- Email: [email protected]
Pressure sensing using a completely flexible organic transistor
Biosensors and Bioelectronics 22 (2007) 2775–2779
Pressure sensing using a completely flexible organic transistor I. Manunza a,b,∗ , A. Bonfiglio a,b a
CNISM, University of Cagliari, Department of Electric and Electronic Engineering, Italy b INFM-CNR, S3 NanoStructures and BioSystems at Surfaces, Modena, Italy
Received 30 June 2006; received in revised form 9 January 2007; accepted 24 January 2007 Available online 3 February 2007
Abstract In this paper, we report on pressure sensors based on completely flexible organic thin film transistors (OTFTs). A flexible and transparent plastic foil (Mylar) is employed both as substrate and gate dielectric. Gold source and drain electrodes are patterned on the upper side of the foil while the gate electrode lies on the opposite side; a vacuum-sublimed pentacene film is used as active layer. The pressure dependence of the output current has been investigated by applying to the gate side of the device a mechanical stimulus by means of a pressurized airflow. Experimental results show a reversible dependence of the current on the pressure. The data analysis suggests that the current variations are due to pressure-induced variations of mobility, threshold voltage and possibly contact resistance. The drain current variation is reproducible, linear and reversible even though it displays a hysteresis. Moreover, the sensor responds very fast to the mechanical stimulus (i.e. within tens–hundreds of milliseconds) but the time required to reach the steady state is much higher (tens–hundreds of seconds). Electrical characteristics with and without applied pressure have been carried out in air without any extra ad hoc read-out circuit or equipment. The reported devices show potential advantages of flexibility of the structure, low cost and versatility of the device structure for sensor technologies. Many innovative and attractive applications as wearable electronics, e-textiles, e-skin for robots can be considered. © 2007 Elsevier B.V. All rights reserved. Keywords: Organic field effect transistor; Pressure detection; Field effect sensor; Organic semiconductor
1. Introduction Organic electronics has received a lot of attention in recent years since the performance of organic thin-film transistors as well as OLED (Organic Light Emitting Diode), solar cells etc., is considerably improved (Nathan and Chalamala, 2005). Organic field-effect transistors can be assembled on plastic films at room temperatures; they are potentially inexpensive to manufacture and mechanically flexible. The features mentioned above make them promising for low-cost and large area electronics applications (Forrest, 2004; Mabeck and Malliaras, 2006; Someya et al., 2005). In particular, for a broad range of sensing applications there is an increasing demand for small, portable and low-cost sensors. Transistor-based sensors are active devices: this implies that they can be electrically characterized in a more complex way compared with passive mono-parameter devices.
∗
Corresponding author at: CNISM, University of Cagliari, Department of Electric and Electronic Engineering, Italy. E-mail address: [email protected] (I. Manunza). 0956-5663/$ – see front matter © 2007 Elsevier B.V. All rights reserved. doi:10.1016/j.bios.2007.01.021
They are multi-parameter sensors: I–V characteristics can be used to extract a set of variables in order to characterize their response to the parameter to be sensed. Finally, active sensors combine in the same device both switching and sensing functions. Therefore they can be used to obtain quite easily a sensing matrix of limited size and optimized reliability. Although the mobility of organic semiconductors is much lower (even three orders of magnitude) than that of poly- and single-crystalline silicon, the slower speed is tolerable for most applications of large-area sensors (Crone et al., 2000; Someya et al., 2004). In fact, during past years, many innovative and attractive applications have been proposed for sensing technology, for instance chemical, biological (Torsi et al., 2000; Bartic et al., 2003; Loi et al., 2005) and humidity sensors (Qiu et al., 2003; Torsi et al., 2001). On the opposite, only a few examples of mechanical sensors have been reported so far (Someya et al., 2004, 2005; Kawaguchi et al., 2005; Darlinski et al., 2005). Someya et al., reported (Someya et al., 2004, 2005; Kawaguchi et al., 2005) a flexible pressure sensor matrix in which organic transistor active matrixes were used to read pressure data from piezoresistive elements. In this case, pentacene
2776
I. Manunza, A. Bonfiglio / Biosensors and Bioelectronics 22 (2007) 2775–2779
based organic transistors were employed to address a pressuresensitive rubber element. Darlinski et al. (2005) reported about the pressure-sensitivity of an OTFT that acts as sensor element on a rigid substrate; in this case, the mechanical force is applied directly to the entire device (channel + source and drain contacts) by means of a micro-needle. In this article we propose a flexible OTFT structure where the organic semiconductor device performs both sensing and switching functions. By exploiting the properties of this structure, the sensor can be combined with any kind of substrate or it can be used as a free-standing device. Taking advantage of the full mechanical flexibility of the insulating sheet, attractive developments can be foreseen for this “electronic film”, for instance, a matrix combining different sensing devices (chemical, temperature, etc.) for robotic skin applications.
