Home Search Collections Journals About Contact us My IOPscience Polysilicon-based flexible temperature sensor for brain monitoring with high spatial resolution This content has been downloaded from IOPscience Please scroll down to see the full text 2017 J Micromech Microeng 27 025001 (http://iopscience.iop.org/0960-1317/27/2/025001) View the table of contents for this issue, or go to the journal homepage for more Download details: IP Address: 130.133.8.114 This content was downloaded on 25/01/2017 at 11:26 Please note that terms and conditions apply You may also be interested in: Evaluation of microelectrode materials for direct-current electrocorticography Chunyan Li, Raj K Narayan, Pei-Ming Wu et al Chemicapacitors as a versatile platform for miniature gas and vapor sensors Robert Blue and Deepak Uttamchandani Progress towards biocompatible intracortical microelectrodes for neural interfacing applications Mehdi Jorfi, John L Skousen, Christoph Weder et al Micromachine-based humidity sensors with integrated temperature sensors for signal drift compensation Chia-Yen Lee and Gwo-Bin Lee A 100 m diameter capacitive pressure sensor with 50MPa dynamic range Xin Luo and Yogesh B Gianchandani Cantilever-like micromechanical sensors Anja Boisen, Søren Dohn, Stephan Sylvest Keller et al SU-8-based microneedles for in vitro neural applications Ane Altuna, Gemma Gabriel, Liset Menéndez de la Prida et al Design, simulation and experimental validation of a novel flexible neural probe for deep brain stimulation and multichannel recording Hsin-Yi Lai, Lun-De Liao, Chin-Teng Lin et al Bidirectional micromachined flow sensor featuring a hot film made of amorphous germanium Samir Cerimovic, Almir Talic, Roman Beigelbeck et al Journal of Micromechanics and Microengineering J Micromech Microeng 27 (2017) 025001 (8pp) doi:10.1088/1361-6439/aa4e99 Polysilicon-based flexible temperature sensor for brain monitoring with high spatial resolution Zhizhen Wu1,4, Chunyan Li2, Jed Hartings3, Sthitodhi Ghosh1, Raj Narayan2 and Chong Ahn1 Microsystem and BioMEMS Laboratory, University of Cincinnati, Cincinnati, USA Cushing Neuromonitoring Laboratory, Feinstein Institute for Medical Research, Manhasset, USA Department of Neurosurgery, University of Cincinnati, Cincinnati, USA E-mail: wuzn@mail.uc.edu Received 28 July 2016, revised 10 November 2016 Accepted for publication 18 November 2016 Published December 2016 Abstract Temperature is one of the most important variables in brain monitoring, since changes of focal brain temperature are closely coupled to cerebral physiology and pathophysiological phenomena in injured brain In this work, a highly accurate temperature sensor with polysilicon thermistors has been developed on flexible polyimide for monitoring brain temperature with high spatial resolution The temperature sensors have a response time of 1.5 s and sensitivity of −0.0031 °C−1 Thermal hysteresis of the sensor in the physiological temperature range of 30–45 °C was found to be less than 0.1 °C With silicon nitride as the passivation layer, the temperature sensor exhibits drift of less than 0.3 °C for d in water In vivo tests of the sensor show a low noise level of 0.025 ± 0.03 °C, and the expected transient increases in cortical temperature associated with cortical spreading depolarization The temperature sensor developed in this work is suitable for monitoring brain temperature with the desired high sensitivity and resolution Keywords: polysilicon thin film, flexible temperature microsensor, brain temperature monitoring (Some figures may appear in colour only in the online journal) 1. Introduction Despite these studies, a systematic investigation of temperature variation across the injured human brain is still needed An accu rate and high spatial resolution method to monitor the brain, characterizing temperature at the level of local microvasculature and functional cell groups, would aid in understanding patho logical processes developing in the injured brain Furthermore, it is known that the injured brain is extremely sensitive and vul nerable to small temperature variations Therefore, in order to improve temperature control and treatment of the injured brain using focal brain thermotherapy [7, 8], accurate measurement of the temperature profile is