Highly Stretchable Carbon Nanotube Transistors with Ion Gel Gate ...
Supporting Information
Highly Stretchable Carbon Nanotube Transistors with Ion Gel Gate Dielectrics Feng Xu†, Meng-Yin Wu‡, Nathaniel Safron†, Susmit Singha Roy†, Robert M. Jacobberger†, Dominick J. Bindl†, Jung-Hun Seo‡, Tzu-Hsuan Chang‡, Zhenqiang Ma‡, Michael S. Arnold†,* †
Department of Materials Science and Engineering and ‡Department of Electrical and Computer
Engineering, University of Wisconsin-Madison, Madison, WI 53706, USA
1
Methods Semiconducting carbon nanotube isolation. Semiconducting single-walled carbon nanotubes (CNTs) are prepared by methods adapted from those of Nish et al.1 and detailed in our previous work2. Solutions of 1 mg/mL CNTs (SouthWest Nanotechnologies, SG65i Lot# SG65i-L38) and 2 mg/mL poly(9,9dioctylfluorenyl-2,7-diyl) (PFO, American Dye Source) in toluene are ultrasonicated using a ½ inch tip horn ultrasonicator for 30 minutes at 40% max power. The resulting PFO-CNT slurry is ultracentrifuged for 15 min at 300,000g in a swing bucket rotor (ThermoFisher TH-641) to remove CNT bundles, unselected CNT chiralities, and catalyst particles. The top 90% of the resulting supernatant is decanted and filtered through a 5.0 µm PVDF membrane filter. Toluene is selectively removed via vacuum-distillation, and the resulting PFO/CNT solid is redissolved into hot tetrahydrofuran (THF). CNTs are selectively sedimented from the PFO-rich THF solution by centrifugation in a fixed-angle rotor at 50,000g at 4 °C for 24 hours. The PFO-rich supernatant is discarded and the semiconducting CNT-rich pellet is redispersed in THF in order to iterate this PFO-removal process, until the pellet contains CNTs in a PFO:CNT weight ratio less than 2:1, as determined via optical absorption, using calibration standards for the optical cross section of PFO and optical cross sections for the CNT S1 transition as reported by Hertel et al.3. Immediately prior to film casting, the CNT solutions are briefly agitated using microtip ultrasonication for 10 seconds at 10% max amplitude. Semiconducting carbon nanotube film deposition and transfer. Initially, PFO wrapped CNTs dispersed in chlorobenzene are doctor-blade cast onto clean glass on a hot-plate set to 110 °C in an argon glovebox, followed by annealing at 150 °C for 10 min to remove residual solvent. The spatially averaged thickness of the CNT film is roughly 5 nm, as determined by optical absorption spectroscopy using a thicker, calibration film that is measured using both profilometry and via absorption spectroscopy.
PDMS
substrates with thickness of 0.5 mm are prepared using Sylgard 184 (Dow Corning) by mixing the “base” and the “curing agent” at a ratio of 10:1. The mixture is first placed in a vacuum oven to remove air bubbles, and then thermally cured at 65 °C for 12 h. Rectangular slabs of suitable sizes are cut from the resulting 2
cured piece. Prior to the transfer of the CNT film, PDMS substrates are cleaned by ultrasonication in isopropyl alcohol for 5 min and dried under flowing N2. For transfer, a PDMS strip is brought into conformal contact with CNT/glass substrate. After brief contact, the PDMS substrate is peeled from the glass. The CNT film is then adhered to the PDMS substrate. Preparation of ion gel. The ion gel solution is prepared by co-dissolving the triblock copolymer poly(styrene-block methylmethacrylate-block-styrene) (PS-PMMA-PS, Polymer Source Inc. MPS = 4.3 kg/mol, MPMMA = 12.5 kg/mol, MW = 21.1 kg/mol) and the ionic liquid 1-ethyl-3-methylimidazolium bis(trifluoromethylsulfonyl)imide ([EMIM][TFSI], Sigma Aldrich) in ethyl acetate (Sigma Aldrich), following procedures adapted from Pu et al.4 The weight ratio of the polymer, ionic liquid, and solvent are maintained at 0.7:9.3:20. This mixture is stirred for 4 hr at 1000 rpm. The well-mixed solution is then filtered using 200 nm poly(tetrafluoroethylene) filters before use. Device fabrication. After the