2. Device fabrication and experimental set-up The device consists of a pentacene-based substrate-free structure, with gold bottom-contact source and drain electrodes as shown in Fig. 1(a). A 1.6 m thick Mylar foil (Du Pont, dielectric constant of 3.3, dielectric rigidity of 105 V/cm) is used as gate insulator and also as mechanical support of the whole device. First of all the Mylar foil is clamped to a cylindrical plastic frame (2.5 cm in diameter) in order to obtain a suspended membrane with both sides available for processing. The foil is then cleaned with acetone, washed with deionised water and finally dried with a nitrogen flux. Then, bottom-contact gold electrodes (nominal thickness 100 nm) with W/L = 100 (W = 5 mm and L = 50 m are the channel width and length, respectively) are patterned on the upper side of the flexible dielectric foil, using a standard photolithographic technique, while the gold gate electrode is patterned on the opposite side. Since the Mylar is transparent to UV light, source and drain may be used as shadow mask for the gate patterning. Therefore, a thin photoresist layer is spin-coated on the lower side of the Mylar layer and then it is exposed to UV light projected through the Mylar itself. In this way, source and drain electrodes act as a mask for the UV light and a perfect alignment between source and drain and the impressed photoresist is
obtained. After the development process, a gold layer (nominal thickness 100 nm) is vacuum-sublimed and patterned by means of lift-off etching with acetone. In this way, we obtained an autoalignment of source, drain and gate contacts by simply using a one-mask photolithographic process, and as a consequence all the parasitic capacitance effects due to source-drain and gate metal overlapping are drastically limited. Further experimental details on the fabrication of the transistor structure have been reported elsewhere (Bonfiglio et al., 2003). In order to study the influence of structural effects (in particular of the contact/semiconductor interface) on the pressure sensitivity, we have also realized, on the same insulating layer, couples of bottom-contact and top-contact devices with the same active layer. In this case, source and drain contacts are patterned by means of soft lithography using the conductive polymer PEDT:PSS (poly-(3,4-ethylene dioxythiophene):polystyrene sulfonic acid) as described in Cosseddu and Bonfiglio (2006). Source and drain contacts of the bottomcontact device are first patterned on a portion of the Mylar foil. Then, the semiconductor layer is grown by vacuum-sublimation on top of the whole insulating foil and finally source and drain electrodes of the top-contact structure are patterned aside the bottom-contact device by means of microcontact-printing (CP) as well. Pentacene film with a thickness of 50 nm is grown by ˚ vacuum-sublimation at a nominal deposition flux of about 1 A/s. Pentacene (Sigma–Aldrich) has been used as received. To study the correlation between the pressure applied on the device and its electrical performance, we have used the experimental set-up shown in Fig. 1(b): a pressurization chamber provided with a manometer and an air valve. The chamber has a circular aperture on the top and the device is clamped at the edge (2.5 cm in diameter) of a cylindrical frame fixed at the aperture. When the pressurized airflows into the chamber it deforms the freestanding device. Electrical characteristics with and without applied pressure have been carried out at room temperature in air, by using a HP 4155 Semiconductor Parameter Analyzer without any extra ad hoc read-out circuit or equipment.
3. Results and discussions The device has the typical behaviour of an organic p-type field effect transistor, working in accumulation mode. The
Fig. 1. (a) Basic structure of the device. Inset: detail of the channel area; (b) experimental set-up for testing the pressure sensitivity of the device.
I. Manunza, A. Bonfiglio / Biosensors and Bioelectronics 22 (2007) 2775–2779
2777
Fig. 2. (a) Comparison between ID vs. VD curves with different pressure states; (b) ID vs. time curve with different applied pressures; (c) reversibility of ID vs. time.
