves can travel either on the surface of the piezoelectric material (surface acoustic waves - SAW) or through the piezoelectric layer (bulk acoustic waves - BAW). Both modes of propagation SAW and BAW can be used for biological sensing. For biological applications the acoustic wave-based MEMS devices are integrated in a microfluidic system and the sensing area is coated with a biospecific layer. When a bioanalyte interacts with this sensing layer, physi- cal, chemical, and/or biochemical changes are produced. Typically, mass and viscosity changes of the biospecific layer can be detected by analyzing changes in the acoustic wave properties such as velocity, attenuation and resonant frequency of the sensor. An important advantage of the acous- tic wave biosensors is simple electronic readout that characterizes these sensors. The measurement of the resonant frequency or time delay can be performed with high degree of precision using conventional electronics. 8.5 Bulk Acoustic Wave Sensors Bulk acoustic wave resonators have been widely used for mass detec- tion in the form of quartz crystal microbalances (QCM). As shown in

298 Advanced Biomaterials and Biodevices Liquid Electrodes Quartz Figure 8.4 Schematic of bulk acoustic wave resonator. Thickness shear oscillation Figure 8.4, the structure of the QCM is fairly simple, the quartz disc is sandwiched between two electrodes (often gold) and the top electrode is exposed to the biological sample. When oscillating voltage is applied between the electrodes, the oscillating electric field generates mechan- ical acoustic waves propagating through the substrate. The mechanical waves resonate at a specific resonance frequency, f0. The resonance fre- quency is dependent on the thickness of the piezoelectric layer. When biological samples are placed on top of the gold electrode, the samples dampen the mechanical acoustic waves, creating a shift in resonant fre- quency, Df. The Sauerbrey equation correlates the changes of the resonant frequency of an acoustic wave resonator with the mass deposited on it. The acoustic wave propagating on a piezoelectric substrate is generated and received using IDTs. In the case of a biosensor resonator, the cell to be analyzed or the anti- body layer for protein marker detection are added on the IDTs. This will cause a shift of the resonant frequency due to the increasing of mass, where fi and fo are the are the resonant frequencies before and after loading the sensor. In Equation 8.3, the Sauerbrey equation is defined as f 2 f02 m 2.26 106 f 2 m (8.3) A qq 0 A where Δf = f0-fi From Eq 8.3, the change Δf of the resonant frequency of the piezoelec- tric crystal is directly proportional to the mass loaded on the acoustic wave resonator, where Δm is expressed in g and Δf and f0 in Hz (Skládal, 2003). Due to the simple structure of the piezoelectric resonator, which requires only electrodes and the piezoelectric layer, they are very easy to

Label-Free Biochips 299 miniaturize. Recent technological advancement of MEMS processes allows the fabrication of thin piezoelectric films and the integration of acoustic wave based devices, and electronics on a common silicon substrate. A state-of-the-art example of a MEMS-based BAW resonator also known as the film bulk acoustic wave resonator (FBAR) is shown in Figure 8.5. In this device, the piezoelectric layer is a thin layer of ZnO, which generates resonance frequency of 1.5 GHz. This sensor was used to detect a cancer biomarker for prostate cancer, human prostate-specific antigen (hPSA) (Zhao et al., 2014). Mouse monoclonal antibody (anti-hPSA) was used to bind hPSA to the top gold electrode. The FBAR hPSA sensor has mass sen- sitivities of 1.5ng/cm2. Antigen binding experiments using FBAR sensors demonstrated that FBARs have the capability to precisely detect antigen binding, thereby making FBARs an attractive low cost alternative to exist- ing cancer diagnostic sensors. Another example of bulk acoustic wave biosensor is a cell-based bio- sensor used to test water toxicants in drinking water (Figure 8.6). This Top electrode Via (contact hole) Piezoelectric film 2 m ZnO Bottom electrode 2 m SiO2 Piezoelectrically active area 500 m Si Membrane released by DRIE Figure 8.5 Cross section of film bulk acoustic wave sensor for detection of human prostate-specific antigen (Zhao et al., 2014). Enclosed BAECs Perfusion inlet culturing well Array of miniaturized Perfusion enclosed PDMS culturing outlet wells Common QCM and AT-cut quartz ECIS sensitive electrode wafer ECIS counter electrode Figure 8.6 Multiparametric sensor array integrated with miniaturized enclosed polymeric culturing chambers (Voiculescu et al., 2013).

300 Advanced Biomaterials and Biodevices biosensor incorporates both acoustic wave mass-sensing and impedance spectroscopy on the same device. Bovine aortic endothelial cells were seeded on the gold electrodes and used to detect the presence of differ- ent toxicants: aldicarb, nicotine and ammonia. The presence of toxicants will kill the BAECs and cause changes in both the resonance frequency and impedance measurements. Linear correlation between resonant fre- quency shifts and concentration of toxicants were obtained using this sensor. The usage of dual, simultaneous detection methods (impedance and mass-sensing) allows cross-validation and reduces false-positives during measurements. 8.6 Surface Acoustic Wave Mass Sensors As mentioned earlier, another form of acoustic wave mass sensors uti- lize surface waves. This type of sensor has its transducers only on top of the piezoelectric layer, making it easier to fabricate. SAW sensors also typically operate at higher resonance frequencies (200 MHz – 1GHz) compared to QCMs, making them more sensitive to mass changes as sensitivity is dependent on resonant frequency. The SAW’s transducers are often in the form of interdigital transducers (IDTs) and the spac- ing between the alternating transducers determines the resonance fre- quency of the device. When alternating voltage is applied the input IDTs induce surface waves in the piezoelectric material. The surface waves are detected by the output IDT and induce an alternating current. Depending on the cut of the piezoelectric material, the surface acous- tic waves can have either compressional (Figure 8.7A) or shear (Figure 8.7B) components. When used as a biosensor, the surface between these two sets of IDTs is covered with a biological layer sensitive to the analyte to be detected. The absorption of the analyte on the sensitive layer will produce a time delay in the acoustic wave propagation. The main disadvantage of the Rayleigh wave based devices when used as biosensors is the degradation of perfor- mance due to liquid damping. In liquid the quality factor Q drops (usually more than 90% reduction) and negatively affects the device sensitivity. Since most of the biological applications are performed in liquid only very few Rayleigh wave acoustic wave devices could be integrated in microfluidic channels, without significant degradation of the sensor performance.

Label-Free Biochips 301 Figure 8.7 A. Schematic of surface acoustic wave device with Rayleigh waves. B. Schematic of surface acoustic wave device with shear horizontal surface waves. To counteract liquid damping, shear horizontal surface acoustic waves (SH-SAW) have been used. This mode of surface wave is less susceptible to liquid damping because the lateral waves are less damped compared to the compressional waves. Recently, researchers in Tsukuba, Japan have reported successful employment of a SH-SAW device for protein immu- nosensing. The interaction between thrombin and a thrombin-binding aptamer (TBA) on a PEG/TBA coimmobilized surface was studied immobilizing the aptamers between the two input and output IDTs as shown in Figure 8.8. The SH-SAW sensor is very sensitive and its detection limit of throm- bin by the optimized TBA-T(40)/PEG-b-PAMA(23) surface was around 7.5 pmol. This sensor complements existing detection mechanisms such as surface plasmon resonance and the QCM as the SH-SAW sensor detects both mass as well as viscosity changes on the thin active layer.

302 Advanced Biomaterials and Biodevices Drop the solution (30 L) Thrombin binding Thrombin aptamer PEG-b-PAMA Gold surface Thin gold film Wave direction Input IDT Quartz substrate Output IDT Noise-cancelling slit Figure 8.8 Illustration of the SH-SAW sensor chip, thrombin binding on the gold active surface and the propagation of the shear horizontal wave on the quartz substrate (Horiguchi, Miyachi, & Nagasaki, 2013). 8.7 Conclusion and Future Prospects It can be seen that with the increasing popularity of portable point-of- care devices, label-free biosensors will become more and more important. Label-free sensors offer real-time measurements, less preparation of sam- ples and minimization of nonspecific adsorption of unwanted materials. Reduction of adsorption of nonspecific materials is especially important because it creates noise in the already small output signal. Conventional assay methods such as radioimmunoassay (RIA), latex immunoassay and enzyme-linked immunosorbent assay (ELISA) all produce end-point mea- surements, making it difficult for real-time measurements and requires multiple sample preparations for continuous analysis. This chapter illustrated two different label-free methods of detection namely electrochemical sensing and acoustic wave mass sensing. State-of- the art examples were described for each technique and special focus was given to devices, which can be fabricated using MEMS fabrication tech- niques. A mature technology in the field of automotive sensors, MEMS lab-on-chips are fast gaining popularity as a healthcare diagnostics due to their portability, parallel and real-time measurement capabilities and low- cost, batch production fabrication process. Label-free MEMS biosensors with novel functions have promising pros- pects in the field of clinical and non-clinical diagnostics. When coupled

Label-Free Biochips 303 with nanostructures, label-free biosensors are highly sensitive; creating pos- sibilities for a new generation of sensors for biological and infectious agents in early-stage detection of disease and threats. Accessibility to the nanome- ter range allows researchers to understand various biomolecule-transducer interactions with interesting nanomaterials, which is especially important in the field of electrochemistry. The nanostructures can be functionalized with specific agents to bind target molecules or can be doped with electroni- cally active materials to enhance charge transfer at the surface. The usage of customized biochips as the electrodes allow the sensors to be designed as arrays which results in parallel real-time monitoring of multiple analytes. These type of sensors provide the capabilities measuring accurate, ampli- fied signals from nL samples which opens the doors for multi-disciplinary researchers to create a new generation of portable, accurate, innovate bio- sensor arrays for health care and environment monitoring. References 1. Arlett, J., Myers, E., & Roukes, M. (2011). Comparative advantages of mechanical biosensors. Nature Nanotechnology, 6(4), 203–215. 2. Ayliffe, H. E., Frazier, A. B., & Rabbitt, R. (1999). Electric impedance spectroscopy using microchannels with integrated metal electrodes. Microelectromechanical Systems, Journal Of, 8(1), 50–57. 3. Han, A., Yang, L., & Frazier, A. B. (2007). Quantification of the heterogeneity in breast cancer cell lines using whole-cell impedance spectroscopy. Clinical Cancer Research, 13(1), 139–143. 4. Horiguchi, Y., Miyachi, S., & Nagasaki, Y. (2013). High-performance surface acoustic wave immunosensing system on a PEG/aptamer hybridized surface. Langmuir, 29(24), pp 7369–7376. 5. Keese, C. R., & Giaever, I. (1994). A biosensor that monitors cell morphology with electrical fields. Engineering in Medicine and Biology Magazine, IEEE, 13(3), 402–408. 6. Li, B., Hsieh, Y., Chen, Y., Chung, Y., Pan, C., & Chen, Y. (2013). An ultra- sensitive nanowire-transistor biosensor for detecting dopamine release from living PC12 cells under hypoxic stimulation. Journal of the American Chemical Society, 135(43), pp 16034–16037. 7. Nakazato, K. (2013). “Potentiometric, amperometric, and impedimetric CMOS biosensor array”, State of the Art in Biosensors - General Aspects, InTech, 2013. 8. Satyanarayana, S. (2005). Surface Stress and Capacitive MEMS Sensor Arrays for Chemical and Biological Sensing, (Doctoral dissertation, University of California). 9. Schneider, M., Marison, I. W., & von Stockar, U. (1996). The importance of ammonia in mammalian cell culture. Journal of Biotechnology, 46(3), 161–185.