CNT films are transferred and buckled, Au/Cr (25 nm/4 nm) electrodes are deposited through a shadow mask using a thermal evaporator onto substrates restretched to a strain of 15%. Following the release of the strain, the substrates are heated at 60 °C in a glove box and ion gel solution is drop-cast onto the channel area as well as the area between the gate electrodes and channels. The devices are then heated at 105 °C in the glovebox for 1 hr to remove residual solvent and water. Device characterization. SEM imaging is conducted using a LEO-1550VP field-emission scanning electron microscope at 3 kV of accelerating voltage. Atomic force microscopy (AFM) imaging is performed using Veeco MultiMode scanning probe microscopy. Raman spectroscopy is conducted on an Aramis Horiba Jobin Yvon Confocal Raman Microscope with 532 nm laser light, a 1 μm2 spot size, and in a VV configuration with both the laser and the detected Raman signal polarized in the same direction. Angle-dependent scans are performed using a rotation stage, in which the scanned area and focus vary slightly with each scan; strain-dependent scans are performed with two perpendicular VV scans, where the scanned area and focus are the same for each scan. The capacitance of the ion gel is measured using an
3
Agilent LCR Meter (4284A). The devices are stretched by a uniaxial tensile stage with two sliders driven by a single right and left-hand threaded lead screw. Electrical characteristics are measured on a probe station by a Keithley Sourcemeter (2636A) in air.
4
Buckling of a CNT film transferred to an unstrained PMDS substrate Fig. S1a shows a SEM image of a CNT film after transfer onto an unstrained PDMS substrate. After two stretch-and-release cycles, periodic buckles appear on the surface of the CNT film (Fig. S1b), in contrast to the unbuckled surface before stretching. Fig. S1c and d show SEM micrographs of an already buckled CNT film on a PDMS substrate subsequently strained by 10% and 20%, respectively, and imaged in the stretched state. Initially, the wrinkles extend perpendicular to the direction of tensile strain. However, these wrinkles gradually disappear with increased strain (Fig. S1c) and eventually develop to extend parallel to the strain (Fig. S1d). a
d d
20%
b
0%
c
10%
Figure S1. a,b, SEM images of the transferred CNT film before (a) and after buckling (b). c,d, SEM images of the already buckled CNT film re-strained by 10% (c), and 20% (d).
5
Formation of cracks and tears in CNT films when overstretched Fig. S2 shows SEM images of a CNT film on a PDMS substrate (prepared using the procedure outlined in Fig. 1 by transferring CNTs to unstrained PDMS) after stretching to 30% and release. In addition to the wrinkles, cracks and tears develop in the film at this strain, which decrease the performance of the FETs. Stretching to > 50% is possible without tears using the alternative method in which CNTs are transferred to strained PDMS in order to increase the initial degree of buckling (as demonstrated in Figs. 3-5).
Cracks
Figure S2. SEM images showing the formation of cracks and tears after the CNT film is stretched and released from 30%.
6
Stretching of CNT FETs Figure S3 shows the optical photographs of CNT FETs before and after stretching to 15% along the longitudinal direction of the channel by a uniaxial tensile stage.
a
b
Figure S3. a,b, Optical photographs of CNT FETs before (a) and after (b) stretching to 15% strain.
7
Transconductance versus applied strain for the stretchable CNT FETs in Fig. 2 Fig. S4 plots the transconductance of the stretchable CNT FET from Fig. 2 at VD = -0.1 V as a function of applied strain. The transconductance gradually decreases with applied strain up to 18%. However, when the strain increases to 20%, the transconductance drops significantly (likely due to crack or tear formation as shown above in Fig. S2).
Figure S4. Transconductance versus applied strain for the CNT FET in Fig. 2.