conductance of the organic semiconductor in the channel region is switched on and off by the gate electrode, which is capacitively coupled through the dielectric layer. The gate bias (VG ) controls the current (ID ) flowing between the source and drain electrodes under an imposed bias (VD ). Applying a negative gate voltage causes an accumulation of carriers and increases the conductance. Typical recorded values of hole mobilities are in the range of 10−3 to 10−2 cm2 /V s, while threshold voltages are in the order of some tens of Volts. Fig. 2(a) shows the drain current ID versus the drain voltage VD at different values of VG and with different pressures applied. The characteristic shows a decrease in the current when pressure is applied. The observed sensitivity concerns both the triode region and the saturation region. Fig. 2(b) and (c) show the time variation of ID (in the linear region) while pressure is applied to the device. In Fig. 2(b) the current variation in response to different values of the applied pressure is displayed, while Fig. 2(c) shows what happens when the same value of pressure is applied and then removed according to a time sequence. As it can be observed, the drain current variation is reproducible, linear and reversible despite the presence of a hysteresis. In addition, it can be observed that the sensor responds very fast to the mechanical stimulus (i.e. within tens–hundreds of milliseconds) but the time required to reach the steady state is much higher (tens–hundreds
of seconds) with a significant variability from sample to sample. This time instability and hysteresis are mainly to be attributed to the intrinsic properties of organic semiconductors. Possible parasitic capacitance effects (that normally produce hysteresis and response delay in the curves) have been minimized through the auto-alignment of contacts; on the other hand, it is worth noting that the device has been measured in air and it is not encapsulated, therefore we cannot exclude long term degradation effects in the semiconducting layer due to humidity, oxygen, etc. Indeed these issues deserve more investigation, since faster dynamics, reproducibility and reversibility are obviously desirable for practical applications. To explain the possible cause of the observed sensitivity of the current to the pressure, it is necessary to take into consideration all parameters affecting the drain current of a field effect transistor. This current is satisfactorily described in the linear regime by the following Eq. (1): W (VD − Rs ID )2 ID = − μCi (VG − Vth )(VD − Rs ID ) − L 2 (1) Variations of W/L ratio, mobility μ, insulating layer capacitance Ci , threshold voltage Vth and contact resistance Rs can
Fig. 3. (a) IDmax /IDmax vs. pressure; (b) average mobility vs. pressure; (c) average threshold voltage vs. pressure; (d) IDmax /IDmax vs. pressure (one sample, five consecutive tests).
2778
I. Manunza, A. Bonfiglio / Biosensors and Bioelectronics 22 (2007) 2775–2779
in principle be related to the pressure-induced variation of the drain current. Firstly, W and L may undergo some modifications when the device is deformed, but it should be noticed that both the mechanical stimulus applied by means of an airflow and the insulating layer are mechanically isotropic. Therefore, the width W and the length L of the transistor channel change at the same extent, thus keeping constant their ratio. Secondly, a possible thinning of the insulating film thickness di should induce an increase in the insulating layer capacitance Ci , being Ci = εi /di . This increase in the insulating layer capacitance is expected to result in an increase in the drain current. On the contrary, we always observe a decrease in the drain current, meaning that this effect, if present, is always negligible. Therefore, the observed sensitivity must be attributed to a different, non-geometrical effect. In order to verify this hypothesis, mobility and threshold voltage have been extracted by fitting our data with Eq. (1). Data showing maximum current (VD = −100 V, VG = −100 V), mobility and threshold voltage is plotted against pressure in Fig. 3. The maximum current has been plotted in terms of relative variation with respect to the value recorded without applying a pressure. This value varies from sample to sample (this is absolutely normal in organic semiconductor based devices not encapsulated and measured in air), nevertheless there is an uncontroversial linear dependence of the current on the pressure. The extracted values of mobility and threshold voltage show a similar linear dependence but standard deviation is higher. This can be partially attributed to the fact that mobility and threshold voltage result from an extrapolation that is affected by possible failures of the fitting model. In Fig. 3(d), we show the relative variation of IDmax in the same sample for different measurements (periodically repeated in time during several hours). Again we can observe that there is a linear and reproducible behaviour,
Fig. 4. Hysteresis in ID vs. VG curves with different applied pressures. Inset: detail for −40 V < VG < −70 V.
despite the fact that the data is extracted from curves taken during prolonged measurement sessions (6–7 h), thus causing mechanical and electrical stress to the device. Another interesting observation can be done on ID –VG curves, shown in Fig. 4; here the sweeping of the gate voltage from positive to negative values and vice versa causes in the drain current a hysteresis phenomenon, that enlarges when pressure is increased. This fact suggests that the pressure-induced changes in drain current can be related not only to a direct dependence of the semiconductor conductivity on the pressure applied to the device but also to the distribution and activity of trap states that typically affect contact resistances, threshold voltages (Horowitz, 2004; Pernistich et al., 2004; Schroeder et al., 2003) and hysteresis (Lindner et al., 2005). To further clarify these observations, we decided to perform experiments on devices with different structures. In particular,
Fig. 5. Comparison between (a) threshold voltage vs. pressure; (b) mobility vs. pressure; hysteresis in ID vs. VG curves with different applied pressures in (c) top-contact and (d) bottom-contact devices.