304 Advanced Biomaterials and Biodevices 10. Skládal, P. (2003). Piezoelectric quartz crystal sensors applied for bioanalyti- cal assays and characterization of affinity interactions. Journal of the Brazilian Chemical Society, 14(4), 491–502. 11. Turner, A. P. (2013). Biosensors: Sense and sensibility. Chemical Society Reviews, 42(8), 3184–3196 12. Voiculescu, I., Li, F., Liu, F., Zhang, X., Cancel, L. M., Tarbell, J. M., et al. (2013). Study of long-term viability of endothelial cells for lab-on-a-chip devices. Sensors and Actuators B: Chemical, 182, 696–705. 13. Xiao, C., & Luong, J. H. (2005). Assessment of cytotoxicity by emerging impedance spectroscopy. Toxicology and Applied Pharmacology, 206(2), 102–112. 14. Zhao, X., Pan, F., Ashley, G. M., Garcia-Gancedo, L., Luo, J., Flewitt, A. J., et al. (2014). Label-free detection of human prostate-specific antigen (hPSA) using film bulk acoustic resonators (FBARs). Sensors and Actuators B: Chemical, 190, 946–953.

9 Polymer MEMS Sensors V.Seena1,2,*, Prasenjith Ray1, Prashanthi Kovur1,3, Manoj Kandpal1 and V. Ramgopal Rao1 1Centre of Excellence in Nanoelectronics, Department of Electrical Engineering, Indian Institute of Technology, Bombay, India 2Department of Avionics Engineering, Indian Institute of Space Science and Technology, Thiruvananthapuram, India 3Department of Chemical and Materials Engineering, University of Alberta, Edmonton, Canada Abstract The evolution of today’s sensors based on the micro/nano electromechanical sys- tems (MEMS/NEMS) happened due to the revolutions in the well-established microelectronics technology. Though silicon is considered to be the primary material in microelectronics and hence in MEMS, many classes of MEMS devices have been realized using other potential materials like polymers. A class of MEMS sensors named nanomechanical cantilevers find applications in the realization of many physical, chemical, and biological sensors. Improved sensitivity, reliability and also cost effectiveness of such sensor platforms have been achieved by the use of polymer materials, along with the employment of smart and compatible trans- duction techniques. This chapter summarizes our research work on development of polymer MEMS cantilever sensor platforms with four novel integrated electrical transduction mechanisms. In these techniques, the mechanical parameters of the polymer (SU-8) MEMS sensors can be translated into electrical output using (1) SU-8/CB nanocomposite (a piezoresistive approach), (2)integrated organic field effect transistor of CantiFET (a strain sensitive transistor approach), (3) integrated Al doped ZnO TFT (a strain sensitive thin film transistor approach) or (4) SU-8/ ZnO nanocomposite ( a piezoelectric approach). *Corresponding author: [email protected] Ashutosh Tiwari and Anis N. Nordin (eds.) Advanced Biomaterials and Biodevices, (305–342) 2014 © Scrivener Publishing LLC 305

306 Advanced Biomaterials and Biodevices KeyWords: Polymer MEMS, polymer microcantilever, nanomechanical cantile- ver, SU-8, CantiFET, AZO thin film transistor, polymer nanocomposite, SU-8/ CB, piezoresistive microcantilever, piezoelectric microcantilever, SU-8/ZnO 9.1 Introduction A sensor can be defined as a device that converts a non-electrical, physical or chemical input into an electrical output signal. Sensors can be classi- fied according to the energy domain of its primary input/output, as elec- trical, thermal, radiation, mechanical, magnetic and bio/chemical sensors [1]. The well-established integrated circuit industry played a major role in creating an environment suitable for the development of microsystems known as microelectromechanical systems (MEMS)[2]. MEMS sensors are used in various industrial, consumer, defence and biomedical appli- cations. Microaccelerometers, pressure sensors and microarrays are some of the commercially available MEMS sensors. MEMS being a technol- ogy derived from microelectronics, these miniature MEMS sensors hold advantages such as, low cost of production due to very large production volume, easy integration with required instrumentation on microelectron- ics chips, arraying capability enabling multiplexed measurements, greater portability, robustness and low power consumption. Many tools used in the design and manufacturing of MEMS devices are borrowed from the conventional IC industry. Hence silicon is considered to be the primary material even in MEMS, though MEMS systems using other materials like polymers, metals and ceramics have been demonstrated. There have always been demands for the detection of very low levels of a large number of chemical and biological substances in application areas such as environmental monitoring, healthcare, biomedical technology, clinical analysis and food processing. For example, in the case of home- land security applications, recent increase in security concerns in public places like airports and public transports have increased the demand for low-cost portable, efficient, and easy to use explosive sensing technology. The emergence of MEMS and nanotechnology along with these demands enabled the development of a class of microsensor systems called as micro- fabricated bio/chemical sensors. Bio/chemical sensors combine a bio/chemi- cal recognition element coupled to a physical transducer [3]. A biological or chemical recognition element recognizes a specific target analyte and does not recognize other analytes, which imparts selectivity to the sensor. The transducer translates the bio/chemical-recognition event into measur- able quantities such as change in electrical signal, an optical emission, a

Polymer MEMS Sensors 307 mechanical motion etc. In the case of biosensors, the recognition element may be an enzyme, antibody, antigen, living cells, tissues, etc. and in the case of chemical sensors, these can be any chemical substance specific for the target analyte. A class of MEMS sensors known as microcantilever sensors (nanome- chanical cantilever sensors) came into existence as atomic force micros- copy (AFM) probes [4] and they have a very good potential as platform for the development of many physical [5], chemical [6], and biological sensors [7–9]. Microcantilever based bio/chemical sensors work on the principle of conversion of the bio/chemical recognition event into nano- mechanical motion [10]. The cause for nanomechanical motion can be due the free energy change on the surface of the sensor due to the reaction of the target analyte with the receptor molecule or due to the mass change on addition of the target analyte bound to the microcantilever surface (Figure 9.1). These nanomechanical cantilever sensors offer many orders of magni- tude higher sensitivity in comparison to other commonly used bio/chemi- cal sensors such as quartz crystal microbalances (QCM), flexural plate wave oscillators (FPW), and surface acoustic wave devices (SAW) [11]. Large surface to volume ratio, miniaturization and mass production at a relatively low cost, several modes of operation and label free detection, feasibility for fabrication of multi-element sensors arrays supporting high degree of parallelization, ease of integration of microcantilever sensor with on-chip electronic circuitry are the distinct advantages of the nanome- chanical cantilever sensors that support their candidature for bio/chemical sensing applications. The sensing operation modes of microcantilevers are classified based on their principles in translating the recognition event into nanomechanical motion. Typically there are three modes of operation such as static mode, dynamic mode and heat mode [12]. In static mode, the bending of the microcantilever upon the molecular adsorption is measured. In dynamic Target molecule Target binding Probe molecule Modified surface Anchor Anchor Deflection h Microcantilever structural layer Figure 9.1 Principle of microcantilever based bio/chemical sensor.

308 Advanced Biomaterials and Biodevices mode, the dependence of resonant frequency of the microcantilever on the mass of the microcantilever is exploited. The heat mode, takes advantage of the bimetallic or bimorph effect that leads to a bending of a biomaterial microcantilever with change in temperature. The performance of a micro- cantilever sensor relies on real-time measurements and the resolution of measurements of cantilever mechanical parameters during sensing opera- tion. Microcantilever transduction schemes are broadly classified as opti- cal and electrical [8]. Microcantilever sensors with an optical transduction scheme are expected to offer the highest sensitivity. However, owing to the practical limitations pertaining to the field deployment of such opti- cal transduction based sensors, the integrated electrical transduction mechanism inside the mechanical element are usually preferred. There are different varieties of electrical transduction schemes reported such as piezoresistive, piezoelectric, capacitive, electron tunnelling technique and embedded MOSFET technique [15]. Most commonly used microcantilever structural materials are single crystalline silicon, polycrystalline silicon, silicon nitride, silicon dioxide and mechanically stable polymers like SU-8, TOPAS and Parylene [13–15]. As per the working principle of microcantilever based surface stress sen- sor, the sensitivity is determined by the stiffness of the cantilever structure and hence by the Young’s modulus of the material. The need for highly sen- sitive and inexpensively fabricated microcantilever sensors motivated the researchers to explore polymer based microcantilever technologies. Among the polymers reported for microfabrication, SU-8 which is an epoxy based polymer developed by IBM, is the most commonly used polymer struc- tural material in MEMS [16–18]. Since 1999 [19], the use of SU-8 poly- mer which is also considered as a high aspect ratio negative photoresist for MEMS applications has been exponentially growing during the last couple of years [20–22]. SU-8 has the ability in forming patterns with wide range of thickness varying from few hundred of nanometres to a few millimetres with high aspect ratios. SU-8 seemed to be a good candidate for structural material for microcantilever sensors with its inherent advantages such as low Young’s modulus, inexpensive and less complex fabrication process, a well understood UV and e-beam resist with low consumption of chemicals and gases and low temperature for fabrication processes making it cost effective and compatibility for integrating sensor with microfluidics. The performance of SU-8 based polymer nanomechanical sensors could be enhanced by incorporating ultra-sensitive electrical transduction schemes that are compatible with SU-8 processing [23–32]. Design and development of SU-8 microcantilever sensors with four novel electrical transduction schemes namely (1) polymer nanocomposite piezoresistive

Polymer MEMS Sensors 309 microcantilever sensors [25–27] (2) Organic CantiFET [28] (3) AZO CantiFET [29] and (4) polymer nanocomposite piezoelectric microcanti- lever sensors [30–32] are discussed here. 9.2 Polymer Nanocomposite Piezoresistive Microcantilever Sensors Microcantilever sensors with an integrated piezoresistor perform electrical transduction of strain by a resultant change in resistance. When a surface functionalized microcantilever with molecules or thin film coatings that are specific to the target molecules is exposed to the ambient containing analyte, the selective molecular interactions create a differential surface stress between the top and bottom surfaces of the microcantilever. This differential surface stress results in a change in resistance of the piezoresis- tive layer. This can be represented using Equation 9.1 [20]. 1+R =−K ZT ⋅ZR ⋅s (9.1) i Eihi ∑ ∑R Z2 +1 hi 2 3ic 2 i Eihi where K is the gauge factor, σS is the surface stress, ZT is the position of top layer, ZR is the position of piezoresistive layer and Ei, hi and Zic are the Young’s modulus, height and position of the ith layer with respect to neutral axis. Based on this expression, the surface stress sensitivity is proportional to the ratio of the gauge factor (K) of the piezoresistive film to the Young’s modulus (E) of the cantilever structural material. As mentioned earlier, usage of compliant polymers such as SU-8 having a low Young’s modulus E (~40 times lower compared to silicon based materials) as cantilever struc- tural materials in place of conventional silicon based materials is expected to improve surface stress sensitivity. Piezoresistive SU-8 microcantilevers with different with integrated gold (Au) [10] and polysilicon strain sensitive layers have been reported earlier [20, 24]. Gold being a metal strain gauge material that has got very low gauge factor, low temperature deposited polysilicon [24] was a better sub- stitute for gold for improving the sensitivity. However the design constraint was that the polysilicon film should be thin enough such that it does not