8
Capacitance change of ion gel film versus tensile strain Figure S5a plots the change of the capacitance of the ion gel as a function of strain during the stretch-and-release process. The capacitance is measured by a LCR meter at 20 Hz. The capacitance approximately linearly decreases with applied strain. The subsequent stretch-andrelease curves follow the same path, showing the reproducibility of the ion gel under cyclic loading and unloading. Previously, it has been observed that by depositing the gel dielectric and gate electrode directly on top of a MoS2 FET channel to create a vertical MoS2 / gel / gate electrode sandwich structure, it is possible to achieve a capacitance that does not vary with strain. 5 This vertical configuration, or a configuration in which the lateral distance of separation between the gate electrode and the channel is reduced, may ultimately be more desirable in order to prevent the decrease in on current with strain. It should also be noted the ion gel film is observed to fracture in some cases for strain as low as 45% (Fig. S5b). Fracture typically occurs between strain of 45 and 60%.
a
b
Crack
Figure S5. a, Change of specific capacitance of ion gel versus strain. b, Optical image showing a tear in the ion gel film when stretched to 45% strain. 9
AFM characterization Figure S6a, b show an AFM image and a corresponding height profile of a buckled CNT film transferred to a PDMS substrate with a prestrain of 50%. It can be seen that the wrinkles vary in height from 10 to 100 nm and have an average period of ~200 nm.
a
b
Figure S6. AFM characterization of a buckled CNT film transferred to a PDMS substrate with a prestrain of 50%. a, Height AFM image. b, The corresponding line-scan profile.
10
Polarized Raman spectroscopy Polarized Raman spectroscopy is utilized to characterize the change in the morphology of the CNT films under strain and after release. Figure S7a plots the G-band intensity as a function of the angle between the polarization of colinear Raman excitation and detection with respect to the direction transverse to the prestrain, for a CNT film following transfer to prestrained PDMS (50%) and strain release. During the measurement, both the laser and the detected Raman signal are polarized in the same direction. In this case, theoretically, the Raman G-band of a CNT exhibits a polarization dependence ~cos4(Θ), where Θ is the angle between the Raman polarization and the CNT orientation6. In Fig. S7a, 0° and 180° on the x-axis corresponds to when the laser and detected Raman signal are both polarized perpendicular to the direction of PDMS prestrain and the corresponding G-band intensity is denoted (I). In Fig. S7a, 90° and 270° on the x-axis corresponds to when the laser and detected Raman signal are both polarized parallel to the direction of PDMS prestrain and the corresponding G-band intensity is denoted (I||). It can be seen that I >> I||. Figure S7b shows I / I|| as a function of applied strain as the PDMS substrate is subsequently stretched again. I / I|| is maximized at no strain and decreases as the PDMS is stretched, reaching 1 at a strain of 50%, which means that the CNT film becomes isotropic. For subsequent stretching > 50%, I / I|| < 1.
11
a
b
. Figure S7. Polarized Raman spectroscopy. a, G-band intensity as a function of the angle between the polarization of colinear Raman excitation and detection with respect to the direction transverse to the prestrain, for a CNT film following transfer to prestrained PDMS (50%) and strain release. I|| is the G-band intensity when the Raman laser and detector are polarized along the direction of prestrain. Iis the G-band intensity when the Raman laser and detector are polarized transverse to the direction of prestrain. b, I / I|| as a function of the applied strain as the PDMS substrate is subsequently stretched again.
12
Characteristics of CNT FETs fabricated from CNTs transferred to strained PDMS The output characteristics of a CNT FET fabricated from CNTs transferred to a strained PDMS substrate (via the approach presented in Fig. 3) at six different gate voltages (VG) are compared in Figure S8a (in the relaxed state prior to re-stretching). The device shows the expected gate modulation of the drain current (ID) in both the linear and saturation regimes. The transfer characteristics in Figure S8b indicate that the CNT FET operates as a typical p-type FETs. The on/off current ratio reaches 3×104 at VD=-0.5V. The field-effect mobility is calculated to be 6.9 cm2/(V·S) at VD = -0.1V. b
a
Figure S8. Characteristics of the CNT FET fabricated from CNTs transferred to strained PDMS, in the relaxed state prior to re-stretching. a, Typical output characteristics. b, Typical transfer characteristics.