required Currently, ultrasound thermometry, microwave radiom etry and magnetic resonance thermometry [9–12] are usually used for non-invasive brain temperature monitoring However, these approaches always suffer from low temporal and spatial Brain temperature reflects the balance of neural metabolic heat production, cerebral blood flow and the temperature of incoming arterial blood [1–3] Studies have shown the temper ature of brain is not homogeneous In humans, the temperatures around the center of the brain are 0.5–1 °C higher than those of the epidural space [4] For the injured brain, the temperature variation throughout brain or between viable tissue and regions of ischemia become more evident Schwab et al [5] showed that temperature in the ventricles exceeded epidural temperature by up to 2.0 °C for patients with severe middle cerebral artery (MCA) infarction Marshall [6] also observed higher temper atures in ischemic brain regions compared to normal brain Author to whom any correspondence should be addressed 1361-6439/17/025001+8$33.00 © 2016 IOP Publishing Ltd Printed in the UK Z Wu et al J Micromech Microeng 27 (2017) 025001 resolution Advanced Proton chemical shift imaging (CSI) sequences in MR thermometry was shown to measure temper ature distributions with 1 min and a low spatial resolution of 3–4 mm [11] Furthermore, artifacts can be induced by dif ferent composition of tissue, tissue motion, and external filed drift [12] Thrippleton et al [13] has reported a discrepancy of 0.7 °C between the temperature measured with MR and the absolute temperature measured by a fiber-optic method Marshall et al [6] performed a MRSI validation study at 1.5 T, finding a standard deviation of 1.2 °C through the repeating measurements on individual voxels which is inaccurate for reliable brain activity monitoring In order to obtain accurate and real-time temperature infor mation, various implantable temperature sensors and probes have been developed using the principles of optical fiber [14], resistance temperature detector (RTD) [15], thermocouple [5, 16], and thermistor [17, 18] However, temperature arrays, which can directly measure the brain spatial temperature vari ation, are scarcely available and technologically challenging Optical fibers are difficult to be integrated into a single plat form, and multiple fibers/needles are needed for thermal mapping which will largely increase the damage caused by fiber insertion [19] Metal based RTDs measure the resistance changes with temperature with high accuracy and repeat ability and multiple RTDs can be easily integrated on one platform However, due to its large pattern size, the RTD lacks the capacity to measure temperature with high spatial resolu tion Thermocouple measures temperature-dependent voltage generated at the junction of two distinct metal wires due to thermoelectric effect, but the thin film based thermocouples always suffer from measurement error due to their low junc tion voltage [20] On the other hand, thermistors with semiconducting mat erials such as amorphous germanium [21], amorphous silicon [22], and vanadium oxide [23] have a large variation of resist ance and a high sensitivity to temperature change, and thus are a good candidate for a brain temperature sensing array with high spatial and temporal resolution Kuttner et al [21] developed amorphous germanium thermistor arrays on glass substrate with a sensitive area of 0.014 mm2 each and an inter distances of 0.4 mm for the investigation of biological temper ature fields Billard et al [23] developed a vanadium oxide thermistor array on glass for localized temperature field meas urements in brain The developed thermistor arrays show a high sensitivity of 0.2–0.4% °C−1 of temperature Nonetheless, both of the works developed thermistor arrays on rigid glass substrate with high mechanical stiffness Cerebral monitoring, however, requires development on a flexible platform that can be inserted into the brain with minimal tissue damage In this work, we have developed a polysilicon thermistor array on flexible polyimide substrate with structural flexibility for brain temperature monitoring Polysilicon thin film has been utilized as the sensing material for thermistors for a long time [24–27] The temperature coefficient of resistance (TCR) of polysilicon can be selected over a wide range, both positive and negative, through developing temperature and selective doping [26] To develop polysilicon on flexible polyimide substrate, polysilicon thin film in this work was formed using Figure 1. Concept and device design: polysilicon thermistors array with four sensing elements (R1 and R2: 20 àmì160 àm, R3 and R4: 200 àmì160 àm), ECoG microelectrodes are placed along for in vivo test of association between ECoG depolarization signal and local temperature change under stimulus aluminum induced crystallization (AIC) process as described in our previous work [28] Polysilicon thermistors have been developed with the dimensions of 20 àmì160 àm and 200 àmì160 àm In vitro tests of the sensor for the sensitivity, response time, thermal hysteresis, resolution, and long term stability have been successfully performed In addition, in vivo tests of the sensor have also been performed to determine noise levels and assess performance in measuring temperature responses to pathophysiologic processes 2. Sensor design and fabrication 2.1. Sensor design Four polysilicon thermistors and two electrocorticography (ECoG) electrodes are designed on a polyimide thin film as shown in figure 1, where the thin polyimide film was used as the flexible substrate To realize high spatial resolution temper ature sensing, two polysilicon thermistor elements with the dimension of 20 àmì160 àm were designed with the interdistance of 450 µm To compare performance, another two polysilicon thermistors with dimension of 200 àmì160 àm were also added The four polysilicon thermistors formed a sensing array for temperature measurement Two ECoG elec trodes with a radius of 25 µm were placed along the thermistor elements to measure associations between neural activity and local temperature The lateral distance between ECoG elec trodes and polysilicon resistors was set as 1 mm 2.2. Polysilicon developed on polyimide with AIC Polysilicon was developed on PI2611 polyimide substrate using aluminum induced crystallization (AIC) process as in our previous work [28] In the aluminum induced crystalliza tion (AIC) process, amorphous silicon and aluminum layers were deposited on the substrate and annealed at a temper ature under the eutectic temperature of 577 °C [29] During annealing, Si atoms dissolve into and diffuse within the alu minum layer The nucleation takes place at the grain boundary of aluminum crystals and terminates when adjacent grains meet, forming a continuous polycrystalline silicon film The presence of aluminum drastically lowers the activation energy Z Wu et al J Micromech Microeng 27 (2017) 025001 (a) (b) (c) (d) (e) (f) mechanical strength and stiffness to PI2611 substrate Silicon nitride film was RF sputtered on the polyimide film by Denton Discovery 24 Sputtering System with a silicon wafer as target and mixed gas of argon and nitrogen as the environmental and reactive gas Optimized sputtering conditions with the distance between target and substrate of inch, base pressure of 10−7 pa, sputtering power of 200 w, and nitrogen-to -argon ratio of 2:1 were used to obtain a dense silicon nitride film which can effectively block the moisture or oxygen The silicon nitride layer we sputtered is a thin layer with the thickness of 100 nm Sputtered silicon nitride film was etched in buffered HF to test the density The film can be considered as an effective pas sive layer only if the etching rate is less than 10 nm min−1 The etching rate in buffered HF of sputtered silicon nitride film in this work was less than 6 nm min−1 Figure 2. Summary of the microfabrication process (a) Develop PI2611 on silicon wafer, (b) develop polysilicon film and pattern thermistors, (c) devlop aluminum lead and ECoG electrodes, (d) develop another PI2611 layer and silicon nitride, (e) RIE etch to expose ECoG electrodes, (f) peel off PI2611 from silicon nitride at the backside 2.4. Microfabrication process A summary of the microfabrication process is described in figure 2 After obtaining the polysilicon film on a µm thick PI2611 substrate, polysilicon thin film was patterned to form a resistor with the standard photolithography process, followed by development of the electrical lead with 250 nm aluminum evaporation and etching 12 