310 Advanced Biomaterials and Biodevices add to the stiffness of the structure while ensuring that the thin polysilicon does not lead to a decreased signal to noise ratio. Piezoresistive materials with high gauge factor similar to polysilicon and Young’s modulus similar to SU-8 would be the best option. So the main goal of improvement in sensitivity without compromising on the process complexity and thereby ensuring cost effectiveness can only be accomplished by a proper choice of piezoresistive thin films having similar mechanical characteristics and process compatibility with that of SU-8. Composites of polymers and con- ducting fillers such as carbon black (CB) and carbon nanotubes are known to exhibit piezoresistive behaviour as reported in [33, 34]. The suspension of different conducting nanoparticles in SU-8 was reported in [35]. Hence polymer composites based on SU-8 with conductive fillers could be a viable option for the embedded piezoresistive material in SU-8 microcantilevers. Hence, SU-8 microcantilevers with integrated piezoresistor made of nano- composite of SU-8 and Carbon Black have been developed. [23, 25–27]. In such a device, the strain in the microcantilever is measured by the change in resistance of the embedded piezoresistive layer containing conducting nanoparticles. When the microcantilever structure is deformed or strained during a sensing operation, the nanocomposite layer responds to the strain by increasing the distance (in the case of tensile strain) between the indi- vidual conductive fillers. This eventually disturbs the conducting network and leads to an increase in resistivity of the composite layer. These devices exhibited very good sensitivity and could overcome the limitations of other schemes discussed earlier. 9.2.1 Preparation and Characterization of SU-8/CB Nanocomposite The polymer nanocomposite was prepared by ultrasonic mixing of Carbon Black (CB), Conductex 7067 Ultra (Columbian Chemicals) in SU-8 and Microchem Nanothinner from. The mixing parameters were optimized to get a uniformly dispersed spin coatable polymer nanocomposite. The qual- ity of dispersion of CB in SU-8 was characterized using a Dynamic Light Scattering (DLS) system, BI-200SM from Brookhaven instruments. The samples were prepared with different sonication energies for DLS analy- sis and the size distributions of CB nanoparticles in these different sam- ples were analyzed. Based on these analysis sonication energy of 3 kJ was adopted for preparation of SU-8/CB nanocomposite [27]. Similar to SU-8, SU-8/CB nanocomposite were patterned using standard photolithograhy process with an increased UV dose. In addition to the regular process steps for patterning of SU-8, a post development ultrasonic cleaning of the

Polymer MEMS Sensors 311 4 UV Exposure with 3 Pre-bake appropriate dose Post-bake 5 2 Spin coat SU-8/CB Development in 6 nanocomposite SU-8 developer Dispersion of Sonicate in IPA to 1 CB in SU-8 remove CB residues 7 Sonication Probe Controller Sample Ice water bath Figure 9. 2 Microfabrication processing steps for SU-8/CB nanocomposite. Optical micrographs of the SU-8/CB nanocomposite pattern before and after final step of sonication are shown. patterned SU-8/CB samples had to be performed in order to remove the carbon residue which is otherwise seen to be adsorbed all over the sample. The overview of these process steps along with optical micrograph of the composite at intermediate steps is given in Figure 9.2. Nanocomposite samples were coated with 10 nm of gold for subsequent analysis in scan- ning electron microscope (SEM). The SEM micrograph also confirms the uniform dispersion of CB nanoparticles in a SU-8 polymer matrix. As prepared SU-8/CB nanocomposite thin films were mechani- cally characterized using nanoindentation technique with a Hysitron Triboscope. This characterization was essential to understand whether nanoparticle filler loading would potentially change the Young’s modu- lus and hardness of SU-8 nanocomposite. Samples with varying CB con- centrations were prepared on a silicon substrate. The technique is based on the analysis of elastic and plastic deformation in the material while driving an indenter into the material whose mechanical characterization has to be done. When the indenter is withdrawn from the material, the elastic deformation is recovered. Indentations were carried out using a Berkovich diamond indenter (Ei = 1141 GPa and νi= 0.07) with maxi- mum load (Pmax) varying from 80 μN to 600 μN. Maximum loads applied to the thin film samples were varied so that the indentation depth was

312 Advanced Biomaterials and Biodevices 700 (A) Pmax 8.5 (B) Young’s modulus 0.7 600 8.0 Hardness 0.6 0.5 500 7.5 0.4 0.3 Load (μN) 400 dP E (GPa) 7.0 H (GPa) s =dh 0.2 100 N Loading 0.1 240 300 200 N 6.5 300 N 200 400 N 440 N Unloading 100 500 N 6.0 600 N 5.5 0 hmax 80 10% of Thickness 0 50 100 150 200 250 120 160 200 Displacement (nm) hmax (nm) Figure 9.3 (A) Load versus indentation depth for different loads. Inset shows the scanned images of a set of Berkovich indents at the SU-8 surface (B) Young’s modulus and Hardness of SU-8 as a function of indentation depth[27]. limited to within 10% of film thickness. The lower limit of Pmax is decided in such a way that indentation depth is at least four times more than the roughness of the sample. Polymers are known to show their peculiar viscoelastic behavior while indenting and in order to minimize the effect of viscoelasticity high loading and unloading rates were used [36]. The viscoelastic behaviour if present would be indicated with the presence of a “nose” in the force curve due to the increase in penetration depth even during the unloading portion of the load curve. Load-indentation depth curves for different Pmax values are shown in figure 9.3 along with an array of Berkovich indents on the SU-8 surface. It can be noticed that viscoelastic behaviour is not seen in these load curves. Olivar Pharr method [37] was used to extract the reduced modulus, Er and hardness, H. Young’s modulus tends to increase for higher depth of indentation, whereas hardness does not show a significant variation with indenta- tion depths. It is widely accepted that, for depths of indentation greater than 10% of film thickness, substrate interaction effects are observed. This can be clearly seen in the figure, where the dotted line indicates 10% of film thickness. For hmax values beyond that, there is a sharp increase in the modulus values, which could be attributed to the effect of the stiffer substrate, silicon (E = 170 GPa). Hence a better analysis of indentation of compliant films on stiffer substrates using a modified King’s analysis was carried out. Modulus values of SU-8 nanocomposite films were found to increase with increasing CB filler loading. (Figure 9.4) This agrees reasonably well

Polymer MEMS Sensors 313 Young’s modulus (GPa) 10.0 Modulus from nanoindentation 9.5 Parallel mixing model 9.5 Guth’s theory 10 8.5 8.0 24 6 8 7.5 CB vol % 7.0 6.5 6.0 5.5 5.0 4.5 4.0 0 Figure 9.4 Young’s Modulus of SU-8/CB composite as a function of CB Vol %. Here all the samples were indented with Pmax = 450 mN [27]. with two existing theories for particulate filled polymer nanocomposites with spherical fillers [38, 39]. Electrical characterization of SU-8/CB nanocomposites was carried out to understand the conduction behaviour of the nanocomposite at different CB loading and find out the usable range of concentrations of CB in SU8. For this, SU-8/CB resistors of varying CB concentration with gold contacts were fabricated. Current voltage (I-V) characteris- tics of SU-8/CB resistors with different CB filler loadings were analysed. I-V characteristics of samples with lower concentration samples exhib- ited symmetric and a non-ohmic behaviour and the characteristics tend to become linear for samples with filler loading well above percolation threshold[25]. Figure shows IV characteristics for SU-8/CB composites with 4.9 CB vol. % and 9.4 CB vol. %. The I-V characteristics for differ- ent CB concentrations were found to fit well with theoretical model that explains the conduction behaviour of nanocomposites as tunnelling of electrons from one particle to next particle inside the polymer matrix. The tunnelling of electrons through the gaps separating the carbon aggregates can reasonably justify the observation of translation of non- ohmic to nearly ohmic (quasi-ohmic) behaviours in IV characteristics with increase in CB vol%[27]. The nanocomposite resistors were also characterized to investigate the variability in the resistance values. It was observed that the percentage

314 Advanced Biomaterials and Biodevices 4.0 × 10–8 SU-8/CB SU-8_CB_4.9 6.0 × 10–4 5 × 10–6 Mean value of resistance (A) Curve fit (Theoretical) (B) 80 2.0 × 10–8 SU-8_CB_9.4 3.0 × 10–4 4 × 106 70 % SD Current [A] 0.0 Current [A]3 × 106 60 Resistance [ohm] 50 –2.0 × 10–8 0.0 Variability [%] 40 –4.0 × 10–8 Curve fit details 2 × 106 30 –6 Equation y = A* (x^n)exp(–B/x) + y0 20 10 7.0 7.5 8.0 8.5 9.0 Adj. R-squar 0.99777 Standard Error –3.0 × 10–4 CB Vol [%] Constant Value y0 –9.93907E-11 6.06433E-11 1 × 106 A 2.50549E-10 8.38804E-13 B 0.09368 0.02139 n 3 0 –6.0 × 10–4 0 24 6 6.5 –4 –2 0 7.0 7.5 8.0 8.5 9.0 Concentration CB Vol% Voltage [V] Figure 9.5 (A) I-V characteristics for SU-8/CB composites (width = 120 um, Length = 30 um) with 4.9 CB vol.% and 9.4 CB vol.% along with the theoretical curve fit (B) Electrical characterization of nanocomposite resistors showing the variability in resistance as a function of CB concentration[27]. variability in resistance decreases with increase the CB vol. % and the vari- ability is < 30 % for samples with CB concentrations > 8 vol. %. 9.2.2 Design and Fabrication of Polymer Nanocomposite Cantilevers The designs for polymer nanocomposite microcantilevers were chosen in order to improve the surface stress sensitivity, common mode rejection and mechanical stability. Dimensions of microcantilevers were decided based on stiffness and resonance frequency of microcantilever. There are the two important mechanical properties of beams that decide the mechanical sensitivity and stability under external vibrations. The thicknesses of indi- vidual SU-8 layers were chosen in such a way that the piezoresistor layer lie away from neutral axis. However the microfabrication constraints for SU-8 were also considered. SU-8 cantilever die contains two sets of measure- ment and reference cantilevers. For the SU-8/CB nanocomposite piezoresistor, optimum carbon black filler loading in the range of 8 – 9 vol. % was chosen based on the following considerations formulated based on the analysis of characterization data of nanocomposite thin films: i. Low Young’s modulus: for this, the CB concentration is kept as low as possible; ii. UV patternability of the SU-8/CB nanocomposite with mini- mum possible edge roughness: the quality of patterns worsens for SU-8/CB samples for CB filler loadings > 9 vol.%.