13
Transconductance versus strain for the device in Fig. 4. Figure S9 plots the transconductance (VD= -0.1 V) of the device presented in Fig. 4 as a function of applied strain. It can be seen that the transconductance gradually decreases with increasing applied strain up to 50.8%. However, the transconductance drops significantly when the strain increases from 50.8% to 57.2%.
Figure S9. Transconductance versus strain for the device in Fig.4.
14
Failure We examine the stretchability of all three components of our devices individually to better understand device failure. For the ion gel dielectric layer, the stretchability is discussed above (Fig. S5). For the source, drain and gate electrodes, buckled Au/Cr films are used, which have been reported to be highly stretchable on PDMS substrates. Their applications as stretchable conductors or interconnects have been well studied7, 8. Here, we measure the resistance of buckled Au/Cr films as a function of the applied strain (Fig. S10a). With prestrain of 15%, the buckled Au/Cr films exhibit stable resistance up to strain of 20%. When the strain increases to 65%, the resistance increases by only 4.3 times. In all cases, the electrode resistance is substantially less than channel resistance and thus the electrodes are not the limiting factor in any of the devices presented, here. Regarding the stretchability of the CNT channel, the electrical properties of CNT films connected by two buckled Au/Cr electrodes are also characterized during the stretching process. For a CNT film buckled with the first approach (Fig. 1a), the resistance increases rapidly after the strain exceeds 27% (Fig. S10b), which is attributed to the fracture of the CNT film, as shown in Fig. S2. The point of fracture varies with the CNT film thickness, density, and uniformity. For CNT films buckled with the second approach (Fig. 3a), the resistance gradually decreases with the increase of strain, as shown in Fig. S10c. During the stretching process, the CNT alignment may improve (Fig. S7), leading to the increase of the contact area between CNTs or their bundles and a decrease in resistance. Thus, it can then be concluded that the failure of the FETs is not caused by failure of the electrodes or the CNT films but rather the fracture of the ion gel (Fig. S5b).
15
a
b
c
Figure S10. a,b,c, Resistance change of the buckled Au/Cr film (a), CNT film buckled with the first (Fig. 1a) approach versus strain (b), and the CNT film buckled with the second (Fig. 3a) approach (c) as a function of applied strain.
16
Demonstration of stretchabiltiy Figure S11 shows the transfer characteristics of the FETs adhered to a textile before and after stretching to 30%. The FETs are fabricated from CNTs transferred to prestrained PDMS (50%).
Figure S11. Transfer characteristics of the FETs adhered to a textile before and after stretching to 30%.
Supporting References 1. Nish, A.; Hwang, J. Y.; Doig, J.; Nicholas, R. J., Nature Nanotech. 2007, 2, 640. 2. Shea, M. J.; Arnold, M. S., Appl. Phys. Lett. 2013, 102. 3. Schoppler, F.; Mann, C.; Hain, T. C.; Neubauer, F. M.; Privitera, G.; Bonaccorso, F.; Chu, D. P.; Ferrari, A. C.; Hertel, T., J. Phys. Chem. C 2011, 115, 14682. 4. Pu, J.; Yomogida, Y.; Liu, K. K.; Li, L. J.; Iwasa, Y.; Takenobu, T., Nano Lett. 2012, 12, 4013. 5. Pu, J.; Zhang, Y. J.; Wada, Y.; Wang, J. T. W.; Li, L. J.; Iwasa, Y.; Takenobu, T., Appl. Phys. Lett. 2013, 103, 023505 6. Gommans, H. H.; Alldredge, J. W.; Tashiro, H.; Park, J.; Magnuson, J.; Rinzler, A. G., J. Appl. Phys. 2000, 88, 2509. 7. Lacour, S. P.; Wagner, S.; Huang, Z. Y.; Suo, Z., Appl. Phys. Lett. 2003, 82, 2404. 8. Lacour, S. P.; Jones, J.; Wagner, S.; Li, T.; Suo, Z. G., Proc. IEEE 2005, 93, 1459.