nm/120 nm Ti/Au film was then deposited and patterned to form the ECoG pattern Another µm thick PI2611 layer was coated on top of the fabricated metal and polysilicon patterns Then, a 100 nm thickness of silicon nitride layer was sputtered on the top of the PI2611 layer as the final passivation layer To expose the ECoG pattern and contact pads, dry etching with CF4 and O2 plasma was performed to etch away both PI2611 and the passivation layer of silicon nitride Finally the device was peeled off from the silicon wafer and a 100 nm thick silicon nitride layer was sputtered on the backside of the device So both front side and backside of the device were protected with the passivation layer In addition, with silicon nitride film at both sides of the device, the possible residual stress in silicon nitride film can be balanced The device was laser cut (Oxford lasers) into a desired size and bonded with connectors using silver paste Epoxy was applied to further strengthen the contact between the connector and the contact pads Pt nanoparticles (NP) were electrochemically deposited on ECoG electrodes with optimized roughness factor and iridium oxide was elec trodeposited to form heterostructured ECoG electrodes The impedance of ECoG electrodes was characterized by per forming electrochemical impedance spectroscopy (EIS) based on the procedures described in our previous paper [35] The average impedances of Pt/IrOx electrodes are 4.72 ± 1.06 kΩ (n = 6) at 1 kHz Figure 3 shows the fabricated device and pat tern under a microscope required for crystallization and the phenomenon can partly be explained by Hiraki’s screening model [30] To start the microfabrication process, PI2611 (HD Microsystem) was spun-on a silicon wafer and cured in a programmable oven at 400 °C for 30 min with a ramp rate of °C min−1 to obtain a µm PI2611 film Aluminum and amorphous silicon thin layers with equal thickness of 200 nm were then sputtered on the cured PI2611 film The sample was annealed at 400 °C with a ramping rate of °C min−1 for 1 h in nitrogen gas After completion of the annealing process, silicon and aluminum layers totally exchanged their positions and the newly formed polysilicon film was exposed by etching away the top aluminum film with aluminum etchant Thus, a very stable polysilicon film was obtained on the flexible poly imide substrate through the low processing temperature 2.3. Silicon nitride sputtering High water absorption or permeation as well as high oxygen permeation of polyimide can limit the performance of devices using polyimide as a substrate in water or air with high humidity [31], especially for BioMEMS applications in an aqueous environment (e.g brain) The water absorption and oxygen permeation can cause drift or instability of the sensing signal throughout the measurement duration Thus, an addi tional passivation layer is necessary to prevent or decrease the moisture and oxygen diffusion As the passivation layer, thin films such as aluminum oxide (Al2O3), silicon oxide (SiO2) or silicon nitride (SiNX) have been deposited on the polyimide substrate to reduce the diffusion of water or humidity [31–34] Here we used silicon nitride (SiNX) film with the thickness of 100 nm as the passivation layer The coefficient of thermal expansion (CTE) of silicon nitride thin film (3.3 ppm °C−1) is very close to that of polyimide substrate PI2611 (3 ppm °C−1), thus the stress induced by mismatch of thermal expansion under temperature change can be minimized A thin film with the thickness of 100 nm was chosen to minimize its effect of 3. In vitro measurement The developed flexible temperature sensor with polysilicon thermistor array was fully characterized in water for its sensi tivity, response time, thermal hysteresis, resolution, and long Z Wu et al J Micromech Microeng 27 (2017) 025001 Figure 4. Sensitivity test: the sensor demonstrates sensitivity of of −0.0031 °C−1 for polysilicon thermistor R1 and R2, and −0.0025 °C−1 for R3 and R4 Figure 3. Fabricated devices: (a) fabricated flexible devices and (b) sensor pattern under microscope term stability The temperature sensor was submersed (sub merging depth: 3 cm) in a water bath (1 l) on a hot plate (Cole Parmer Series 04644) A high accuracy biomedical temper ature probe (WPI® BAT-21) was attached