Pressure Sensors for Printed Blast Dosimeters Jurgen Daniel +, Tse Nga Ng, Sean Garner and Ana Claudia Arias Palo Alto Research Center (PARC) Palo Alto, CA 94304, USA + Daniel@parc.com John Coleman, Jianzhong Liu and Ronald Jackson * Naval Medical Center San Diego San Diego, CA, USA * Ronald.Jackson@med.navy.mil Abstract Disposable printed sensor tapes are being developed to record the magnitude of explosive blasts in the battlefield. The goal is to detect and mitigate the possible occurrence of traumatic brain injury in soldiers. This paper presents results on the pressure sensors to measure the blast pressure and the blast noise. The sensors are based on piezoelectric polymers and their fabrication is compatible with roll-to-roll fabrication methods to enable low cost. I. INTRODUCTION Printing technologies are promising for the fabrication of large-area and low-cost electronics devices, particularly when combined with roll-to-roll (R2R) processes. We have previously used inkjet printing to fabricate active-matrix transistor backplanes for reflective displays such as electronic paper [1,2]. Here, we apply a related technology towards printed sensor systems. Our goal was to use technologies that are compatible with flexible substrates to fabricate disposable blast dosimeter tapes that can be attached to a soldier s helmet. Blast events due to improvised explosive devices (IEDs) are a major cause of traumatic brain injury (TBI) in the battlefield [3]. The tape-like dosimeter is designed to measure and record the force and the number of blast events during one week in order to enable early administration of medical care. The disposable tape would be replaced with a new one after reading out the stored data. The goal was to implement measurements of blast pressure, blast noise, acceleration and light on the sensor tape together with printed circuitry to process and store the sensor data. An integrated flexible battery would function as the power source. This paper focuses on the pressure sensors to record the extent of the blast pressure wave and the acoustic pressure (or sound pressure) of an explosion. Printed amplifier [4], data acquisition and memory circuitry [5] was developed in parallel to this work. II. SENSOR DESIGN The blast dosimeter was designed so that pressure and acoustic sensors would cover the ranges of 5-100 psi and 100-180 db (equivalent to 0.00029-5.2 psi). Both types of sensors Figure 1. Printed blast pressure and acoustic sensors connected to conventional charge amplifiers and mounted on rigid aluminum holders for test experiments. have a similar design, with the acoustic sensors being slightly larger due to the lower pressure range. Sets of ten blast pressure and ten acoustic sensors were fabricated, calibrated and tested. Figure 1 shows the sensors attached to rigid aluminum holders for test experiments. A sensing concept based on piezoelectric polymers was chosen for the pressure and acoustic sensors because of attributes such as low power consumption, flexibility, reduced sensor drift and compatibility with solution or web processing. Particularly the need for low power operation favored piezoelectric sensing. Also the reduced sensitivity to static strain, which may occur in a flexible device, was an important factor [6]. While piezoelectric sensors cannot measure absolute pressure changes, they are well suited for short-term transients such as blast overpressure. The design of the pressure sensors is based on a suspended sensor membrane over a cavity. Deflection of the membrane due to a pressure change causes strain in the piezo-polymer film and therefore induces a charge which can be measured as
a voltage across the film. This voltage V is dependent on the film capacitance and the input impedance of the readout electronics. In the ideal case of infinitely high input impedance, the voltage is given by V= g31 in-plane stress film thickness, (1) where g31 is the piezoelectric coefficient [7]. Piezoelectric materials are also sensitive to temperature changes because of the pyroelectric effect. Therefore, the sensors were designed with an added thermal mass over the piezoelectric film to delay and weaken the temperature signal. This is particularly important for measuring blast events where the blast wind can cause temperature changes. For both types of sensors, a rugged design was essential to survive the maximum targeted pressure of 100 psi. Moreover, the blast pressure sensors had to be fast enough to capture the peak pressure of a blast wave. The developed blast pressure sensors had a calculated resonance frequency around 21 khz which enables them to capture pressure peaks with rise-times well below 1 ms. III. FABRICATION Fabrication methods that would allow R2R processing, including printing, lamination, die cutting and laser machining were chosen to enable inexpensive manufacturing. Due to the different sensitivity ranges for blast pressure and acoustic sensors, two different fabrication approaches were chosen (Fig. 2). One is based on the lamination of a piezoelectric foil, the other one on a solution process to deposit the piezo material. Although eventually only one method may be selected, this path also enabled us to evaluate the advantages and disadvantages of either process. A. Solution Process The pressure sensors were fabricated by depositing a polyvinylidene fluoride - trifluoroethylene (PVDF-TrFE) copolymer