17
Highly Stretchable Carbon Nanotube Transistors with Ion Gel Gate Dielectrics Feng Xu†, Meng-Yin Wu‡, Nathaniel Safron†, Susmit Singha Roy†, Robert M. Jacobberger†, Dominick J. Bindl†, Jung-Hun Seo‡, Tzu-Hsuan Chang‡, Zhenqiang Ma‡, Michael S. Arnold†,* †
Department of Materials Science and Engineering and ‡Department of Electrical and Computer
Engineering, University of Wisconsin-Madison, Madison, WI 53706, USA
1
Methods Semiconducting carbon nanotube isolation. Semiconducting single-walled carbon nanotubes (CNTs) are prepared by methods adapted from those of Nish et al.1 and detailed in our previous work2. Solutions of 1 mg/mL CNTs (SouthWest Nanotechnologies, SG65i Lot# SG65i-L38) and 2 mg/mL poly(9,9dioctylfluorenyl-2,7-diyl) (PFO, American Dye Source) in toluene are ultrasonicated using a ½ inch tip horn ultrasonicator for 30 minutes at 40% max power. The resulting PFO-CNT slurry is ultracentrifuged for 15 min at 300,000g in a swing bucket rotor (ThermoFisher TH-641) to remove CNT bundles, unselected CNT chiralities, and catalyst particles. The top 90% of the resulting supernatant is decanted and filtered through a 5.0 µm PVDF membrane filter. Toluene is selectively removed via vacuum-distillation, and the resulting PFO/CNT solid is redissolved into hot tetrahydrofuran (THF). CNTs are selectively sedimented from the PFO-rich THF solution by centrifugation in a fixed-angle rotor at 50,000g at 4 °C for 24 hours. The PFO-rich supernatant is discarded and the semiconducting CNT-rich pellet is redispersed in THF in order to iterate this PFO-removal process, until the pellet contains CNTs in a PFO:CNT weight ratio less than 2:1, as determined via optical absorption, using calibration standards for the optical cross section of PFO and optical cross sections for the CNT S1 transition as reported by Hertel et al.3. Immediately prior to film casting, the CNT solutions are briefly agitated using microtip ultrasonication for 10 seconds at 10% max amplitude. Semiconducting carbon nanotube film deposition and transfer. Initially, PFO wrapped CNTs dispersed in chlorobenzene are doctor-blade cast onto clean glass on a hot-plate set to 110 °C in an argon glovebox, followed by annealing at 150 °C for 10 min to remove residual solvent. The spatially averaged thickness of the CNT film is roughly 5 nm, as determined by optical absorption spectroscopy using a thicker, calibration film that is measured using both profilometry and via absorption spectroscopy.
PDMS
substrates with thickness of 0.5 mm are prepared using Sylgard 184 (Dow Corning) by mixing the “base” and the “curing agent” at a ratio of 10:1. The mixture is first placed in a vacuum oven to remove air bubbles, and then thermally cured at 65 °C for 12 h. Rectangular slabs of suitable sizes are cut from the resulting 2
cured piece. Prior to the transfer of the CNT film, PDMS substrates are cleaned by ultrasonication in isopropyl alcohol for 5 min and dried under flowing N2. For transfer, a PDMS strip is brought into conformal contact with CNT/glass substrate. After brief contact, the PDMS substrate is peeled from the glass. The CNT film is then adhered to the PDMS substrate. Preparation of ion gel. The ion gel solution is prepared by co-dissolving the triblock copolymer poly(styrene-block methylmethacrylate-block-styrene) (PS-PMMA-PS, Polymer Source Inc. MPS = 4.3 kg/mol, MPMMA = 12.5 kg/mol, MW = 21.1 kg/mol) and the ionic liquid 1-ethyl-3-methylimidazolium bis(trifluoromethylsulfonyl)imide ([EMIM][TFSI], Sigma Aldrich) in ethyl acetate (Sigma Aldrich), following procedures adapted from Pu et al.4 The weight ratio of the polymer, ionic liquid, and solvent are maintained at 0.7:9.3:20. This mixture is stirred for 4 hr at 1000 rpm. The well-mixed solution is then filtered using 200 nm poly(tetrafluoroethylene) filters before use. Device fabrication. After the CNT films are transferred and buckled, Au/Cr (25 nm/4 nm) electrodes are deposited through a shadow mask using a thermal evaporator onto substrates restretched to a strain of 15%. Following the release of the strain, the substrates are heated at 60 °C in a glove box and ion gel solution is drop-cast onto the channel area as well as the area between the gate electrodes and channels. The devices are then heated at 105 °C in the glovebox for 1 hr to remove residual solvent and water. Device characterization. SEM imaging is conducted using a LEO-1550VP field-emission scanning electron microscope at 3 kV of accelerating voltage. Atomic force microscopy (AFM) imaging is performed using Veeco MultiMode scanning probe microscopy. Raman spectroscopy is conducted on an Aramis Horiba Jobin Yvon Confocal Raman Microscope with 532 nm laser light, a 1 μm2 spot size, and in a VV configuration with both the laser and the detected Raman signal polarized in the same direction. Angle-dependent scans are performed using a rotation stage, in which the scanned area and focus vary slightly with each scan; strain-dependent scans are performed with two perpendicular VV scans, where the scanned area and focus are the same for each scan. The capacitance of the ion gel is measured using an
3
Agilent LCR Meter (4284A). The devices are stretched by a uniaxial tensile stage with two sliders driven by a single right and left-hand threaded lead screw. Electrical characteristics are measured on a probe station by a Keithley Sourcemeter (2636A) in air.
4
Buckling of a CNT film transferred to an unstrained PMDS substrate Fig. S1a shows a SEM image of a CNT film after transfer onto an unstrained PDMS substrate. After two stretch-and-release cycles, periodic buckles appear on the surface of the CNT film (Fig. S1b), in contrast to the unbuckled surface before stretching. Fig. S1c and d show SEM micrographs of an already buckled CNT film on a PDMS substrate subsequently strained by 10% and 20%, respectively, and imaged in the stretched state. Initially, the wrinkles extend perpendicular to the direction of tensile strain. However, these wrinkles gradually disappear with increased strain (Fig. S1c) and eventually develop to extend parallel to the strain (Fig. S1d). a
d d
20%
b
0%
c
10%
Figure S1. a,b, SEM images of the transferred CNT film before (a) and after buckling (b). c,d, SEM images of the already buckled CNT film re-strained by 10% (c), and 20% (d).
5
Formation of cracks and tears in CNT films when overstretched Fig. S2 shows SEM images of a CNT film on a PDMS substrate (prepared using the procedure outlined in Fig. 1 by transferring CNTs to unstrained PDMS) after stretching to 30% and release. In addition to the wrinkles, cracks and tears develop in the film at this strain, which decrease the performance of the FETs. Stretching to > 50% is possible without tears using the alternative method in which CNTs are transferred to strained PDMS in order to increase the initial degree of buckling (as demonstrated in Figs. 3-5).
Cracks
Figure S2. SEM images showing the formation of cracks and tears after the CNT film is stretched and released from 30%.
6
Stretching of CNT FETs Figure S3 shows the optical photographs of CNT FETs before and after stretching to 15% along the longitudinal direction of the channel by a uniaxial tensile stage.
a
b
Figure S3. a,b, Optical photographs of CNT FETs before (a) and after (b) stretching to 15% strain.
7
Transconductance versus applied strain for the stretchable CNT FETs in Fig. 2 Fig. S4 plots the transconductance of the stretchable CNT FET from Fig. 2 at VD = -0.1 V as a function of applied strain. The transconductance gradually decreases with applied strain up to 18%. However, when the strain increases to 20%, the transconductance drops significantly (likely due to crack or tear formation as shown above in Fig. S2).
Figure S4. Transconductance versus applied strain for the CNT FET in Fig. 2.