Polymer MEMS Sensors 315 Figure 9.6 Fabrication process flow (1) First layer of SU-8 (2) Cr/Au for contacts (3) SU-8/ CB composite layer (4) encapsulating SU-8 (5) Thick SU-8 die base (6) Release of cantilever die from the substrate. iii. Good strain sensitivity: SU-8/CB composites with CB con- centrations just above the percolation threshold are expected to give the maximum strain sensitivity. iv. Ohmic conduction behaviour and minimum variability in conduction: the variability in SU-8/CB resistance values decreases with increase in CB concentration. The fabrication of this device structure involves the defining of indi- vidual layers on dummy substrate silicon, and the final release of the whole polymer structure from the silicon substrate by etching away the sacrificial layer. The detailed process flow schematic is shown in Figure 9.6. Starting substrate was a piranha cleaned silicon substrate with a 200 nm of silicon dioxide that was indented to be the released layer. A 500 nm of SU-8 was patterned to define the microcantilever structure and the die with contact vias. Subsequently gold (200 nm) contact wires and pads with chrome (10 nm) as the adhesion support layer were patterned. Piezoresistive layer of SU-8/CB nanocomposite film was patterned follow- ing the procedure explained in the previous section. This was followed by the patterning of SU-8 (1.6 um) which formed the encapsulation layer for the piezoresistor. The device patterns after SU-8/CB resistor and encapsula- tion lithography steps are shown in Figure 9.7(A). Final layer was for form- ing the SU-8 anchor die (~150 μm) for the cantilevers. The photograph of a part of processed silicon wafer after this final lithographic step is shown in figure 7(B). The release of polymer device chips from the dummy substrate was performed by wet etching of the silicon dioxide layer in buffered oxide

316 Advanced Biomaterials and Biodevices (A) (C) (B) (D) Figure 9.7 (A) SU-8/CB resistor patterns on “V” shaped and “U” shaped cantilever areas after lithography process 4 (B)Photograph of the processed wafer after the final lithographic step showing arrays of polymer devices attached to the dummy substrate just before the release(C) Arrays of polymer nanocomposite microcantilever device chips after release process (D) SEM image of one of the device chips containing 4 cantilevers. etch (BOE). The released device chips and SEM image of one of the fabri- cated devices are also shown in Figures 9.7(C) and 9.7(D). These polymer microcantilevers were about 3 μm thick and the SEM micrograph confirms the stress free nature of these free standing polymer nanocomposite struc- tures achieved through an optimization of baking parameters for individ- ual layers of SU-8 at different levels of lithography. 9.2.3 Characterization of Polymer Nanocomposite Cantilevers The spring constant of the SU-8 nanocomposite microcantilever was extracted using microcantilever beam bending technique using a nanoin- denter, which provides high load and displacement measurement sen- sitivity. Moreover the spring constant of the indenter (diamond) is very high in comparison to the microcantilever structure. So this is a direct and simple measurement technique, since one need not consider the case of two springs in series as done in spring constant measurement using standard AFM[40]. The nanoindenter was used to apply load to the tip of the microcantilever leading to the displacement of the microcantilever. Using the indenter software itself, the indenter was engaged with polymer nanocomposite microcantilever and the indentation was performed. The load and unload segments showing the load and displacement as a func- tion of time are shown in figure 9.8(A). The indentation was performed

Polymer MEMS Sensors 317 (A) Load and unload segments 1600 (B) Vib velocity Average spectrum 2.0 Force 1400 80 1.5 Displacement 60 Frequency 1.0 40 22.6 kHz 0.5 1200 20 Force [μN] Magnitude 0 Displacement [nm]1000 100.25 m/s Magnitude [μm/s] Load 800 Indenter 600 400 200 0 20 40 Frequency [kHz] –200 20 40 60 80 Time [seconds] Figure 9.8 (A) Load and unload segment of force and displacement characteristics of SU-8 nanocomposite microcantilever obtained from nanoindenter. Insets: (1) Schematic of measurement.(B) Resonance frequency plot from Laser Doppler Vibrometer. in displacement control mode. For reliable spring constant measurements, the peak load or displacement should be such that the cantilever does not undergo plastic deformation. From this, the load versus displacement curve was plotted and the spring constant, ‘k’ was extracted from the slope of this curve. Since the indenter tip was placed with an offset from the tip; the actual spring constant of the cantilever structure was calculated from the slope of the curve using equation. k = slope × (L − l) 3 (9.2) L Here L is the length of the cantilever structure, l is the distance from the tip of the cantilever where load is applied as indicated in the inset of the plot. The spring constant extracted from measurements was 0.44 N/m[27]. Experiments were conducted for resonance frequency measurements using a Polytec Laser Doppler Vibrometer (LDV). Microcantilevers were actuated at different frequencies and the laser beam from the LDV is directed at the cantilever surface and the vibration amplitude and fre- quency are extracted from the doppler shift of the laser beam frequency due to the motion of the cantilever (Figure 9.8B). Measured resonant fre- quency for “U” shaped cantilevers were is 22.6 kHz which is very close to the analytically calculated value (~ 20 kHz). The resonance frequency was well above the lower limit for external vibrational noise immunity. The piezoresistive microcantilevers were electromechanically character- ized to demonstrate the piezoresistive behaviour. This was performed by

318 Advanced Biomaterials and Biodevices 0.05 Experimental data Linear fit 0.04 (Slope = 1.16 × 10–3 [ m]–1) 0.03R/R 4.0 @ 0 m 0.02 0.01 Current [μA] 4.1 @ 10 m 0.00 @ 20 m –0.01 4.2 @ 30 m @ 40 m 4.3 4.4 4.5 4.6 1.90 1.92 1.94 1.96 1.98 2.0 Bias voltage [V] 0 10 20 30 40 Deflection (μm) Figure 9.9 Electromechanical characterization plot for polymer nanocomposite microcantilever. deflecting the tip of the microcantilever with a calibrated micromanipu- lator needle with simultaneous measurement of resistance using Keithley 4200 source measuring unit. ΔR/R for a polymer nanocomposite micro- cantilever plotted as a function of deflection as shown in Figure 9.9 indi- cates a deflection sensitivity of 1.1 ppm/nm. The extracted gauge factor was approximately 90. The surface stress sensitivity was calculated using Eq 9.1. The surface stress sensitivity is 7.6 × 10–3 [N/m] -1 which is greater than that of an optimized silicon microcantilever and one order of magnitude higher than that of polymer microcantilevers with Au as the strain gauge [20] reported earlier. 9.3 Organic CantiFET Polymer nanomechanical sensor platforms using SU-8 nanocomposite cantilevers discussed in the previous section find applications in low cost biochemical sensing owing to their high sensitivity. However, that the nano- composite materials realized using particle dispersion techniques might suffer from process variability issues. So a new transduction scheme was explored that addressed this aspect by introducing a strain sensitive poly- mer layer in a SU-8 nanomechanical cantilever sensor with reduced process complexity and process variability. Pentacene which is a well-studied organic

Polymer MEMS Sensors 319 semiconductor that is commonly used as a channel material in organic field effect transistors (OFET), is reported to exhibit good strain sensitivity [41, 42] and also has a Young’s modulus nearly matching with that of SU-8. The most popular deposition process for Pentacene is vacuum sublimation. This allows uniform deposition of pentacene layers and the deposition process is done at room temperature making it suitable for polymer substrates [43]. OFETs have a huge potential in realizing large-area, mechanically flexible, lightweight and low-cost devices and circuits such as paper-like displays, radio frequency identification tags, and large area sensors on flexible substrates. However, the effect of strain on the electrical behaviour of pentacene based OFETs has always been a concern while discussing the reliability of OFETs for flexible electronics applications and different device design ideas to overcome this effect is an on-going research for realizing ultra-flexible OFET devices and circuits [44–46]. There are a very few reports available in literature that inves- tigated the effect of bending induced strain on change in current of pentacene OFETs [43–47, 48]. This negative aspect of dependence of electrical behav- iour on bending induced strain in pentacene OFETs for flexible electronics could be utilized in a positive way in realizing sensor applications using pen- tacene. The idea was to integrate pentacene based OFETs as strain sensor in place of simple piezoresistor configuration for microcantilevers. This was highly beneficial as the resistivity of pentacene thin films were known be very high and the transistor configuration would support the arrays of sensors with built-in switching matrix using these integrated OFETs. The device concept is to embed a pentacene based OFET inside an SU-8 microcantilever as shown in Figure 9.10. This novel device named Encapsulation layer SourcMeicrocOHaringgthailn-ekivcGesraeGtsmaetrtiduecicoetnluedrcautlrciltcaoyrer Drain ION = I0 ION = I0 – I Nanomechanical motion I Figure 9.10 Concept of an ‘Organic CantiFET’ device [28].

320 Advanced Biomaterials and Biodevices as ‘Organic CantiFET’, can be thought of as a low cost polymer counter- part of a similar device in silicon, a MOSFET embedded microcantilever reported earlier [49, 50]. When SU-8 microcantilevers with embedded OFET undergo nanomechanical motion during sensing events, the strain sensitive organic semiconductor pentacene responds to this by changing its mobility and hence the drain current of the transistor. The change in drain current can be measured and recorded using appropriate signal con- ditioning circuitry in order to perform the sensing operation. A simple differential amplifier circuit with sensing and reference transistors could be used in a complete sensor configuration. Such an arrangement of organic CantiFET retains the advantages offered by Wheatstone bridge used for piezoresistive sensors. 9.3.1 Process Integration of Organic CantiFET The CantiFET device chip consisted of two CantiFET devices for dif- ferential current measurement scheme, in which one of the devices was considered as the measurement cantilever and other being the reference cantilever [28]. Arrays of such CantiFET device chips were fabricated. The transistor covered the whole length of cantilever so that the design is appropriate for nanomechanical cantilever sensors involving surface stress measurements. The thickness of the SU-8 gate dielectric has been chosen based on the preliminary study conducted on the current voltage (I-V) characteristics of back gate back contact (BGBC) pentacene OFETs using SU-8 as dielectric [28]. The process integration of pentacene OFET with solution processed polymer materials like SU-8 is a real challenge as pentacene is known to get degraded by organic solvents. For this, a novel process sequence is followed, in which, one develops all the required layers except the penta- cene layer for OFET on the SU-8 cantilever and then deposits pentacene layer after the releasing the device chips from the substrate to form the final organic cantiFET. This process has inherent advantages such as (1) this avoids the exposure of pentacene layer to organic solvents used in SU-8 lithography (2) obviates the need for patterning of pentacene layer which is also known to be a non-trivial problem. The schematic of this fabrication sequence is given in figure and process steps are detailed as below. The process started with an oxidized silicon wafer in which the silicon dioxide layer served as the sacrificial layer for Figure 9.11(C2)). Gate elec- trode and contact pads of the OFET were formed using gold (Au) with a