I. Manunza, A. Bonfiglio / Biosensors and Bioelectronics 22 (2007) 2775–2779
we assembled couples of bottom-contact and top-contact devices as previously described. The goal of the experiment was to check if these different structures behave differently when a pressure is applied and to see which parameter is mostly affected. The different metal–semiconductor interface is expected to affect the behaviour of the electrical characteristics of the transistors even if no pressure is applied. The same organic semiconductor is deposited on both devices, in order to have the same thickness and quality of the semiconductor in the channel of each transistor. The geometry of the contacts is also exactly the same; and in conclusion the only difference between the two devices is the interface between semiconductor and contacts. The results of these measurements are shown in Fig. 5. In the bottom-contact device we can observe a sharp dependence of the threshold voltage (Fig. 5(a) left Y axis) and hysteresis (Fig. 5(d)) on the applied pressure, whereas for the top-contact device no correlation has been found between threshold voltage and pressure (Fig. 5(a) right Y axis) and also the hysteresis phenomenon is negligible in this case (Fig. 5(c)). Contact resistance effects cannot be evaluated with the same model in top and bottom-contact devices: in the bottom-contact structure the semiconductor layer is grown on top of source and drain electrodes, so there is a direct carrier injection from the electrode to the device channel and the contribution to Rs may be attributed to the contact between electrodes and semiconductor, while in the top-contact structure in addition to the contribution to Rs due to source and drain contacts, there is a further voltage drop due to the fact that the current injected from the electrodes must vertically travel through the semiconductor layer before reaching the channel. For these reasons a direct comparison of the pressure-dependence of the contact resistance is not possible and needs further investigations. The different dependence of threshold voltage and hysteresis on pressure in bottom-contact and top-contact devices confirms the role of insulator/semiconductor and metal/semiconductor interfaces in determining the pressure sensitivity of the device. On the other hand, mobility has a very similar behaviour in the two devices (see Fig. 5(b)) indicating also a direct contribution of the semiconductor in the channel to the observed sensitivity. 4. Conclusions We realized a pressure sensor based on a completely flexible organic thin film transistor. We have observed a marked sensitivity of the drain current to the elastic deformation induced by a pressure applied on the device channel. The electrical charac-
2779
teristics indicate that the output current varies quickly, linearly and reversibly in response to the mechanical stimulus applied to the device. The time required to reach the current stable state after applying the stimulus and the high drain and gate voltages required for the device operation are the main drawbacks of the structure. Measurements recorded on different devices with the same structure and on devices with different structures suggest that the underlying mechanism responsible for the observed sensitivity could be related not only to a pressure-induced decrease of pentacene mobility in the channel, but also to interface effects at the contacts/semiconductor boundary. Work is in progress to further clarify these issues. Acknowledgments The authors acknowledge the European Commission for founding the project under the Programme IST-IP, VI FP Integrated Project No. 26987 “PROETEX”, and the Project FIRB RBAU01THZA 002 of the Italian Ministry of Research. References Bartic, C., Campitelli, A., Borghs, S., 2003. Appl. Phys. Lett. 82, 475–477. Bonfiglio, A., Mameli, F., Sanna, O., 2003. Appl. Phys. Lett. 82, 3550–3552. Cosseddu, P., Bonfiglio, A., 2006. Appl. Phys. Lett. 88, 23406–23408. Crone, B., Dodabalapur, A., Lin, Y.-Y., Filas, R.W., Bao, Z., LaDuca, A., Sarpeshkar, R., Katz, H.E., Li, W., 2000. Nature 403, 521–523. Darlinski, G., B¨ottger, U., Waser, R., Klauk, H., Halik, M., Zschieschang, U., Schmidt, G., Dehm, C., 2005. J. Appl. Phys. 97, 93708–93710. Forrest, S.R., 2004. Nature 428, 911–918. Horowitz, G., 2004. J. Mater. Res. 19, 1946–1962. Kawaguchi, H., Someya, T., Sekitani, T., Sakurai, T., 2005. IEEE J. Solid State Circuits 40, 177–185. Lindner, Th., Paasch, G., Scheinert, S., 2005. J. Appl. Phys. 98, 114505:1–114505:9. Loi, A., Manunza, I., Bonfiglio, A., 2005. Appl. Phys. Lett. 86, 103512–103514. Mabeck, J.T., Malliaras, G., 2006. Anal. Bioanal. Chem. 384, 343–353. Nathan, A., Chalamala, B.R., 2005. Proc. IEEE 93 (8), 1391–1510. Pernistich, K.P., Haas, S., Oberhoff, D., Goldmann, C., Gundlach, D.J., Batlogg, B., Rashid, A.N., Schitter, G., 2004. J. Appl. Phys. 96, 6431–6438. Qiu, Y., Hu, Y., Dong, G., Wang, L., Xie, J., Ma, Y., 2003. Appl. Phys. Lett. 8, 1644–1646. Schroeder, R., Majewski, L.A., Grell, M., 2003. Appl. Phys. Lett. 83, 3201–3203. Someya, T., Sekitani, T., Kato, Y., Iba, S., Kawaguchi, H., Sakurai, T., 2004. PNAS 101, 9966–9970. Someya, T., Kato, Y., Sekitani, T., Iba, S., Noguchi, Y., Murase, Y., Kawaguchi, H., Sakurai, T., 2005. PNAS 102 (35), 12321–12325. Torsi, L., Dodabalapur, A., Sabbatini, L., Zambonin, P.G., 2000. Sens. Actuators B 67, 312–316. Torsi, L., Dodabalapur, A., Cioffi, N., Sabbatini, L., Zambonin, P.G., 2001. Sens. Actuators B 77, 7–11.