8
Capacitance change of ion gel film versus tensile strain Figure S5a plots the change of the capacitance of the ion gel as a function of strain during the stretch-and-release process. The capacitance is measured by a LCR meter at 20 Hz. The capacitance approximately linearly decreases with applied strain. The subsequent stretch-andrelease curves follow the same path, showing the reproducibility of the ion gel under cyclic loading and unloading. Previously, it has been observed that by depositing the gel dielectric and gate electrode directly on top of a MoS2 FET channel to create a vertical MoS2 / gel / gate electrode sandwich structure, it is possible to achieve a capacitance that does not vary with strain. 5 This vertical configuration, or a configuration in which the lateral distance of separation between the gate electrode and the channel is reduced, may ultimately be more desirable in order to prevent the decrease in on current with strain. It should also be noted the ion gel film is observed to fracture in some cases for strain as low as 45% (Fig. S5b). Fracture typically occurs between strain of 45 and 60%.
a
b
Crack
Figure S5. a, Change of specific capacitance of ion gel versus strain. b, Optical image showing a tear in the ion gel film when stretched to 45% strain. 9
AFM characterization Figure S6a, b show an AFM image and a corresponding height profile of a buckled CNT film transferred to a PDMS substrate with a prestrain of 50%. It can be seen that the wrinkles vary in height from 10 to 100 nm and have an average period of ~200 nm.
a
b
Figure S6. AFM characterization of a buckled CNT film transferred to a PDMS substrate with a prestrain of 50%. a, Height AFM image. b, The corresponding line-scan profile.
10
Polarized Raman spectroscopy Polarized Raman spectroscopy is utilized to characterize the change in the morphology of the CNT films under strain and after release. Figure S7a plots the G-band intensity as a function of the angle between the polarization of colinear Raman excitation and detection with respect to the direction transverse to the prestrain, for a CNT film following transfer to prestrained PDMS (50%) and strain release. During the measurement, both the laser and the detected Raman signal are polarized in the same direction. In this case, theoretically, the Raman G-band of a CNT exhibits a polarization dependence ~cos4(Θ), where Θ is the angle between the Raman polarization and the CNT orientation6. In Fig. S7a, 0° and 180° on the x-axis corresponds to when the laser and detected Raman signal are both polarized perpendicular to the direction of PDMS prestrain and the corresponding G-band intensity is denoted (I). In Fig. S7a, 90° and 270° on the x-axis corresponds to when the laser and detected Raman signal are both polarized parallel to the direction of PDMS prestrain and the corresponding G-band intensity is denoted (I||). It can be seen that I >> I||. Figure S7b shows I / I|| as a function of applied strain as the PDMS substrate is subsequently stretched again. I / I|| is maximized at no strain and decreases as the PDMS is stretched, reaching 1 at a strain of 50%, which means that the CNT film becomes isotropic. For subsequent stretching > 50%, I / I|| < 1.
11
a
b
. Figure S7. Polarized Raman spectroscopy. a, G-band intensity as a function of the angle between the polarization of colinear Raman excitation and detection with respect to the direction transverse to the prestrain, for a CNT film following transfer to prestrained PDMS (50%) and strain release. I|| is the G-band intensity when the Raman laser and detector are polarized along the direction of prestrain. Iis the G-band intensity when the Raman laser and detector are polarized transverse to the direction of prestrain. b, I / I|| as a function of the applied strain as the PDMS substrate is subsequently stretched again.
12
Characteristics of CNT FETs fabricated from CNTs transferred to strained PDMS The output characteristics of a CNT FET fabricated from CNTs transferred to a strained PDMS substrate (via the approach presented in Fig. 3) at six different gate voltages (VG) are compared in Figure S8a (in the relaxed state prior to re-stretching). The device shows the expected gate modulation of the drain current (ID) in both the linear and saturation regimes. The transfer characteristics in Figure S8b indicate that the CNT FET operates as a typical p-type FETs. The on/off current ratio reaches 3×104 at VD=-0.5V. The field-effect mobility is calculated to be 6.9 cm2/(V·S) at VD = -0.1V. b
a
Figure S8. Characteristics of the CNT FET fabricated from CNTs transferred to strained PDMS, in the relaxed state prior to re-stretching. a, Typical output characteristics. b, Typical transfer characteristics.