Polymer MEMS Sensors 321 Figure 9.11 Schematic of fabrication process for polymer CantiFET (C1) Sacrificial layer (C2)First layer of SU-8 defining the cantilever and contact vias.(C3) Au electrode patterning for gate of transistor (C4) SU-8 as gate dielectric (C5) Au electrodes defining the source and drain of OFET (C6) Thick SU-8 layer defined for anchor or chip of the cantiFET device (C7) Release of the CantiFET device from substrate and pentacene and encapsulation layer deposition (C8) final device structure showing cantilevers , source drain and gate contacts through contact vias. thin layer of chrome (Cr) as the adhesion layer (Figure 9.11(C3)). The gate dielectric of the OFET was defined using SU-8 layer with a thickness of 900 nm (Figure 9.11(C4)). This process was followed by the formation of source and drain electrodes and the contacts using Cr/Au (Figure 9.11(C5)). Then a thick (more than 100 μm) SU-8 layer was patterned to form the anchor or the device chip and the frames for holding arrays of CantiFET device chips (Figure 9.11(C6)). Subsequently, the polymer device chips were released from the silicon substrate using isotropic etching of silicon dioxide in a buffered hydro- fluoric acid. The final step in the fabrication of organic CantiFET was the deposition of pentacene on the released SU-8 nanomechanical device to complete the integration of an OFET. Pentacene deposition was done using thermal evaporation of purified pentacene (Sigma Aldrich) at a deposition rate of 1 Ǻ/second at a base vacuum of 2 × 10–6 mbar and sub- strate temperature of 65°C. Photograph of an array of released organic CantiFET chips along with the scanning electron micrographs (SEM) of one of the CantiFET devices is shown in figure 9.12. The pentacene OFET in the SU-8 cantilever has been encapsulated using a very thin (~ 15 -20 nm) silicon nitride layer deposited using a hot wire chemical vapour deposition (HWCVD) method. The silicon nitride encapsulation layer was kept thin enough so as not to increase the stiffness of the can- tilever structure.

322 Advanced Biomaterials and Biodevices Figure 9.12 (A) Photographs of released arrays of organic CantiFETs (B) SEM micrograph of the fabricated CantiFET device (C) Bottom and enlarged view of cantilever portion of the CantiFET from SEM showing the inter digitated source drain electrode configuration (Type 1 & 3 CantiFET) (D) Top and enlarged view of cantilever portion of the CantiFET from SEM [28]. 9.3.2 Characterization of Organic CantiFET The organic CantiFETs were electrically characterized inside a shielded, vibra- tion isolated probe station. The electrical measurements were performed at room temperature under ambient atmospheric conditions using semiconduc- tor parameter analyzer. Output and transfer characteristics for these devices were recorded. The extracted saturation field effect mobility (μ) and threshold voltage (VTH) were found to be 3.7 × 10–4 cm2/Vs and -11.5 V respectively. These integrated OTFTs exhibited low gate leakage and good switching characteris- tics (Gate current density =1.2 × 10–7 A/cm2 @ VGS=-40V and ION/IOFF = 2.2 × 103). The output characteristics of encapsulated Organic CantiFET exhibited a linear increase in the drain current at low drain bias (Figure 9.13(b)). This is a clear indication of the existence of good ohmic contact at the interface of source/drain electrodes and the organic semiconductor. At high drain biases, proper saturation of drain current was also observed. The mechanical characterization of CantiFET devices were performed by beam bending technique using a nanoindenter tool following the pro- cedure explained in the previous section for nanocomposite microcantile- vers. The spring constant obtained using this method is 0.4 N/m[28]. In order to verify the suitability of CantiFET devices for biochemical sensing applications, the fabricated devices were electromechanically char- acterized to extract the surface stress sensitivity. The CantiFET device was

Polymer MEMS Sensors 323 (A) Organic cantiFET: Transfer characteristics (B) Organic cantiFET: Output characteristics 8.0 × 10–9 VgS = –40V 10–8 IDS 1.2 × 10–4 6.0 × 10–9 VgS = –30V I 1/2 DS Drain current [A] I 1/2 DS Drain current [A] 10–9 8.0 × 10–5 4.0 × 10–9 VgS = –20V 10–10 4.0 × 10–5 10–11 –30 –20 –10 0.0 2.0 × 10–9 VgS = –10V 0 –40 Gate voltage [V] 0 VgS = –0V 0.0 –40 –30 –20 –10 Drain voltage [V] Figure 9.13 I-V characteristics of organic CantiFETs after silicon nitride encapsulation. (A) Transfer characteristics (B) Output characteristics [28]. subjected to different levels of compressive strain using calibrated micro- manipulators and the I-V characteristics of the device was recorded after each level of strain, ε. It was observed that these organic CantiFET devices exhibit a good strain sensitive behavior with the drain current increasing with compressive strain [28]. The percentage change in drain current, field effect mobility and threshold voltage were plotted as a function of change in the strain for the bias condition of VGS = VDS = -40 V (Figure 9.14). From this it can be observed that the transistor parameter that is strongly influenced by the strain is the field effect mobility and not the threshold voltage. This agrees well with the present understanding of hopping transport in penta- cene. The compressive strain is expected to reduce the hole hopping energy barrier due to the decrease in hopping distance [41]. The devices exhibited very high strain sensitivity (ΔI/I per unit strain) of the order of 103 .The sur- face stress sensitivity for organic CantiFET devices (ΔI/I in ppm per unit surface stress in mN/m) extracted from the electromechanical characteriza- tion was 401 ppm [mN/m]-1. The minimum detectable surface stress for CantiFET sensors were predicted by measuring the noise levels in these devices. A battery operated low noise trans-impedance preamplifier (Stanford Research 570) with gain varying from 10–3 to 10–12 A/V was used to provide the gate bias to the CantiFETs probed inside a shielded probe station. The output or the drain terminal was connected to another SR570 in order to provide drain bias and to measure and amplify noise levels in the drain current. A spectrum analyzer (SR 750) was used to record the noise power spectrum in frequencies ranging from 1Hz to a few kHz. The noise current level calculated for 1/f noise frequency range was 1.46 pA. The performance of CantiFET was compared with other relevant nano- mechanical sensors such as SU-8 microcantilevers with integrated gold as

324 Advanced Biomaterials and Biodevices I0(VGS = VDS = –40V) = 19.05 nA 250 3.0 × 10–8 ε = 1.5 × 10–4 200 2.0 × 10–8 ε = 3.1 × 10–4 [%] Change ε = 4.7 × 10–4 Drain current [A] ε = 6.2 × 10–4 150 1.0 × 10–8 100 0.0 –30 –20 –10 0 I/I [%] 50 –40 Gate voltage [V] 0 VTH/VTH [%] Exponential fit 0.02 0.04 0.06 0.08 0.10 Strain [%] Figure 9.14 Percentage change in drain current, saturation field effect mobility and threshold voltage are plotted as a function of percentage strain. Inset : Transfer characteristics of organic CantiFETs under different strain conditions. Inset shows the IDS-VDS characteristics (@VDS = -40 V) under identical strain conditions. strain gauge [20] and the MOSFET embedded silicon cantilevers [49]. The deflection sensitivity of organic CantiFET sensors was at least 50 times higher compared to the SU-8 microcantilevers with gold as a strain gauge and 15 times higher compared to that of MOSFET embedded silicon cantilevers. The extracted surface stress sensitivity value for organic CantiFET is at least three orders of magnitude higher in comparison to that of SU-8 microcantilevers with integrated Au strain gauge (0.3 ppm [mN/m]-1) and 50 times higher than that of SU-8 nanocomposite cantilevers. With the high surface stress sensitiv- ity and low noise levels it should be possible to detect surface stress values down to 0.2 mN/m. This performance can be attributed to the integration of OFET as a strain sensor that consists of a very thin (~50 nm) strain sensitive organic semiconductor layer pentacene in the channel region. This capability of detecting low surface stress makes the organic CantiFET a suitable candi- date for many applications in biochemical sensor developments. 9.4 Polymer Microcantilever Sensors with Embedded Al-doped ZnO Transistor Organic CantiFET devices exhibited very high sensitivity, in addition to other benefits of embedded transistor configuration for strain transduction.

Polymer MEMS Sensors 325 However, pertaining to the stability of the embedded organic FET, the best option could be to replace the organic semiconductor with a non-organic channel material having a good process compatibility with that of SU-8. So Al- doped zinc oxide (AZO) as a channel later for CantiFET applications have been explored [29]. The process of this device started with RCA clean wafer followed by ther- mal oxidation to grow a 900 nm silicon di-oxide as sacrificial layer. Next the encapsulation layer and contact via of the cantilever die was fabricated using UV-exposure of SU-8 polymer layer. Using sputtering, Cr/Au layer of thickness 10nm/90nm was deposited and patterned for gate electrode and contact via. Another SU-8 layer of 900 nm was patterned for gate dielec- tric. In the next process step, source/drain layer was fabricated as described for gate electrode. A 150μm thick anchor was patterned using SU-8 2100. The cantilever die was released from the wafer by sacrificial oxide etching using 5:1 BHF. The semiconducting layer of Al-doped ZnO was deposited on the backside of the cantilever using sputtering. The process flow is given in Figure 9.15 A. The optical image is shown in Figure 9.15 B. The output characteristics for the ZnO TFT embedded CantiFET device were recorded for gate voltage in the range of 0–40V (Figure 9.16a). The Figure 9.15 A. Fabrication process for AZO integrated SU-8 microcantilevers (a) Sacrificial layer (b) SU-8 2002 encapsulation layer (c) Contact Pad and Gate electrode pattering using Cr/Au (d) Gate dielectric pattering (e) Source/Drain contact of Cr/Au (f) Thick SU8 pattering as anchor layer (g) Release of the device from the silicon wafer. B. Optical image of ZnO TFT embedded cantilever[29].