Pressure sensing using a completely flexible organic transistor I. Manunza a,b,∗ , A. Bonfiglio a,b a
CNISM, University of Cagliari, Department of Electric and Electronic Engineering, Italy b INFM-CNR, S3 NanoStructures and BioSystems at Surfaces, Modena, Italy
Received 30 June 2006; received in revised form 9 January 2007; accepted 24 January 2007 Available online 3 February 2007
Abstract In this paper, we report on pressure sensors based on completely flexible organic thin film transistors (OTFTs). A flexible and transparent plastic foil (Mylar) is employed both as substrate and gate dielectric. Gold source and drain electrodes are patterned on the upper side of the foil while the gate electrode lies on the opposite side; a vacuum-sublimed pentacene film is used as active layer. The pressure dependence of the output current has been investigated by applying to the gate side of the device a mechanical stimulus by means of a pressurized airflow. Experimental results show a reversible dependence of the current on the pressure. The data analysis suggests that the current variations are due to pressure-induced variations of mobility, threshold voltage and possibly contact resistance. The drain current variation is reproducible, linear and reversible even though it displays a hysteresis. Moreover, the sensor responds very fast to the mechanical stimulus (i.e. within tens–hundreds of milliseconds) but the time required to reach the steady state is much higher (tens–hundreds of seconds). Electrical characteristics with and without applied pressure have been carried out in air without any extra ad hoc read-out circuit or equipment. The reported devices show potential advantages of flexibility of the structure, low cost and versatility of the device structure for sensor technologies. Many innovative and attractive applications as wearable electronics, e-textiles, e-skin for robots can be considered. © 2007 Elsevier B.V. All rights reserved. Keywords: Organic field effect transistor; Pressure detection; Field effect sensor; Organic semiconductor
1. Introduction Organic electronics has received a lot of attention in recent years since the performance of organic thin-film transistors as well as OLED (Organic Light Emitting Diode), solar cells etc., is considerably improved (Nathan and Chalamala, 2005). Organic field-effect transistors can be assembled on plastic films at room temperatures; they are potentially inexpensive to manufacture and mechanically flexible. The features mentioned above make them promising for low-cost and large area electronics applications (Forrest, 2004; Mabeck and Malliaras, 2006; Someya et al., 2005). In particular, for a broad range of sensing applications there is an increasing demand for small, portable and low-cost sensors. Transistor-based sensors are active devices: this implies that they can be electrically characterized in a more complex way compared with passive mono-parameter devices.
∗
Corresponding author at: CNISM, University of Cagliari, Department of Electric and Electronic Engineering, Italy. E-mail address: [email protected] (I. Manunza). 0956-5663/$ – see front matter © 2007 Elsevier B.V. All rights reserved. doi:10.1016/j.bios.2007.01.021
They are multi-parameter sensors: I–V characteristics can be used to extract a set of variables in order to characterize their response to the parameter to be sensed. Finally, active sensors combine in the same device both switching and sensing functions. Therefore they can be used to obtain quite easily a sensing matrix of limited size and optimized reliability. Although the mobility of organic semiconductors is much lower (even three orders of magnitude) than that of poly- and single-crystalline silicon, the slower speed is tolerable for most applications of large-area sensors (Crone et al., 2000; Someya et al., 2004). In fact, during past years, many innovative and attractive applications have been proposed for sensing technology, for instance chemical, biological (Torsi et al., 2000; Bartic et al., 2003; Loi et al., 2005) and humidity sensors (Qiu et al., 2003; Torsi et al., 2001). On the opposite, only a few examples of mechanical sensors have been reported so far (Someya et al., 2004, 2005; Kawaguchi et al., 2005; Darlinski et al., 2005). Someya et al., reported (Someya et al., 2004, 2005; Kawaguchi et al., 2005) a flexible pressure sensor matrix in which organic transistor active matrixes were used to read pressure data from piezoresistive elements. In this case, pentacene
2776
I. Manunza, A. Bonfiglio / Biosensors and Bioelectronics 22 (2007) 2775–2779
based organic transistors were employed to address a pressuresensitive rubber element. Darlinski et al. (2005) reported about the pressure-sensitivity of an OTFT that acts as sensor element on a rigid substrate; in this case, the mechanical force is applied directly to the entire device (channel + source and drain contacts) by means of a micro-needle. In this article we propose a flexible OTFT structure where the organic semiconductor device performs both sensing and switching functions. By exploiting the properties of this structure, the sensor can be combined with any kind of substrate or it can be used as a free-standing device. Taking advantage of the full mechanical flexibility of the insulating sheet, attractive developments can be foreseen for this “electronic film”, for instance, a matrix combining different sensing devices (chemical, temperature, etc.) for robotic skin applications.
