advances.sciencemag.org/cgi/content/full/3/9/e1700159/dc1 Supplementary Materials for Ultratransparent and stretchable graphene electrodes Nan Liu, Alex Chortos, Ting Lei, Lihua Jin, Taeho Roy Kim, Won-Gyu Bae, Chenxin Zhu, Sihong Wang, Raphael Pfattner, Xiyuan Chen, Robert Sinclair, Zhenan Bao The PDF file includes: Published 8 September 2017, Sci. Adv. 3, e1700159 (2017) DOI: 10.1126/sciadv.1700159 Experimental section Additional supporting information fig. S1. Optical microscopy images of monolayer MGG on SiO2/Si substrates at different magnifications. fig. S2. SEM images of mono-, bi-, and trilayer MGGs on the SiO2/Si wafers. fig. S3. SEM images of graphene film covered with spray-coated CNTs. fig. S4. Comparison of two-probe sheet resistances and transmittances @550 nm of mono-, bi- and trilayer plain graphene (black squares), MGG (red circles), and CNTs (blue triangle). fig. S5. Sheet resistances of mono-, bi-, and trilayer MGGs. fig. S6. Optical transmittances of MGGs and multilayer plain graphene. fig. S7. Normalized resistance change of mono- and bilayer MGGs (black) and G (red) under ~1000 cyclic strain loading up to 40 and 90% parallel strain, respectively. fig. S8. Calculation of relative areal capacitance change as a function of strain. fig. S9. Optical microscopy image of trilayer MGG on SEBS elastomer. fig. S10. SEM image of trilayer MGG on SEBS elastomer after strain, showing a long scroll cross over several cracks. fig. S11. AFM images of various graphene structures on SEBS elastomer after 100% strain. fig. S12. AFM image of trilayer MGG on very thin SEBS elastomer at 20% strain, showing that a scroll crossed over a crack. fig. S13. Optical microscopy observation and simulation of graphenes on SEBS under strain.
fig. S14. Contact resistances of monolayer G/CNTs and Au/CNTs at different gate voltages. table S1. Mobilities of bilayer MGG single-walled carbon nanotube transistors at different channel lengths before and after strain. table S2. Summary of recent work on all-carbon transistors. References (54 58) Other Supplementary Material for this manuscript includes the following: (available at advances.sciencemag.org/cgi/content/full/3/9/e1700159/dc1) movie S1 (.mp4 format). Demonstration of stretchable LED control units by allcarbon transistors.
Experimental section Preparation of Plain Graphene CVD graphene was grown on Cu foils (Alfa Aesar, 99.999%) under a constant pressure of 0.5 mtorr with 50 sccm CH4 and 20 sccm H2 as precursors at 1000 C. Both sides of Cu foil were covered by monolayer graphene. To transport the sample, we clamped graphene/cu foil in between weighing paper and then glass slides. To maintain the flatness of graphene/cu foil at every step is critical in achieving clean transfer. A thin layer of PMMA (Microchem, A4, 2000 rpm) was spin-coated on one side of Cu foil, and O2 plasma (Micro, 200sccm air, 100 mtorr, 30 seconds) was used to etch the graphene on other side of Cu foil. Subsequently, the whole film was soaked in 0.1 mol/l ammonium persulfate ((NH4)2S2O8) solution for about 2 hours to etch away the Cu foil. The as-obtained PMMA/Graphene film was washed in DI-water several times, and laid onto target substrates. Right after the film was dried on the substrate, it was sequentially soaked in acetone, 1:1 acetone:ipa, and IPA for 30 seconds each to remove PMMA. To stack the following layers with repeated graphene transfer, the adhesion between the initial layer of graphene film and substrates is enhanced by either heating the sample at 100 C for 15 min or keeping it in vacuum overnight. This is to avoid the detachment of underlying graphene film from the substrates during PMMA removal. For a clean transfer, we normally chose to dry PMMA film in vacuum at room temperature, and afterwards soak the final graphene stacks in acetone for several days to remove PMMA residues as much as possible.