film from solution onto a sensor foil. The polymer was spin-coated to a thickness between 1 and 5 µm and slotcoating or inkjet printing may be used in a R2R process. The polymer was then heated to form polar beta-phase crystallites and subsequently oriented by corona charge poling. A top electrode was patterned via inkjet printing of conductive silver ink or PEDOT:PSS conductive polymer [8,9]. For prototyping purposes, some of the top contacts were also deposited by sputtering of Pd/ Au through a shadow mask. The sensor foil was then aligned and laminated to a spacer foil in which a 6 mm diameter hole was formed. This bilayer foil is laminated to a substrate. The top electrode and the stainless steel bottom electrode are connected to the input of an amplifier circuit with silver ink. An elastomer layer is formed on the sensor membrane as thermal mass to reduce the sensor temperature sensitivity. By choosing an elastomeric material, the increase of the mechanical stiffness of the sensor membrane is minimized. Figure 2. Fabrication of printed blast pressure and acoustic sensors using a combination of lamination and printing methods. All of these techniques are compatible with roll-to-roll processing. B. Lamination Process The acoustic sensors were fabricated by using a commercial PVDF piezoelectric foil. A foil with a thin metal layer on both sides was laminated onto a spacer foil in which a 12 mm diameter hole was formed. This double-layer structure was then laminated onto a substrate. The top electrode for the PVDF film was patterned via laser ablation or by spark erosion. After connecting the electrodes to an amplifier circuit with silver ink, an elastomer layer was formed on the top surface of the sensor foil as thermal mass. C. Process Comparison Both fabrication processes have advantages and disadvantages. Solution processing would be easier to integrate into a R2R process because the alignment of subsequent layers is simpler than in a lamination process. However, the annealing time to form the beta-phase in the piezoelectric copolymer can be long, making it problematic for a R2R process. Promising in this respect are some newer varieties of piezoelectric copolymers which require shorter annealing times. The polarization of the material can be optimized in a separate process when the lamination process is used. D. Sensor Readout Electronics To test the sensors, a high impedance operational amplifier was attached to a flexible printed circuit patterned from Pyralux (Dupont) material. The copper traces were formed by jet-printing wax as an etch mask and then etching the unprotected areas [10]. The sensors and amplifier circuits were bonded to aluminum holders to provide rigid support during sensor testing. The connection between the sensors and the amplifier circuit was established with conductive silver ink. Sets of ten sensors were fabricated and tested to show the
repeatability of the fabrication process. In addition, one blast pressure sensor was combined with a printed inverter circuit. IV. MEASUREMENTS AND RESULTS The sensors were tested and calibrated with custom-built equipment to provide the required acoustic and blast pressure ranges. In all the measurements, commercial reference sensors were placed as close as possible to the printed sensors. In verification experiments, the sensors were examined for deviations from the calibration. A. Thermal Sensitivity As described earlier, it was necessary to add a thermal mass in form of an elastomer to the front surface of the piezoelectric sensor membrane in order to reduce the response to thermal transients (pyroelectric response). The beneficial effect is illustrated in Fig. 3. An illuminated optical fiber was used as a heat source and passed twice across the printed sensor and a piezoresistive reference sensor next to it. Without elastomer layer, the printed sensor showed a pronounced pyroelectric signal. It was much reduced after adding the elastomer layer to the sensor. The piezoresistive reference sensor also shows some temperature sensitivity which serves as a measure for the magnitude of the heat pulse. B. Blast Pressure Sensor Testing and Calibration The pressure sensors were calibrated using a pressure chamber with a fast acting valve. Printed sensor and piezoresistive reference sensor were mounted in a small chamber near the exit port of the valve. With this setup, pressure pulses with a rise time around 5 ms and a length of ~10 ms were achieved over a pressure range of ~1-100 psi. The pressure pulses were very reproducible. A calibration curve for each of the ten sensors was established by applying ten pressure pulses near each of the values of 5, 10, 15, 25, 50 and 100 psi. The pulse profiles were displayed on an oscilloscope and the peak values were recorded. A second order curve fit was used to establish the parameters for the calibration curves. A subset of four sensors was then selected by an independent observer and tested against the calibration using similar pressure pulses over a range of pressure values. Fig. 4a shows a plot of the calibration curve and the data points from the verification. In Fig. 4b, the error of the sensor reading is plotted. Within the specified pressure range of 5-100psi, the verification values were well within the allowed +/-10% error interval. At around 1psi, which was outside the specified pressure range, the error was larger. By modifying