to the temperature sensor and used to monitor the temperature in real-time The probe tip and the polysilicon thermistor were positioned at the same height during the test to ensure they were exposed to the same environment Polysilicon thermistor’s resistance was measured by a digital multimeter (Agilent 34461A) Resistance and temperature data were acquired using Labview 3.1. Sensitivity The sensitivity of the polysilicon thermistor was obtained by measuring the sensing element’s resistance variation while changing the environmental temperature Temperature ranging from room temperature (~21 °C)–60 °C was attained by increasing the hot plate’s temperature with a ramping rate of 60 °C h−1 Four polysilicon elements for each dimension were tested The resistance change with standard deviation (SD) is shown in figure Polysilicon thermistors’ resist ance decreases with temperature in a linear fashion For R1 and R2 with the dimension of 20 àmì160 àm and resist ance of 36.5 ± 0.3 kΩ, the sensitivity is −0.0031 °C−1 −4 (SD: 10 ), while for R3 and R4 with the dimension of 200 àmì160 àm and resistance of 3.7 ± 0.2 kΩ, the sensitivity is −0.0025 °C−1 (SD: 1.4 × 10−4) Though the sensitivity of the developed polysilicon is about ten-orders lower than other thermistor materials such as amorphous germanium and silicon, it is comparable with gold-based RTD [15] that we previously developed for the brain temperature sensing and is sensitive enough for brain monitoring Figure 5. Response time test: the time constant and response time are found to be around 0.8 s and 1.5 s for R1 and R2, and 0.8 s and 2.2 s for R3 and R4 3.2. Response time The temporal response of the polysilicon temperature sensor when exposed to instantaneous change in environment temperature is defined by two measures, the time constant and response time Time constant is the time to reach 63.2% of the complete step change in temperature Response time is the time to reach 99.5% of the final temperature in a step change The ability to track process changes depends on the sen sor’s thermal mass and proximity to the process The time response for the sensor was determined by moving the sensor from the ambient air (~25 °C) into a water bath at 40 °C The temperature range of 25–40 °C totally covered the physiologic temperature of brain Z Wu et al J Micromech Microeng 27 (2017) 025001 Figure shows the time course of the resistance change of the temperature sensor measured for three samples The time constant and response time are found to be around 0.8 and 1.5 s for R1 and R2, and 0.8 s and 2.2 s for R3 and R4 Since brain temperature changes on a longer time scale, the response time of the developed polysilicon thermistor is fast enough for brain temperature sensing 3.3. Thermal hysteresis Three cycles in the temperature range of 30–45 °C (which includes physiological temperature of brain) were applied to examine thermal hysteresis Figures 6(a) and (b) shows the relationship between temperature and resistance for R1 and R2, and R3 and R4 during the cycles, respectively As shown in the figure, the thermal hysteresis is less than 0.1 °C for polysilicon elements R1 and R2 after three cycles, while for R3 and R4, the thermal hyster esis reaches °C 3.4. Resolution Temperature steps of 0.1 °C were realized by a slow ramping rate of °C h−1 of the hot plate The resistance of the temper ature sensor was recorded when the reference temperature probe (BAT-21) showed an increase of 0.1 °C Recorded temperature and resistance are shown in figure 7 The sensor’s resistance changes for both R1 and R2 and R3 and R4 with each temperature step clearly show that the sensor can realize a resolution of 0.1 °C, thus meeting requirements for brain temperature monitoring 3.5. Long term stability Figure 6. Thermal cycle test: the thermal hysteresis is less than 0.1 °C for the temperature sensor for for R1 and R2 (a), and °C for R3 and R4 (b) after three thermal cycles from 30 °C to 45 °C Long term instability of polysilicon resistor can be induced by the grain boundaries in polysilicon film Grain boundaries con tain numerous dangling bonds, which can either trap a charge carrier or be saturated by an atom [36] Trapped carriers give rise to a potential barrier, which increases the resistivity [37] When tested in aqueous