2. Device fabrication and experimental set-up The device consists of a pentacene-based substrate-free structure, with gold bottom-contact source and drain electrodes as shown in Fig. 1(a). A 1.6 m thick Mylar foil (Du Pont, dielectric constant of 3.3, dielectric rigidity of 105 V/cm) is used as gate insulator and also as mechanical support of the whole device. First of all the Mylar foil is clamped to a cylindrical plastic frame (2.5 cm in diameter) in order to obtain a suspended membrane with both sides available for processing. The foil is then cleaned with acetone, washed with deionised water and finally dried with a nitrogen flux. Then, bottom-contact gold electrodes (nominal thickness 100 nm) with W/L = 100 (W = 5 mm and L = 50 m are the channel width and length, respectively) are patterned on the upper side of the flexible dielectric foil, using a standard photolithographic technique, while the gold gate electrode is patterned on the opposite side. Since the Mylar is transparent to UV light, source and drain may be used as shadow mask for the gate patterning. Therefore, a thin photoresist layer is spin-coated on the lower side of the Mylar layer and then it is exposed to UV light projected through the Mylar itself. In this way, source and drain electrodes act as a mask for the UV light and a perfect alignment between source and drain and the impressed photoresist is
obtained. After the development process, a gold layer (nominal thickness 100 nm) is vacuum-sublimed and patterned by means of lift-off etching with acetone. In this way, we obtained an autoalignment of source, drain and gate contacts by simply using a one-mask photolithographic process, and as a consequence all the parasitic capacitance effects due to source-drain and gate metal overlapping are drastically limited. Further experimental details on the fabrication of the transistor structure have been reported elsewhere (Bonfiglio et al., 2003). In order to study the influence of structural effects (in particular of the contact/semiconductor interface) on the pressure sensitivity, we have also realized, on the same insulating layer, couples of bottom-contact and top-contact devices with the same active layer. In this case, source and drain contacts are patterned by means of soft lithography using the conductive polymer PEDT:PSS (poly-(3,4-ethylene dioxythiophene):polystyrene sulfonic acid) as described in Cosseddu and Bonfiglio (2006). Source and drain contacts of the bottomcontact device are first patterned on a portion of the Mylar foil. Then, the semiconductor layer is grown by vacuum-sublimation on top of the whole insulating foil and finally source and drain electrodes of the top-contact structure are patterned aside the bottom-contact device by means of microcontact-printing (CP) as well. Pentacene film with a thickness of 50 nm is grown by ˚ vacuum-sublimation at a nominal deposition flux of about 1 A/s. Pentacene (Sigma–Aldrich) has been used as received. To study the correlation between the pressure applied on the device and its electrical performance, we have used the experimental set-up shown in Fig. 1(b): a pressurization chamber provided with a manometer and an air valve. The chamber has a circular aperture on the top and the device is clamped at the edge (2.5 cm in diameter) of a cylindrical frame fixed at the aperture. When the pressurized airflows into the chamber it deforms the freestanding device. Electrical characteristics with and without applied pressure have been carried out at room temperature in air, by using a HP 4155 Semiconductor Parameter Analyzer without any extra ad hoc read-out circuit or equipment.
3. Results and discussions The device has the typical behaviour of an organic p-type field effect transistor, working in accumulation mode. The
Fig. 1. (a) Basic structure of the device. Inset: detail of the channel area; (b) experimental set-up for testing the pressure sensitivity of the device.
I. Manunza, A. Bonfiglio / Biosensors and Bioelectronics 22 (2007) 2775–2779
2777
Fig. 2. (a) Comparison between ID vs. VD curves with different pressure states; (b) ID vs. time curve with different applied pressures; (c) reversibility of ID vs. time.
conductance of the organic semiconductor in the channel region is switched on and off by the gate electrode, which is capacitively coupled through the dielectric layer. The gate bias (VG ) controls the current (ID ) flowing between the source and drain electrodes under an imposed bias (VD ). Applying a negative gate voltage causes an accumulation of carriers and increases the conductance. Typical recorded values of hole mobilities are in the range of 10−3 to 10−2 cm2 /V s, while threshold voltages are in the order of some tens of Volts. Fig. 2(a) shows the drain current ID versus the drain voltage VD at different values of VG and with different pressures applied. The characteristic shows a decrease in the current when pressure is applied. The observed sensitivity concerns both the triode region and the saturation region. Fig. 2(b) and (c) show the time variation of ID (in the linear region) while pressure is applied to the device. In Fig. 2(b) the current variation in response to different values of the applied pressure is displayed, while Fig. 2(c) shows what happens when the same value of pressure is applied and then removed according to a time sequence. As it can be observed, the drain current variation is reproducible, linear and reversible despite the presence of a hysteresis. In addition, it can be observed that the sensor responds very fast to the mechanical stimulus (i.e. within tens–hundreds of milliseconds) but the time required to reach the steady state is much higher (tens–hundreds
of seconds) with a significant variability from sample to sample. This time instability and hysteresis are mainly to be attributed to the intrinsic properties of organic semiconductors. Possible parasitic capacitance effects (that normally produce hysteresis and response delay in the curves) have been minimized through the auto-alignment of contacts; on the other hand, it is worth noting that the device has been measured in air and it is not encapsulated, therefore we cannot exclude long term degradation effects in the semiconducting layer due to humidity, oxygen, etc. Indeed these issues deserve more investigation, since faster dynamics, reproducibility and reversibility are obviously desirable for practical applications. To explain the possible cause of the observed sensitivity of the current to the pressure, it is necessary to take into consideration all parameters affecting the drain current of a field effect transistor. This current is satisfactorily described in the linear regime by the following Eq. (1): W (VD − Rs ID )2 ID = − μCi (VG − Vth )(VD − Rs ID ) − L 2 (1) Variations of W/L ratio, mobility μ, insulating layer capacitance Ci , threshold voltage Vth and contact resistance Rs can
Fig. 3. (a) IDmax /IDmax vs. pressure; (b) average mobility vs. pressure; (c) average threshold voltage vs. pressure; (d) IDmax /IDmax vs. pressure (one sample, five consecutive tests).