326 Advanced Biomaterials and Biodevices 1.2 × 10–9 Vgs = 40V 1.6 × 10–9 Vds = 40V 1.0 × 10–9 1.4 × 10–9 Drain current (A) 8.0 × 10–10 Vgs = 35V Drain current (A) 1.2 × 10–9 0 10 20 30 6.0 × 10–10 1.0 × 10–9 Gate voltage (V) 4.0 × 10–10 Vgs = 30V 8.0 × 10–10 2.0 × 10–10 Vgs = 25V 6.0 × 10–10 0.0 4.0 × 10–10 Vgs = 20V 2.0 × 10–10 VVVgggsss = 15V = 10V 0.0 = 5V Vgs = 0V 0 5 10 15 20 25 30 35 40 40 (a) Drain voltage (V) (b) Figure 9.16 (a) Drain current versus drain voltage for different gate voltages (b) Drain current versus gate voltage plot of AZO thin film transistor embedded on a cantilever, with SU-8 as a gate dielectric [29]. Drain current (A) 8.0 × 10–9 Vgs = 40V 1.0 × 10–8 Drain current (A) ε = 0 Vds = 40V 7.5 × 10–9 8.0 × 10–9 7.0 × 10–9 ε=0 6.0 × 10–9 ε = 7 ×10–2 6.5 × 10–9 ε = 7 ×10–2 4.0 × 10–9 ε = 14 ×10–2 6.0 × 10–9 ε = 14 ×10–2 2.0 × 10–9 ε = 21 ×10–2 5.5 × 10–9 ε = 21 ×10–2 ε = 28 ×10–2 5.0 × 10–9 ε = 28 ×10–2 0.0 ε = 35 ×10–2 4.5 × 10–9 ε = 35 ×10–2 4.0 × 10–9 (b) 0 10 20 30 40 3.5 × 10–9 Drain voltage (V) 3.0 × 10–9 2.5 × 10–9 2.0 × 10–9 1.5 × 10–9 1.0 × 10–9 5.0 × 10–9 0.0 –5.0 × 10–10 0 5 10 15 20 25 30 (a) Drain voltage (V) Figure 9.17 (a) Drain current versus drain voltage for different strain values for AZO thin film transistor embedded on a cantilever (b) Drain current plotted as a function of gate voltage for different strain values. transfer characteristics of the ZnO CantiFET device were recorded at a drain voltage of 40V (Figure 9.16b). The electromechanical characterization was performed by inducing stress on the cantilever beam using micromanipulator and for different bending values, output characteristics (Figure 9.17a) and transfer charac- teristics (Figure 9.17b) were recorded. The change in current can be attrib- uted to the combination of piezoresistive and piezoelectric effects. The strain generated in the semi-conducting film leads to a change in the energy spacing. This is expected to change the effective mass of the charge carriers causing a change in mobility. Therefore, the change in drain current in the CantiFET devices, when subjected to strain, is mainly caused by the change

Polymer MEMS Sensors 327 in mobility. Because of strain generated in the semi-conducting film, energy spacing decreases, which in turn helps to decrease the effective mass of the charge carriers. Effective mass of the charge carriers is inversely propor- tional to mobility. The change in mobility therefore will give rise to a change in the current with mechanical deflection. Also the piezoelectric property of AZO can affect the sensitivity. When the cantilever gets bent by an applied force, because of the piezoelectric effect, a potential gets generated which will help to reduce the surface potential barrier allowing additional carrier injection into the channel, thereby increasing the drain current sensitivity. 9.5 Piezoelectric Nanocomposite (SU-8/ZNO) Thin Films Studies and Their Integration with Piezoelectric MEMS Devices The field of nanocomposite materials has been attracting researchers due to the flexibility in their design, low cost, ease of fabrication and tunable physi- cal and chemical properties [27, 51–54]. The key drive for development of nanocomposites is the need for combining desirable properties of different materials. For example, one may wish to enhance the piezoelectric sensitivity of nanocomposite thin film but simultaneously also want to integrate these thin films with micro scale fabrication technology. Thus, for many applica- tions, one might need two conflicting desirable properties and such require- ments can only be achieved with combinations of useful properties of two or more materials. Based on the concept of nanocomposite formulation, there has been significant interest in research of lithographically patterned nano- composite polymer such as SU-8 [25–27, 35, 56]. It is commonly known as a negative photoresist, which is used as a structural material for realizing the MEMS structures. The advantage of SU-8 is its low Young’s modulus and photo patterning properties which makes it an obvious candidate to be used as a polymer matrix for nanocomposite realization [27, 55]. Over the last few decades, people have demonstrated SU-8 based nanocomposite with different physical and chemical properties by incorporating various inorganic fillers such as carbon black nanoparticles, silver nano particles, carbon nanotubes and magnetic nanoparticles [25–27, 30–32]. Concept of piezoelectric com- posite was first proposed by Newman who presented various models of con- nectivity of ceramics in polymer matrix for explaining piezoelectricity origin in such type of systems [51–52]. There have been attempts in the past to real- ize the piezoelectric composite materials by utilizing a polymer matrix with piezoelectric ceramics inclusion [53]. Recently, piezoelectric nanocomposite

328 Advanced Biomaterials and Biodevices with the inorganic filler such has BaTiO3, ZnO nanoparticles was proposed and tested for their piezoelectric properties [30–32, 57–58]. This opened up many interesting applications of these piezoelectric nanocomposite materials in piezoelectric MEMS devices. Due to limited lifespan of batteries and their biocompatibility issues, today there is a need of development of self-powered devices. Energy harvesting from ambient vibrations by piezo-electric MEMS microstructures can be an attractive option for self-powering sensor net- works [59–60]. We have explored a novel SU-8/ZnO polymer composite route which enables a simple and low cost fabrication of piezoelectric MEMS devices. Integrating ZnO into a photosensitive SU-8 polymer matrix not only retains the highly desired piezoelectric properties from the ZnO, but also combines the photo-patternability and optical transparency of the SU-8 poly- mer. The applications for these nanocomposite thin films were demonstrated with fabrication of piezoelectric MEMS cantilever device which can harvest mechanical vibrations and convert them to electrical signals. 9.5.1 Fabrication and Mechanical Characterization SU-8/ ZnO piezoelectric nanocomposite was formulated by mixing com- mercially available ZnO nanoparticles of dimensions 50–100 nm into the photosensitive SU-8 polymer (Microchem) matrix. In order to make uni- form dispersion of ZnO nanoparticles, these particles were dispersed in 1 ml of Cycolpentanone (SU-8 Thinner, Microchem) and were subsequently probe sonicated for 10 minutes. Thereafter, 1 ml of SU8–2002 was added to the dispersed solution of ZnO nanoparticles and was again sonicated for 25 mins. To compensate the heat generated during the sonication process, ice bath was used. Thin films of nanocomposite with various ZnO wt % ranging from 5–20 were prepared by spinning process. These films were then spin coated on small pieces of silicon and quartz at 2000 rpm for 30 sec and prebaked at 70 °C for 3–4 min and at 90 °C for 4–7 min. The sam- ples were then UV exposed, to activate the photo-polymerisation reactions followed by post baking cycle which is similar to prebake for crosslinking of nanocomposite. Thin films of SU-8/ZnO were investigated with various materials characterizations techniques to evaluate their mechanical, piezo- electric and photopatternabilty [32]. 9.5.1.1 Optimization of UV Dose and Transmission Characteristic of SU-8/ZnO Nanocomposite Thin Films It was observed that with the increasing wt % of inorganic filler in the SU-8, its physical properties changes [54]. For SU-8 based nanocomposite

Polymer MEMS Sensors 329 Irradiation dose (mJcm–2) 600 (a) 500 Transmittance (%)400 100 300 80 60 40 SU-8 (b) 20 SU-8/ZnO (15%) SU-8/ZnO (20%) 15 Wt%–480 mJcm2 0 400 600 800 Wavelength (nm) 200 100 0 5 10 15 20 Wt % of ZnO Figure 9.18 Optimized irradiation dose required for photo-patterning of the nano- composite films as a function of various ZnO filler concentration. Inset diagrams (a) Transmission characteristics of the nanocompositethin films. (b) SEM image of the patterned structures with optimized 15 wt % ZnO filler concentration [32]. thin films, photopatternabilty is the key issue that needs to be optimized in terms of inorganic filler concentration and requires UV dose. It was observed that SU-8 /ZnO nanocomposite retains their photopatternabilty properties up to 15 wt % of ZnO nanoparticles and then subsequently losses this property [32]. The optimized UV dose for various W % of ZnO nanoparticles are shown in Figure 9.18. The loss of photopatternabilty prop- erties can be explained with their transmission characteristics as shown in inset Figure  9.18a. It could be seen from transmission characteristics that transmittance of SU-8, which is originally more than 80% decreases drastically with increasing ZnO nanoparticles content and reaches almost 0% as particles concentration reaches about 20 wt %. The loss of litho- graphic patterning properties of nanocomposite beyond 20 Wt % can be due to high content of ZnO nanoparticles , which absorbs the major frac- tions of UV dose and inhibits the cation photo polymerization reaction in SU-8 [61–62]. 9.5.1.2 Piezoelectric Response of Nanocomposite Thin Films The other focus of research in SU-8 / ZnO nanocomposite thin films is to evaluate their piezoelectric response for their applications in realization of piezoelectric devices. Piezoforce microscopy (PFM) was used for probing

Amplitude RMS (pm)330 Advanced Biomaterials and Biodevices 200 150 100 50 0 0 2 4 6 8 10 12 Voltage RMS (V) Figure 9.19 Piezoelectric measurements of SU8/ZnO nanocomposite thin films on the surface showing almost linear relation between the PR amplitude and drive amplitude [32]. the piezoelectric activity of embedded ZnO particles in nanocomposite thin films. A linear relationship was observed between the PR amplitude and applied PR drive which varied between 2 to 10 V. For recording the PFM images the alternate current voltage with modulated frequency 14.2 KhZ was applied to PtIr5 coated cantilever tip. Resonance effect [63] was mini- mized by keeping the modulated frequency lower than the cantilever reso- nance frequency (66.67 kHz). The effective piezoelectric coefficient d33 was observed in range of 15–23 pm/V which is higher as compared to bulk [32, 64–65]. The high piezoelectricity of the nanocomposite thin films could be due to small size and shape of ZnO nanoparticles [64–67]. Recently, some studies showed that as the size goes down in nanoscle dimension the surface to volume ratio goes high which significantly impacts the atomic polariza- tions of surface atoms [65]. The other contribution in enhancement of piezo- electric properties is the shape anisotropic nature of nanoparticles [66–67]. 9.5.1.3 Mechanical Properties of Nanocomposite Thin Films: The advantage of SU-8 for realization of MEMS structures over the Silicon material is its low young modulus and low temperature processing [27, 54, 68]. It was observed that the mechanical properties of SU-8 changes with incorporation of inorganic filler such as CB and ZnO [27, 32]. For real- ization of piezoelectric cantilever MEMS devices with SU-8/ZnO nano- composite, stiffness is a very important parameter; hence the mechanical properties of SU-8/ZnO thin films were studied. The young modulus of SU-8 and SU-8/ZnO thin films was calculated with the formula repre- sented in Equation 9.3[27, 69–70]:

Polymer MEMS Sensors 331 Young modulus (GPa) 20 18 16 5 10 15 20 14 Wt% of ZnO 12 10 8 6 0 Figure 9.20 Young’s modulus measurements of nanocomposite thin films with different ZnO nanoparticles concentration [32]. 1 = 1 − 2 8 + 1 − 2 (9.3) Er SU indenter ES2U 8 Ei2ndenter The results of calculated young modulus of various wt % are shown in Figure 9.20. The calculated young modulus of SU-8 was observed to be 7–8 GPa initially and it increases to 14 Gpa as the ZnO filler concentra- tion reaches 15 wt % [32]. This enhancement in young modulus can be explained with the Guth-Smallwood equation. (Eq. 9.4): [27, 70] Ym = Yme (1 + 2.5VFiller + 14.1VF2iller) (9.4) The above equation shows the exponential enhancements in young modu- lus with respect to filler concentrations. The other factors which may also influence the mechanical properties of nanocomposite are baking temper- ature, dispersion, particles size and morphology [56]. 9.5.2 Fabrication of Polymer (SU-8) Piezoelectric (ZnO) Composite MEMS Cantilevers The complete process flow for fabrication of SU-8/ZnO nanocompos- ite based MEMS cantilevers is shown in Figure 9.21(a). The cantilevers were fabricated using a flip-chip approach involving six mask levels [31]. Silicon wafers were first piranha cleaned and then a 500 nm thick sacrifi- cial layer of silicon dioxide was grown by wet thermal oxidation. A layer of SU-8 2002 was spin coated to obtain a thickness of 1.8 μm, and then

332 Advanced Biomaterials and Biodevices (c) (a) First SU-8 layer Bottom Cr/Au contact Sacrificial layer (d) Thick SU-8 pad Top Cr/Au contact SU-8/ZnO nanocomposite layer (b) SU-8 layer (1.8 m) Cr/Au layer (90 nm) Removal of sacrificial layer Released cantilever die (flip chip) SU-8/ZnO nanocomposite layer Cr/Au layer (90 nm) (2.7 m) Thick SU-8 pad (200 m) Figure 9.21 (a) Schematic illustration of the process for fabricating flexible SU-8/ZnO cantilevers. (b) Schematic of cantilever beam showing various layers and their thicknesses (c) SEM image of SU-8/ZnO nanocomposite cantilevers. (d) Magnified optical image of SU-8/ZnO nanocomposite cantilevers showing Au electrodes [31]. it was lithographically patterned using the regular SU-8 processing steps. This layer acts as contact via layer after release of the cantilevers and also to insulate the top gold layer of the stack. A stack of Cr/Au was then deposited for a total thickness of 90 nm and patterned for contact pads. Then a well dispersed SU-8/ZnO nanocomposite was spin coated to get a thickness of 2.7 μm. This composite layer is patterned by regular SU-8 processing steps. After this step, the deposition and patterning of second electrode layer of Cr/Au was performed. The frame and device die were patterned using SU-8 2100. The frames were finally released from the substrate by etching the sacrificial layer in buffered HF acid solution. The floating SU-8 frames in the HF solution were slowly transferred in to DI water and the Cr layer on the gold contact pads is removed using Cr etchant. The schematic of cantile- ver beam showing various layers used with their thicknesses is represented in figure 9.21(b). The SEM and optical images of fabricated cantilevers are shown in figure(c) and figure (d) respectively. From SEM image, it is noted that the width of cantilever beam is 50 μm and the length is 250 μm. 9.5.3 Characterization of SU-8/ZnO Cantilevers as Vibration Sensors: For an oscillating multilayer cantilever, the resonant frequency is expressed as follows [75]: ( ( ) )f 2 1 EI (9.5) L2 nphnp + php =n 2

Polymer MEMS Sensors 333 where E is Young’s modulus, I is the moment of inertia, h is the thick- ness, w is the width, L is the length, is the density and the subscripts np and p denote the non-piezolayers and SU-8/ZnO piezo layer respec- tively. E—I is given by, ( )1 = 12 En2phn4p + E 2 h 4 Enphnp + E php + 6hnphp + 4hp2) (9.6) w p p + Enp E phnphp (4hn2p EI The elastic modulus of SU-8/ZnO nanocomposite which was calculated using the Hysitron Triboscope is 14 GPa. The fabricated cantilevers were tested for piezoelectric transduction using a mechanical vibrator. Figure 9.22(a) shows the experimental setup used for testing piezoelectric cantilevers. The cantilever was mounted on a shaker which acts as a vibration source to the cantilever. The shaker was Oscilloscope DUT Function Generator Shaker (a) 60 160 30 120 Voltage (mV) VP-P(mV) 0 80 –30 40 –60 1.0 1.5 2.0 2.5 3.0 3.5 4.0 0 234 5 6 Frequency (kHz) (b) Time (ms) (c) Figure 9.22 (a) Schematic of experimental setup (b) The observed oscilloscope traces of output voltage of the piezoelectric nanocomposite based MEMS cantilever during periodic bending and unbending at resonance. (c) The generated peak-peak voltage as function of frequency of vibration.

334 Advanced Biomaterials and Biodevices driven with a function generator with sinusoidal waveform. An oscillo- scope was used to measure the output signal of the piezoelectric canti- lever. When no vibration is set, the measured noise level is 8 mV. At 4.3 kHz vibration, which is much higher than the oscillation frequency of shaker, the cantilever provides the maximum AC peak-peak (Vpeak–peak) output voltage of about 146 mV. The voltage pulses are generated due to induced strains caused by vibration of the cantilever beam. The result- ing oscilloscope trace obtained at this vibration of cantilever is shown in Figure 9.22b. Figure 9.22c shows that the generated voltage pulses vary in sequence with frequency of vibration or induced surface strain. The resonant frequency of SU-8/ZnO cantilever calculated using equation 5 is 4.2 kHz which is very close to the frequency (4.3 kHz) at which maxi- mum signal is observed. The SU-8/ZnO composite in cantilever beam experiences a time- varying change in mechanical stress, alternating between both compres- sive as well as tensile stress resulting in a time-varying generated charge within this nanocomposite layer. Therefore, the cantilever acts as an AC generator while mechanically vibrating. The AC output power calculated using equation (9.7) across 100kΩ load resistor (R) is about 0.025μW: Vpeak− peak 2 Power = 2 2 (9.7) R 9.6 Polymer Nanomechanical Cantilever Sensors for Detection of Explosives The world has seen deadly terrorist attacks over past decade because of explo- sive based weapons, whose potential for damage is known to be tremendous. Most of these explosives are usually very difficult to detect as they have very low vapour pressures. In order to be able to detect the presence of these explo- sive to any degree of success, a detector should be able to perceive 1ppb of explosive in ambient air. Hence building detectors that can smell explosives similar to sniffer dogs has been a challenge that the scientific community has been grappling with over a decade now. In addition to detection technolo- gies with very high sensitivity, there exist demands for explosive sensors with minimum production and deployment costs especially since cost effective technologies can globally revolutionize the battle against terrorist attacks.

Polymer MEMS Sensors 335 The growing technology problem around the globe has prompted the researchers to come up with novel and nifty solutions to quickly detect concealed explosives. Several decades of research that went into the devel- opment of sensors of explosives and hazardous gases helped the research- ers to identify different popular detection techniques such as ion mobility spectrometry (IMS), mass spectrometry etc. However these conventional explosive detection techniques tend to be expensive and bulky besides hav- ing a longer response time. However, in order to efficiently detect explosives at strategic locations and public spots, deployment of multiple sensors is necessary; hence the recent demand is for development of extremely sensi- tive and cost effective sensors that can be mass produced and networked. Gravimetric sensors such as SAW sensors and QCM sensors are not small enough to be deployed as arrays and these sensing technologies require frequency measuring systems that are known to be expensive and large in size. Hence MEMS systems such as microcantilevers have been found to be suitable for explosive detection because of their advantages such as small size, high sensitivity, low power consumption and versatility to integrate multiple explosive detectors in a single miniature package. Different techniques for the detection of explosives in vapour phase using microcantilevers have been reported in the literature [72–76] which includes receptor based detection and receptor free detection. In recep- tor based detection, microcantilevers are functionalized with receptors for molecular recognition and the functionalization procedure is based on the receptor and microcantilever surface chemistry. The molecular recogni- tion mechanisms in these cases are based on the weak interactions that can be reversed making the microcantilever sensors reversible. As per litera- ture, microcantilever that have been used for these applications are mostly silicon or derived materials such as silicon dioxide and silicon nitride. Polymer microcantilever sensor platform being a better sensing platform in terms of sensitivity compared to conventional silicon based microcan- tilever platform would be a good candidate for development for explosive detectors. The polymer nanocomposite microcantilevers were functionalized using a receptor coating, 4-MBA (4-mercaptobenzoic acid). They form hydrogen bond with nitroaromatic based explosive molecules such as TNT. [72]. 4-MBA can easily form a stable monolayer on gold surface through thiol chemistry. Therefore one side of the microcantilever was coated with 30 nm of gold with a 5–7 nm of titanium as the adhesion layer in order to facilitate for a selective functionalization of 4-MBA. The functional- ized microcantilevers along with non-functionalized microcantilevers as reference microcantilevers were then attached to a PCB with respective

336 Advanced Biomaterials and Biodevices (A) SU-8 nanocomposite NMC’s response to TNT vapours (B) 14 ppb 0 30 ppb 46 ppb Output [mV]–20 TNT introduced Output signal (mV)110 N2 OFF (B) TNT ON TNT ON 105 100 95 –40 90 85 80 TNT OFF TNT OFF N2 ON 75 N2 ON Reversibility of the sensor 0 200 400 600 800 1000 Time (seconds) –60 0 30 60 90 Time [seconds] Figure 9.23 (A) Response of a 4-MBA coated polymer nanocomposite microcantilever to different concentrations of TNT vapour in nitrogen. Inset: Response of these devices for alternate cycles of TNT and Nitrogen (B) Explosive detector prototype developed at IIT Bombay based on a piezo-resistive polymer nanocomposite cantilever platform [27]. contacts. The microcantilevers were kept inside a PTFE gas flow cell [27]. The reference and measurement devices connected in bridge circuit were connected to a standard signal conditioning circuit for continuous moni- toring of the microcantilever response. Nitrogen purging was performed before starting the explosive vapour exposure experiment. This helped in bringing down the humidity levels. A calibrated vapour generator containing TNT source was used. Gas flow of 30 SCCM was maintained using a mass flow controller (MFC). The calibrated vapour generator was capable of delivering TNT vapours with concentrations < 10 ppb. Microcantilever response for alternating cycles of TNT (~30 ppb) and nitrogen for three minutes and five minutes respec- tively was recorded as shown in figure 9.23 (A)[26]. It can be observed that the polymer nanocomposite cantilever sensor is reusable even after multiple exposure cycles. The responses of these microcantilevers to dif- ferent concentrations of TNT vapours were also recorded. The output volt- age was found to increase with increase in concentration (figure 9.23(A)). Considering the noise levels, these devices should be able to detect TNT vapours down to a few ppb (~ 6 ppb) concentrations with an approximate sensitivity value of 1 mV/ppb [27]. The hand-held explosive detector pro- totypes developed based on these polymer nanomechanical cantilever sen- sor platform is shown in figure 9.23(B). These detectors could detect TNT and RDX in normal ambient. The analysis presented here for the explosive vapour exposure experiments conducted in controlled environment. However, the experiments conducted