environment, hydrogen may dif fuse in to the polysilicon film and attach itself to the dangling bonds to form weak silicon–hydrogen bond, which easily get broken when the resistor is under thermal and electrical stress [38] The forming and breaking of silicon–hydrogen bonds may cause additional resistance drift In this work, dense sil icon nitride layer has been sputtered on polyimide substrate as the moisture and ion barrier To test the long term stability of the polysilicon thermistors, the developed devices were submersed in a water bath at room temperature and resistances were measured continuously for d Figure 8(a) shows the measured resistance of one poly silicon thermistor and temperature with time and figure 8(b) shows the resistance drift and corresponding temperature drift (with S.E.) over d measured for five samples (three with 20 àmì160 àm and two with 200 àmì160 àm) As shown, the resistance drifts of the polysilicon thermistors were less than 0.08%, corresponding to less than 0.3 °C This result Figure 7. Resolution test: the sensor’s resistance changes with each temperature step clearly shows that the sensor can realize a resolution of 0.1 °C Z Wu et al J Micromech Microeng 27 (2017) 025001 Figure 9. Measured in vivo noise level for polysilicon thermistors: R1 with the dimension of 20 àmì160 àm: 0.0250.03 C; R3 with the dimension of 200 àmì160 àm: 0.070.06 C Figure 8. Long term stability test: (a) shows the resistance of one polysilicon thermistor meausured for d; (b) shows the resistance drift in d measured for five samples The resistance drifts of the polysilicon thermistors were less than 0.08%, which corresponds to less than 0.3 °C Figure 10. Transient response of polysilicon thermistors to in vivo stimulus: brain cortical temperature showed a transient increase (0.35 ± 0.06 for R1 and R2; 0.23 ± 0.07 for R3 and R4; n = 4) and then slowly recovered to its baseline value during spreading depolarization suggests that the developed temperature sensor can achieve a high accuracy of 0.3 °C over d of continuous monitoring in an aqueous environment and is acceptable for brain moni toring applications 3 mm, ML 2 mm, DV −0.5 mm) Recordings began 24 h after device implantation To assess the polysilicon temperature per formance, a spreading depolarization (SD) wave was elicited by needle prick at the frontal craniotomy and the wave was recorded as it propagated across the sensor array All proce dures complied with the NIH guidelines for the care and use of laboratory animals and were approved by the Feinstein Institute for Medical Research Committee on Use and Care of Animals 4. In vivo measurement All surgical procedures were performed in the animal sur gical procedure room at the Feinstein Institute for Medical Research Two male Sprague–Dawley rats (250–400 g) were used in the experiments Animals were anesthetized with 5% isoflu rane and placed in a stereotaxic frame with a base heating pad Animals were maintained on 2.5% isoflurane delivered in room air through a nose cone throughout surgical procedures, and 1.5% isoflurane throughout recordings Two craniotomies were made in the right hemisphere One was made over coordinate at AP 0–6 mm, ML 1–3 mm to place the polysilicon temperature sensing strip on the dura The other was made at AP −2 mm, ML 1–2 mm to induce a spreading depolarization wave by a needle prick to the cortex A 250 µm diameter Ag/AgCl wire reference electrode was implanted in the left hemisphere (AP 4.1. In vivo noise level The in vivo noise levels of the polysilicon temperature sen sors were measured In order to prevent self-heating of the polysilicon elements, 10 µA and 40 µA constant current were applied to R1/R2 polysilicon elements with the resistance of 36.5 ± 0.3 kΩ and R3/R4 with the resistance of 4.0 ± 0.1 kΩ, respectively The experimental results are shown in figure 9 A total of polysilicon elements for each dimension (n = 4) were evaluated In summary, the R1/R2 polysilicon exhib ited less noise (0.025 ± 0.03 °C) than the R3/R4 element (0.07 ± 0.06 °C) by a factor of at least 2.8 Z Wu et al J Micromech Microeng 27 (2017) 025001 References Compared to the noise level of gold-based RTD (0.035 ± 0.04 °C) that we developed previously as a temper ature sensor, the polysilicon thermistors with the