2778
I. Manunza, A. Bonfiglio / Biosensors and Bioelectronics 22 (2007) 2775–2779
in principle be related to the pressure-induced variation of the drain current. Firstly, W and L may undergo some modifications when the device is deformed, but it should be noticed that both the mechanical stimulus applied by means of an airflow and the insulating layer are mechanically isotropic. Therefore, the width W and the length L of the transistor channel change at the same extent, thus keeping constant their ratio. Secondly, a possible thinning of the insulating film thickness di should induce an increase in the insulating layer capacitance Ci , being Ci = εi /di . This increase in the insulating layer capacitance is expected to result in an increase in the drain current. On the contrary, we always observe a decrease in the drain current, meaning that this effect, if present, is always negligible. Therefore, the observed sensitivity must be attributed to a different, non-geometrical effect. In order to verify this hypothesis, mobility and threshold voltage have been extracted by fitting our data with Eq. (1). Data showing maximum current (VD = −100 V, VG = −100 V), mobility and threshold voltage is plotted against pressure in Fig. 3. The maximum current has been plotted in terms of relative variation with respect to the value recorded without applying a pressure. This value varies from sample to sample (this is absolutely normal in organic semiconductor based devices not encapsulated and measured in air), nevertheless there is an uncontroversial linear dependence of the current on the pressure. The extracted values of mobility and threshold voltage show a similar linear dependence but standard deviation is higher. This can be partially attributed to the fact that mobility and threshold voltage result from an extrapolation that is affected by possible failures of the fitting model. In Fig. 3(d), we show the relative variation of IDmax in the same sample for different measurements (periodically repeated in time during several hours). Again we can observe that there is a linear and reproducible behaviour,
Fig. 4. Hysteresis in ID vs. VG curves with different applied pressures. Inset: detail for −40 V < VG < −70 V.
despite the fact that the data is extracted from curves taken during prolonged measurement sessions (6–7 h), thus causing mechanical and electrical stress to the device. Another interesting observation can be done on ID –VG curves, shown in Fig. 4; here the sweeping of the gate voltage from positive to negative values and vice versa causes in the drain current a hysteresis phenomenon, that enlarges when pressure is increased. This fact suggests that the pressure-induced changes in drain current can be related not only to a direct dependence of the semiconductor conductivity on the pressure applied to the device but also to the distribution and activity of trap states that typically affect contact resistances, threshold voltages (Horowitz, 2004; Pernistich et al., 2004; Schroeder et al., 2003) and hysteresis (Lindner et al., 2005). To further clarify these observations, we decided to perform experiments on devices with different structures. In particular,
Fig. 5. Comparison between (a) threshold voltage vs. pressure; (b) mobility vs. pressure; hysteresis in ID vs. VG curves with different applied pressures in (c) top-contact and (d) bottom-contact devices.