13
Transconductance versus strain for the device in Fig. 4. Figure S9 plots the transconductance (VD= -0.1 V) of the device presented in Fig. 4 as a function of applied strain. It can be seen that the transconductance gradually decreases with increasing applied strain up to 50.8%. However, the transconductance drops significantly when the strain increases from 50.8% to 57.2%.
Figure S9. Transconductance versus strain for the device in Fig.4.
14
Failure We examine the stretchability of all three components of our devices individually to better understand device failure. For the ion gel dielectric layer, the stretchability is discussed above (Fig. S5). For the source, drain and gate electrodes, buckled Au/Cr films are used, which have been reported to be highly stretchable on PDMS substrates. Their applications as stretchable conductors or interconnects have been well studied7, 8. Here, we measure the resistance of buckled Au/Cr films as a function of the applied strain (Fig. S10a). With prestrain of 15%, the buckled Au/Cr films exhibit stable resistance up to strain of 20%. When the strain increases to 65%, the resistance increases by only 4.3 times. In all cases, the electrode resistance is substantially less than channel resistance and thus the electrodes are not the limiting factor in any of the devices presented, here. Regarding the stretchability of the CNT channel, the electrical properties of CNT films connected by two buckled Au/Cr electrodes are also characterized during the stretching process. For a CNT film buckled with the first approach (Fig. 1a), the resistance increases rapidly after the strain exceeds 27% (Fig. S10b), which is attributed to the fracture of the CNT film, as shown in Fig. S2. The point of fracture varies with the CNT film thickness, density, and uniformity. For CNT films buckled with the second approach (Fig. 3a), the resistance gradually decreases with the increase of strain, as shown in Fig. S10c. During the stretching process, the CNT alignment may improve (Fig. S7), leading to the increase of the contact area between CNTs or their bundles and a decrease in resistance. Thus, it can then be concluded that the failure of the FETs is not caused by failure of the electrodes or the CNT films but rather the fracture of the ion gel (Fig. S5b).
15
a
b
c
Figure S10. a,b,c, Resistance change of the buckled Au/Cr film (a), CNT film buckled with the first (Fig. 1a) approach versus strain (b), and the CNT film buckled with the second (Fig. 3a) approach (c) as a function of applied strain.
16
Demonstration of stretchabiltiy Figure S11 shows the transfer characteristics of the FETs adhered to a textile before and after stretching to 30%. The FETs are fabricated from CNTs transferred to prestrained PDMS (50%).
Figure S11. Transfer characteristics of the FETs adhered to a textile before and after stretching to 30%.
Supporting References 1. Nish, A.; Hwang, J. Y.; Doig, J.; Nicholas, R. J., Nature Nanotech. 2007, 2, 640. 2. Shea, M. J.; Arnold, M. S., Appl. Phys. Lett. 2013, 102. 3. Schoppler, F.; Mann, C.; Hain, T. C.; Neubauer, F. M.; Privitera, G.; Bonaccorso, F.; Chu, D. P.; Ferrari, A. C.; Hertel, T., J. Phys. Chem. C 2011, 115, 14682. 4. Pu, J.; Yomogida, Y.; Liu, K. K.; Li, L. J.; Iwasa, Y.; Takenobu, T., Nano Lett. 2012, 12, 4013. 5. Pu, J.; Zhang, Y. J.; Wada, Y.; Wang, J. T. W.; Li, L. J.; Iwasa, Y.; Takenobu, T., Appl. Phys. Lett. 2013, 103, 023505 6. Gommans, H. H.; Alldredge, J. W.; Tashiro, H.; Park, J.; Magnuson, J.; Rinzler, A. G., J. Appl. Phys. 2000, 88, 2509. 7. Lacour, S. P.; Wagner, S.; Huang, Z. Y.; Suo, Z., Appl. Phys. Lett. 2003, 82, 2404. 8. Lacour, S. P.; Jones, J.; Wagner, S.; Li, T.; Suo, Z. G., Proc. IEEE 2005, 93, 1459.
17