dimension of 20 àmì160 µm and larger resistance achieved a lower noise level and thus realized higher measurement accuracy [1] Childs C 2008 Human brain temperature: regulation, measurement and relationship with cerebral trauma: part Br J Neurosurg 22 486–96 [2] Busto R et al 1987 Small differences in intraischemic brain temperature critically determine the extent of ischemic neuronal injury J Cereb Blood Flow Metab 7 729–38 [3] Michenfelder J D et al 1991 The relationship among canine brain temperature, metabolism, and function during hypothermia Anesthesiology 75 130–6 [4] Mrozek S et al 2012 Brain temperature: physiology and pathophysiology after brain injury Anesthesiol Res Pract 2012 989487 [5] Schwab S et al 1997 Brain temperature monitoring and modulation in patients with severe MCA infarction Neurology 48 762–7 [6] Marshall I et al 2006 Measurement of regional brain temperature using proton spectroscopic imaging: validation and application to acute ischemic stroke Magn Reson Imaging 24 438–46 [7] Mack M G et al 2002 MR-guided ablation of head and neck tumors Magn Reson Imaging Clin North Am 10 707–13 [8] Kim J H et al 1979 Clinical and biological studies of localized hyperthermia Cancer Res 39 2258–61 [9] Arthur R M et al 2005 Non-invasive estimation of hyperthermiatemperatures with ultrasound Int J Hyperthermia 21 589–600 [10] Hand J W et al 2001 Monitoring of deep brain temperature in infants using multi-frequency microwave radiometry and thermal modelling Phys Med Biol 46 1885–1903 [11] Kuroda K et al 2000 Temperature mapping using the water proton chemical shift: self-referenced method with echoplanar spectroscopic imaging Magn Reson Med 43 220–5 [12] Viola R and Pauly K B 2008 MR thermometry J Magn Reson Imaging 27 376–90 [13] Thrippleton M J et al 2014 Reliability of MRSI brain temperature mapping at 1.5 and T NMR Biomed 2014 183–90 [14] Krafft C et al 2007 Methodology for fiber-optic Raman mapping and FTIR imaging of metastases in mouse brains Anal Bioanal Chem 389 1133–42 [15] Li C et al 2012 Brain temperature measurement: a study of in vitro accuracy and stability of smart catheter temperature sensors Biomed Microdevices 14 109–18 [16] Kiyatkin E et al 2003 Fluctuations in neural activity during cocaine self-administration: clues provided by brain thermorecording Neuroscience 116 525–38 [17] Stewart C et al 2008 The new Licox combined brain tissue oxygen and brain temperature monitor: assessment of in vitro accuracy and clinical experience in severe traumatic brain injury Neurosurgery 63 1159–65 [18] Raymann R J et al 2008 Skin deep: enhanced sleep depth by cutaneous temperature manipulation Brain 131 500–13 [19] Thayne R L et al 1995 An accurate technique for detailed prostatic interstitial temperature-mapping in patients receiving microwave thermal treatment J Endourol 9 339–47 [20] Lee B C et al 2009 Microfabricated neural thermocouple arrays probe for brain research Proc Int Solid-State Sensors, Actuators and Microsystems Conf pp 338–41 [21] Urban G et al 1990 High-resolution thin-film temperature sensor arrays for medical applications Sensors Actuators A 22 650–4 4.2. In vivo transient response The needle prick model was employed as a mechanically induced ‘mini’ traumatic brain injury causing one single spreading depolarization (SD) wave [39] Two needle pricks were performed in two rats with a 1 h interval A representa tive recording is shown in figure 10 Successful induction of the SD wave was confirmed by a DC negative shift (5.9 ± 0.7 mV; n = 4) and a transient suppression of spontaneous neural activity During the SD, the brain cortical temperature showed a transient increase (0.35 ± 0.06 for R1 and R2; 0.23 ± 0.07 for R3 and R4; n = 4) and then slowly recovered to its base line value The temperature sensing elements R1 and R2 with larger resistance showed higher sensitivity than the temper ature sensing elements R3 and R4 5. Conclusions In this work, a flexible polysilicon thermistor array has been developed on thin polyimide substrate for brain temperature monitoring The array has low mechanical stiffness and the ultra-small dimension of developed polysilicon thermistors can be applied for high spatial resolution measurement with sufficient accuracy and resolution Comparison of polysilicon thermistors for