I. Manunza, A. Bonfiglio / Biosensors and Bioelectronics 22 (2007) 2775–2779
we assembled couples of bottom-contact and top-contact devices as previously described. The goal of the experiment was to check if these different structures behave differently when a pressure is applied and to see which parameter is mostly affected. The different metal–semiconductor interface is expected to affect the behaviour of the electrical characteristics of the transistors even if no pressure is applied. The same organic semiconductor is deposited on both devices, in order to have the same thickness and quality of the semiconductor in the channel of each transistor. The geometry of the contacts is also exactly the same; and in conclusion the only difference between the two devices is the interface between semiconductor and contacts. The results of these measurements are shown in Fig. 5. In the bottom-contact device we can observe a sharp dependence of the threshold voltage (Fig. 5(a) left Y axis) and hysteresis (Fig. 5(d)) on the applied pressure, whereas for the top-contact device no correlation has been found between threshold voltage and pressure (Fig. 5(a) right Y axis) and also the hysteresis phenomenon is negligible in this case (Fig. 5(c)). Contact resistance effects cannot be evaluated with the same model in top and bottom-contact devices: in the bottom-contact structure the semiconductor layer is grown on top of source and drain electrodes, so there is a direct carrier injection from the electrode to the device channel and the contribution to Rs may be attributed to the contact between electrodes and semiconductor, while in the top-contact structure in addition to the contribution to Rs due to source and drain contacts, there is a further voltage drop due to the fact that the current injected from the electrodes must vertically travel through the semiconductor layer before reaching the channel. For these reasons a direct comparison of the pressure-dependence of the contact resistance is not possible and needs further investigations. The different dependence of threshold voltage and hysteresis on pressure in bottom-contact and top-contact devices confirms the role of insulator/semiconductor and metal/semiconductor interfaces in determining the pressure sensitivity of the device. On the other hand, mobility has a very similar behaviour in the two devices (see Fig. 5(b)) indicating also a direct contribution of the semiconductor in the channel to the observed sensitivity. 4. Conclusions We realized a pressure sensor based on a completely flexible organic thin film transistor. We have observed a marked sensitivity of the drain current to the elastic deformation induced by a pressure applied on the device channel. The electrical charac-
2779
teristics indicate that the output current varies quickly, linearly and reversibly in response to the mechanical stimulus applied to the device. The time required to reach the current stable state after applying the stimulus and the high drain and gate voltages required for the device operation are the main drawbacks of the structure. Measurements recorded on different devices with the same structure and on devices with different structures suggest that the underlying mechanism responsible for the observed sensitivity could be related not only to a pressure-induced decrease of pentacene mobility in the channel, but also to interface effects at the contacts/semiconductor boundary. Work is in progress to further clarify these issues. Acknowledgments The authors acknowledge the European Commission for founding the project under the Programme IST-IP, VI FP Integrated Project No. 26987 “PROETEX”, and the Project FIRB RBAU01THZA 002 of the Italian Ministry of Research. References Bartic, C., Campitelli, A., Borghs, S., 2003. Appl. Phys. Lett. 82, 475–477. Bonfiglio, A., Mameli, F., Sanna, O., 2003. Appl. Phys. Lett. 82, 3550–3552. Cosseddu, P., Bonfiglio, A., 2006. Appl. Phys. Lett. 88, 23406–23408. Crone, B., Dodabalapur, A., Lin, Y.-Y., Filas, R.W., Bao, Z., LaDuca, A., Sarpeshkar, R., Katz, H.E., Li, W., 2000. Nature 403, 521–523. Darlinski, G., B¨ottger, U., Waser, R., Klauk, H., Halik, M., Zschieschang, U., Schmidt, G., Dehm, C., 2005. J. Appl. Phys. 97, 93708–93710. Forrest, S.R., 2004. Nature 428, 911–918. Horowitz, G., 2004. J. Mater. Res. 19, 1946–1962. Kawaguchi, H., Someya, T., Sekitani, T., Sakurai, T., 2005. IEEE J. Solid State Circuits 40, 177–185. Lindner, Th., Paasch, G., Scheinert, S., 2005. J. Appl. Phys. 98, 114505:1–114505:9. Loi, A., Manunza, I., Bonfiglio, A., 2005. Appl. Phys. Lett. 86, 103512–103514. Mabeck, J.T., Malliaras, G., 2006. Anal. Bioanal. Chem. 384, 343–353. Nathan, A., Chalamala, B.R., 2005. Proc. IEEE 93 (8), 1391–1510. Pernistich, K.P., Haas, S., Oberhoff, D., Goldmann, C., Gundlach, D.J., Batlogg, B., Rashid, A.N., Schitter, G., 2004. J. Appl. Phys. 96, 6431–6438. Qiu, Y., Hu, Y., Dong, G., Wang, L., Xie, J., Ma, Y., 2003. Appl. Phys. Lett. 8, 1644–1646. Schroeder, R., Majewski, L.A., Grell, M., 2003. Appl. Phys. Lett. 83, 3201–3203. Someya, T., Sekitani, T., Kato, Y., Iba, S., Kawaguchi, H., Sakurai, T., 2004. PNAS 101, 9966–9970. Someya, T., Kato, Y., Sekitani, T., Iba, S., Noguchi, Y., Murase, Y., Kawaguchi, H., Sakurai, T., 2005. PNAS 102 (35), 12321–12325. Torsi, L., Dodabalapur, A., Sabbatini, L., Zambonin, P.G., 2000. Sens. Actuators B 67, 312–316. Torsi, L., Dodabalapur, A., Cioffi, N., Sabbatini, L., Zambonin, P.G., 2001. Sens. Actuators B 77, 7–11.