the sensor geometry, this error could be reduced. The performance of the pressure sensors in a simulated blast pressure wave was tested with a blast tube. With a blast tube, fast pressure pulses in the range of a millisecond can be generated by pressurizing a chamber with a membrane window. Upon bursting of the membrane, a pressure wave is directed towards the sensor. Compared to a mechanical valve, significantly faster pressure pulses can be generated with a a) Figure 3. Reduced temperature sensitivity of a printed piezoelectric blast pressure sensor by adding an elastomer layer as thermal mass. A piezoresistive reference sensor is mounted in close proximity. In a) and, the sensor output voltage is shown for the printed and the reference sensor when passing a heat source over the sensors. The temperature sensitivity of the printed sensor is significantly reduced with the elastomer layer attached. In, the heat pulses were larger as noticed by the larger signal of the reference sensor. larger volume of displaced air. However, the shape of the pressure pulse is more complex. Printed sensors were mounted in front of the blast tube with a piezoelectric reference sensor positioned as close as possible. Rigid mounting of the sensors was essential to avoid false signals due to vibration caused by the blast wave. In Fig. 5, the blast pressure curves of the printed and the reference sensor are shown for a blast pressure pulse in the 13-17 psi range. The pressure value was extracted from the highest peak using the calibration of the sensors. Although not exactly the same, the shape of both curves is very similar. C. Acoustic Sensor Testing and Calibration Calibration and verification of the acoustic sensors occurred in an acoustic chamber with a 500 Hz sinusoidal signal from a speaker facing the printed sensor and reference sensor. Multiple measurement series were taken at various sound pressure levels using a commercial piezoelectric acoustic sensor as the reference to determine the sound pressure level. A calibration curve was established for each printed sensor from the first measurement series by using a linear curve fit for the data of sound pressure level (db) versus output voltage of the printed sensor. The data points for the second and third measurement series were then plotted next to the calibration curve as shown in Fig. 6. Two sensors were exposed to blast over pressure (BOP) pulses from a blast tube
a) voltage low, a thin polymer gate dielectric was employed. Before connecting the circuit to the sensor, the operation of the inverter was verified by applying an ac signal to its input and monitoring the output with a high impedance oscilloscope probe. Fig. 7(a) shows a 100Hz input signal and the inverted output with ~0.8 times amplification. The input of the inverter circuit was then electrically connected to the pressure sensor with silver ink. Several pressure pulses of varying magnitude were applied to the sensor membrane. The sensor signal was recorded via the inverter output and is plotted in Fig. 7( against the pressure in the pressure chamber which provides the pressure pulses. Some of the data point scatter was due to the difference between the pressure reading of the pressure chamber and the actual pressure pulse value on the sensor membrane. In this experiment a more accurate pressure reading from a reference sensor next to the printed sensor was not available. Figure 4. Calibration and verification of a printed blast pressure sensor. The measurement error was determined by comparing the pressure values of the printed sensor obtained via the calibration curve with the pressure values of a commercial reference sensor. Pressure pulses with a ~5ms rise time were applied to the sensors. The measurement error for this sensor was well within ~+/-5% for the range of ~5psi to ~90psi. in order to test their ruggedness. The applied pressure pulses were ~90 psi and ~170 psi. Afterwards, theses sensors were measured again to check for a potential permanent drift of the sensor signal and the data points were plotted against the calibration curve. The measurement error was calculated as the deviation of the measured values from the calibration curve. For all measurements from ~100-140 db, the error was well below +/-2%. The BOP pulses did not change the sensor response indicating that the sensors have a robust design. a) D. Pressure Sensor with Printed Amplifier One ultimate goal in the development of the blast dosimeter tape is to fabricate the electronic readout circuit using printing techniques. As a first step towards this, a printed blast pressure sensor was coupled to a printed inverter circuit. A unipolar inverter design was used with two p-type thin-film transistors. The circuit was fabricated by inkjet printing using silver nanoparticle ink for the conductors and organic semiconductor ink for the transistor channel. A similar fabrication process has been reported for complementary inverters [4]. In order to keep the operating Figure 5. Blast pressure curves of a printed blast pressure sensor (a) and a commercial reference sensor (. Both sensors were placed next to each other in front of a blast tube to expose them as close as possible to the same blast pressure. The peak pressure values, extracted from the sensor calibration, were for both sensors in the 13-17 psi range and the shape of the curves for this blast event is rather similar.