two dimensions 20 àmì160 µm and 200 µm × 160 µm showed the thermistors with larger resistance (20 àmì160 àm) had better performance both in vitro and in vivo The developed polysilicon thermistors with dimension of 20 àmì160 àm has achieved good in vitro performance with a sensitivity of −0.0031 °C−1, response time of 1.5 s, a resolution of 0.1 °C, thermal hysteresis less than 0.1 °C, and long term stability with drift less than 0.3 °C for d of continuous operation in water In vivo tests of the poly silicon thermistor showed a low noise level of 0.025 ± 0.03 °C and the expected transient temperature increase associated with cortical spreading depolarization Thus, the polysilicon thermistor array developed here has successfully achieved the high temporal and spatial resolution required for moni toring brain temperature The obtained results further suggest the superior performance of polysilicon thermistor compared to RTD Acknowledgments This research was supported by the US Army CDMRP PH/ TBI Research Program (Awards Nos W81XWH-10-1-0978 and W81XWH-10-1-0977) All the microfabrication was performed at the Engineering Research Center (ERC) Clean Room of the University of Cincinnati Z Wu et al J Micromech Microeng 27 (2017) 025001 [31] Tseng I et al 2013 Effect of magnetron sputtered silicon nitride on the water–vapor-permeation-rate of polyimide thin film Surf Coat Technol 231 496–500 [32] Carcia P F et al 2009 Gas diffusion ultrabarriers on polymer substrates using Al2O3 atomic layer deposition and SiN plasma-enhanced chemical vapor deposition J Appl Phys 106 023533 [33] Iwamori S et al 2003 Silicon oxide gas barrier films deposited by reactive sputtering Surf Coat Technol 166 24–30 [34] Tsai M et al 2013 Properties of magnetron-sputtered moisture barrier layer on transparent polyimide/graphene nanocomposite film Thin Solid Films 544 324–30 [35] Li C et al 2016 Evaluation of microelectrode materials for direct-current electrocorticography J Neural Eng 13 016008 [36] Lu N-C et al 1981 Modeling and optimization of monolithic polycrystalline silicon resistors IEEE Trans Electron Devices 28 818–30 [37] Seto J Y W 1975 The electrical properties of polycrystalline silicon films J Appl Phys 46 5247–54 [38] Rydberg M et al 2000 Effect of post-nitride-passivation processing on the long-term stability of polysilicon integrated circuit resistors J Vac Sci Technol B 18 690–4 [39] Li C et al 2016 Development and application of a microfabricated multimodal neural catheter for neuroscience Biomed Microdevices 18 1–10 [22] Moinuddin A et al 2012 Temperature sensor in a flexible substrate IEEE Sens 12 864–9 [23] Mylee W B et al 2016 A flexible vanadium oxide thermistor array for localized temperature field measurements in brain IEEE Sens 16 2211–12 [24] Obermeier E et al 1992 Polysilicon as a material for microsensor applications Sensors Actuators A 30 149–55 [25] Bendekovi Z et al 1998 Polysilicon temperature sensor 9th Mediterranean Electrotechnical Conf., MELECON 98 (Tel-Aviv) vol pp 362–6 [26] French P J 2002 Polysilicon: a versatile material for microsystems Sensors Actuators A 99 3–12 [27] Wu Z H et al 2006 A new high-temperature thermal sensor based on large-grain polysilicon on insulator Sensors Actuators A 130 129–34 [28] Wu Z et al 2016 Polysilicon thin film developed on flexible polyimide for biomedical applications J Microelectromech Syst 25 585–92 [29] Oliver N et al 1998 Aluminum-induced crystallization of amorphous silicon on glass substrates above and below the eutectic temperature Appl Phys Lett 73 3214 [30] Hiraki A et al 1980 Electronic states of Ni-silicides and its relation to metal-silicide/Si interfaces J Electrochem Soc 127 1542–5 ... injured brain In this work, a highly accurate temperature sensor with polysilicon thermistors has been developed on flexible polyimide for monitoring brain temperature with high spatial resolution. .. Microeng 27 (2017) 025001 (8pp) doi:10.1088/1361-6439/aa4e99 Polysilicon- based flexible temperature sensor for brain monitoring with high spatial resolution Zhizhen Wu1,4, Chunyan Li2, Jed Hartings3,... in cortical temperature associated with cortical spreading depolarization The temperature sensor developed in this work is suitable for monitoring brain temperature with the desired high sensitivity