Polymer MEMS Sensors 337 using microcantilevers functionalized with 4-MBA for detection of TNT and RDX in ambient conditions could support their candidature for detection of explosive vapours in ambient. Moreover, development of some orthogonal approaches that could be employed along with these receptor based micro- cantilever explosive detection schemes is also an on-going research. References 1. J. W Gardner and V.K. Varadan, Osama O. Awadelkarim, Microsensors, MEMS and Smart devices, John Wiley & Sons, 2002. 2. Bhushan, Bharat. Springer handbook of nanotechnology. Springer, 2010. 3. S.P. Mohanty and E. Kougianos, “Biosensors : A Tutorial Review,” IEEE Potentials, 2006, pp. 1–16. 4. King W P et.al., “Design of atomic force microscope cantilevers for com- bined thermomechanical writing and thermal reading in array operation,” J. Microelectromech. Syst., vol. 11, p. 765–74. 5. Berger, R., Lang, H. P., Gerber, C., Gimzewski, J. K., Fabian, J. H., Scandella, L., ... & Güntherodt, H. J. (1998). Micromechanical thermogravimetry. Chemical Physics Letters, 294(4), 363–369 6. S.-hyung S. Lim, D. Raorane, S. Satyanarayana, and A. Majumdar, “Nano- chemo-mechanical sensor array platform for high-throughput chemical analysis,” Sensors And Actuators, vol. 119, 2006, pp. 466–474. 7. Y. Arntz, J.D. Seelig, H.P. Lang, J. Zhang, P. Hunziker, J.P. Ramseyer, E. Meyer, M. Hegner, and C. Gerber, “Label-free protein assay based on a nanome- chanical cantilever array,” vol. 14, 2003, pp. 86–90. 8. L.G. Carrascosa and M. Moreno, “Nanomechanical biosensors : a new sens- ing tool,” Trends Anal. Chem., vol. 25, 2005, p. 196–206. 9. C. Ziegler, “Cantilever-based biosensors,” Analytical and Bioanalytical Chemistry, vol. 379, 2004, pp. 946–959. 10. J. Fritz, M.K. Baller, H.P. Lang, and H. Rothuizen, “Translating Biomolecular Recognition into Nanomechanics,” Science, vol. 288, 2000, pp. 316–318. 11. J.W. Gardner, JehudaYinon, ed., Electronic Noses & Sensors for the Detection of Explosives, NATO Science series, . 12. H.-peter Lang, M. Hegner, and C. Gerber, “Nanomechanical Cantilever Array Sensors,” Handbook of Nanotechnology, Springer, , pp. 443–459. 13. M.A. and B.A. Greve A, Keller S, Vig A L, Kristensen A, Larsson D, Yvind K, Hvam J M, Cerruti M, “Thermoplastic microcantilevers fabricated by nano- imprint lithography,” J. Micromech. Microeng., vol. 20, 2010, p. 015009. 14. L.Y. and X.Y. Katragadda R, Wang Z, Khalid W, “Parylene cantilevers inte- grated with polycrystalline silicon piezoresistors for surface stress sensing,” Appl. Phys. Lett., vol. 91, 2007, p. 083505. 15. A. Boisen, S. Dohn, S.S. Keller, S. Schmid, and M. Tenje, “Cantilever-like micromechanical sensors,” Rep. Prog. Phys., vol. 74, 2011, pp. 1–30.

338 Advanced Biomaterials and Biodevices 16. J.M. Shaw, J.D. Gelorme, N.C. LaBianca, W.E. Conley, and S.J. Holmes, “Negative photoresists for optical lithography,” IBM Journal of Research and Development, vol. 41, 1997, pp. 81–94. 17. J. Brugger, P. Renaud, and P. Vettiger, with M. Despont and H. Lorenz, N.Fahrni, “High aspect ratio ultrathick, negative-tone near-UV photoresist for MEMS applications,” IEEE MEMS, 1997, pp. 518–522. 18. H. Lorenz, M. Despont, M. Fahrni, N. LaBianca, P. Vettiger, and P. Renaud, “SU-8: a low-cost negative resist for MEMS,” J. Micromech. Microeng, vol. 7, 1997, pp. 121–124. 19. Genolet G, Brugger J, Despont M, Drechsler U, Vettiger P, “Soft, entirely photoplastic probes for scanning force microscopy,” Rev. Sci. Instrum., vol. 70, 1999, p. 2398. 20. J. Theysen, A.D. Yalcinkaya, P. Vettiger, and A. Menon, “Polymer-based stress sensor with integrated readout,” Journal of Physics D: Applied Physics, vol. 35, 2002, pp. 2698–2703. 21. J.H.T. Ransley, M. Watari, D. Sukumaran, and R.A. McKendry, A.A. Seshia, “SU8 bio-chemical sensor microarrays,” Microelectronic Engineering, vol. 83, 2006, pp. 1621–1625. 22. M. Nordström, S. Keller, M. Lillemose, A. Johansson, S. Dohn, D. Haefliger, G. Blagoi, M. Havsteen-jakobsen, and A. Boisen, “SU-8 Cantilevers for Bio/ chemical Sensing; Fabrication, Characterisation and Development of Novel Read-out Methods,” Sensors, 2008, pp. 1595–1612. 23. L. Gammelgaard, P.A. Rasmussen, M. Calleja, P. Vettiger, and A. Boisen, “Microfabricated photoplastic cantilever with integrated photoplastic/carbon based piezoresistive strain sensor,” Applied Physics Letters, vol. 88, 2006, p. 113508. 24. N. S. Kale, S. Nag, R. Pinto, V. R. Rao,”Fabrication and Characterization of a Polymeric Microcantilever with an Encapsulated Hotwire CVD Polysilicon Piezoresistor”, J. Microelectromech. Syst.,vol.18, 2009. 25. V. Seena, A. Rajorya, P. Pant, S. Mukherji and V. R. Rao., “Polymer microcan- tilever biochemical sensors with integrated polymer composites for electri- cal detection”, Solid State Sci., vol.11, pp.1606–1611, 2009. 26. V. Seena, A.Rajorya, A.Fernandes,P.Pant, S.Mukherji, V.R.Rao, “Fabrication and characterization of novel polymer composite microcantilever sensors for explosive detection ”, Technical Digest, IEEE 23rd Int. Conf. MEMS, pp. 851 – 854,Hong Kong, Jan24–28, 2010. 27. V.Seena, A.Fernandes, P.Pant, S.Mukherji and V.R.Rao “Polymer nanocom- posite nanomechanical cantilever sensors: material characterization, device development and application in explosive vapour detection”, Nanotechnology, Vol 22, p. 295501, 2011 28. V.Seena, A. Nigam, P. Pant, S Mukherjee and V. Ramgopal Rao, IEEE/ASME Journal of Microelectromechanical Systems vol 21, pp294–301,2012 29. Prasenjit Ray and V. Ramgopal Rao, Appl. Phys. Lett. 102, 064101 (2013). 30. K. Prashanthi, H. Zhang, V. R. Rao, T. Thundat, Phys. Status Solidi RRL, “Local piezoelectric response of ZnO nanoparticles embedded in a photosensitive polymer,”.Vol.6, pp.77–79, (2012).

Polymer MEMS Sensors 339 31. K. Prashanthi, M. Naresh, V. Seena, T. Thundat, V. R. Rao, “A novel photo- plastic piezoelectric nanocomposite for MEMS applications,” J MEMS Lett., Vol.21, pp.259 -261,(2012). 32. M. Kandpal, C. Sharan, P. Poddar, K. Prashanthi, P. R. Apte, V. R. Rao, “Photopatternable nano-composite (SU-8/ZnO) thin films for piezo-electric applications,”. Appl. Phys.Lett., Vol. 101, pp.104102–5 (2012). 33. X.-wu Zhang, Y.I. Pan, Q. Zheng, and X.-su Yi, “Time Dependence of Piezoresistance for the Conductor-,” Polymer, 2000, pp. 2739–2749. 34. M.H.G. Wichmann, S.T. Buschhorn, J. Gehrmann, and K. Schulte, “Piezoresistive response of epoxy composites with carbon nanoparticles under tensile load,” vol. c, 2009, pp. 1–8. 35. H.C. Chiamori, J.W. Brown, E.V. Adhiprakasha, and E.T. Hantsoo, “Suspension of nanoparticles in SU-8 : Processing and characterization of nanocomposite polymers,” Microelectronics Journal, vol. 39, 2008, pp. 228–236. 36. D. Tranchida, S. Piccarolo, J. Loos, A. Alexeev, D. Tranchida, S. Piccarolo, J. Loos, and A. Alexeev, “Mechanical Characterization of Polymers on a Nanometer Scale through Nanoindentation . A Study on Pile-up and Viscoelasticity,” Macromolecules, vol. 40, 2007, pp. 1259–1267. 37. W. C. Oliver and G. M. Pharr,” Measurement of hardness and elastic mod- ulus by instrumented indentation: Advances in understanding and refine- ments to methodology”, J. Mat. Res., vol.19, pp. 3–20, 2004. 38. J. C. Grunlan, D. Rowenhorst, L. F. Francis and W. W. Gerberich ,” Modulus Determination of Polymer Matrix Composites: Comparison of Nanoindentation and Dynamic Mechanical Analysis” Mat. Res. Soc. Symp. vol. 649 39. N. Abdel-Aal, Farid El-Tantawy, A. Al-Hajry and M. Bououdina , “Epoxy Resin/ Plasticized Carbon Black Composites.Part II. Correlation Among Network Structure and Mechanical Properties” Polymer Composites, pp. 804 – 808, 2008. 40. C. Serre, P. Gorostiza, A. Prez-rodriguez, F. Sanz, and J.R. Morante, “Measurement of micromechanical properties of polysilicon microstruc- tures with an atomic force microscope,” Sensors and Actuators, vol. 67, 1998, pp. 215–219. 41. T. Sekitani, Y. Kato, S. Iba, H. Shinaoka, and T. Someya, “Bending experiment on pentacene field-effect transistors on plastic films” Appl. Phys. Letters., vol. 86, p.073511, 2005. 42. T. Ji, S. Jung, and V.K. Varadan, “ Field-Controllable Flexible Strain Sensors Using Pentacene Semiconductors”, IEEE Electron Device Lett., vol.28, 2007. 43. M. Kitamura and Y. Arakawa, “Pentacene-based organic field-effect transis- tors”, J. Phys.: Condens. Matter, vol. 20, pp. 184011, 2008. 44. T. Sekitani, U. Zschieschang, H. Klauk, and T. Someya, “Flexible organic transistors and circuits with extreme bending stability,” Nature Materials, vol. 9, 2010, p. 1015. 45. A. Jedaa and M. Halik, “Toward strain resistant flexible organic thin film transistors,” Applied Physics Letters, vol. 95, 2009, p. 103309.