a) Figure 6. Calibration and verification of a printed acoustic sensor. In order to test its robustness, the sensor was exposed to a blast over pressure (BOP) of ~90psi and ~170psi and then re-measured. The deviation from the calibration curve was well below +/-2%. V. CONCLUSION Pressure sensors were fabricated, calibrated and tested in ranges capable of measuring blast overpressure and sound pressure. The fabrication approaches used are compatible with roll-to-roll processing. The tested sensors showed that the desired pressure ranges can be reliably measured within the specifications. ACKNOWLEDGMENT The authors would like to thank Robert Reuss and George Duchak for advice and helpful discussions. We also like to acknowledge help and advice by the following people at PARC: Chris Paulson, Brent Krusor, Greg Anderson and Jeng Ping Lu. We thank Measurement Specialties, Inc. for providing a sample of PVDF copolymer and Dupont for samples of Pyralux flexible circuit material. This research was partially funded by DARPA, under contract number W81XWH-09-C-0065. Approved for public release, distribution unlimited. REFERENCES [1] J.H. Daniel, et al., Flexible Electrophoretic Displays with Jet-Printed Backplanes, SID 09 Digest, 44.3, 2009, pp.660-663 [2] A.C. Arias, et al., All-additive inkjet-printed display backplanes: Materials development and integration, J. SID, 15/7, 2007, pp.485-490. Figure 7. Printed blast pressure sensor with a printed inverter circuit. In a) the input and output signal of the inverter circuit alone is shown at a frequency of 100Hz. At the chosen voltage settings, the amplification factor of this circuit was ~0.8. In, the inverter and the sensor are connected. The graph shows the sensor voltage measured at the inverter output as a function of the applied pressure. The pressure is applied in form of a pulse originating from a pressurized chamber. [3] E. Singer, Brain Trauma in Iraq, Technology Review, May/June 2008, pp.53 59. [4] T. Ng, et al., Electrical stability of inkjet-patterned organic complementary inverters measured in ambient conditions, App. Phys. Lett., 94, (2009) 233307. [5] T.N. Ng, et al., Degradation mechanisms of organic ferroelectric fieldeffect transistors used as nonvolatile memory, J. Appl. Phys. 106, 094504 (2009) [6] R.H. Brown, The Piezo Solution for Vital Signs Monitoring, Medical Design Technology, 2008, 36-40 [7] P. Ueberschlag, PVDF piezoelectric polymer, Sensor Review, Vol 21, No.2, 2001, pp. 118-125. [8] D. Kim, et al., Direct writing of silver conductive patterns: Improvement of film morphology and conductance by controlling solvent compositions, Appl. Phys. Lett., 89, 264101 (2006) [9] H. Sirringhaus, et al., High-resolution inkjet printing of all-polymer transistor circuits, Science, Vol 290 (2000)2123-2126 [10] W.S. Wong, et al., Amorphous silicon thin-film transistors and arrays fabricated by jet-printing, Appl. Phys. Lett, 80 (2002) 610-612