Preparation of G-CNTs-G Plain monolayer graphene was first transferred from Cu foil onto SiO2/Si wafers using aforementioned method. CNTs produced using the arc discharge method (P2-SWNT from Carbon Solutions) were dispersed in n-methyl 2-pyrrolidone using a probe sonicator (Cole Parmer Ultrasonicator 750 W) at 30% power for 30 minutes. The initial concentration of the mixture was 200 μg/ml. The resulting dispersion was centrifuged at 8000 rpm for 30 minutes, and the top 80% of the solution was aspirated for use in spraycoating. The CNT solution was spraycoated onto previous graphene covered SiO2/Si wafer at 200 C using a commercial airbrush (Master Airbrush, Model SB844- SET). The outlet of the airbrush kept a distance of approximately 15 cm with the surface of G/SiO2/Si substrate at a pressure of 35 psi. The density of CNTs is determined by the spraycoating time. The whole sample of CNTs/G was next left in a vacuum oven about 1 day to enhance its adhesion to substrates. A following graphene layer was transferred to form the sandwich composite of G-CNTs-G. Preparation of Semiconducting SWNTs Semiconducting SWNT solution was prepared using a modified procedure from our published work (53). 5 mg of poly[(9,9- di-n-dodecyl-2,7-fluorendiyl-dimethine)-(1,4- phenylene-dinitrilomethine)] (PF-PD) and 15 mg of raw SWNTs (RN-020, from Raymor Industries Inc.) were mixed in 25 ml of toluene and ultrasonicated for 30 min at an amplitude level of 50% while externally cooled with a dry ice bath. The solution was then centrifuged at 8000 rpm for 5 min and 17 000 rpm (22 000 g) for 25 min at 16 C. 80 % of the supernatants (20 ml) was collected and directly used for device fabrication. The
SWNT networks were fabricated by drop-casting the polymer sorted SWNT solution on SiO 2 wafer and then rinsed with toluene containing a small amount (1 % v/v) of trifluoroacetic acid to remove polymer residues. Toluene rinsing was used to degrade and remove the polymer residues. Morphology characterizations (optical microscopy and scanning electron microscopy) To characterize the morphology of multilayer graphene/graphene scrolls (MGG), we utilized OM, SEM and AFM. To observe the scroll distribution, we transferred MGG on Si substrates with 300-nm thermally-grown SiO2 on top. In optical microscope, the scrolls, showing as purple, are uniformly distributed over the graphene background (fig. S1). SEM was performed with an FEI Magellan 400 XHR scanning electron microscope operated at 1 kv, with a working distance of 3 mm. The surface of MGG is full of rolled up graphene scrolls and with stacked layer number increases, the scrolls became denser (fig. S2). fig. S1. Optical microscopy images of monolayer MGG on SiO2/Si substrates at different magnifications. Darker lines are graphene scrolls and lighter background is covered by monolayer graphene.
fig. S2. SEM images of mono-, bi-, and trilayer MGGs on the SiO2/Si wafers. In zoom-in images, representative scrolls and wrinkles are labeled to highlight their differences. Rolled-up graphene scrolls were replaced by spray-coated CNTs (fig. S3). Figure S3a is an overview image, with fig. S3b and c specifically zoomed into fully CNTs and partially CNTs covered regions.
fig. S3. SEM images of graphene film covered with spray-coated CNTs. (a) shows large-area CNTs/G film and (b,c) focus on all-cnts and partially CNTs- covered graphene regions. The red dash line in (c) marked the border of CNTs only and CNTs/G areas. Electrical and optical properties of MGG To compare resistances of MGG and plain graphene at strain, we first patterned them into strips (~ 300 μm wide and ~ 2000 μm long) by photolithography and O2 plasma on a Si substrate with 300-nm thermally-grown SiO2 on top and then deposited an array of Au electrodes by shadow mask and metallization method. The device array was next detached from the SiO2/Si substrate by etching SiO2 layer in BOE solution (HF:H2O 1:6) and transferred onto target elastomer substrate. To achieve a good contact during stretching test, additional macroscale liquid metal (EGaIn) was carefully connected from Au electrodes using toothpick. The entire sample was stretched in a manual apparatus and their 2-probe resistance changes were in-situ tested at strain perpendicular to the flow direction on a probe station with a semiconductor analyzer (Keithley 4200-SCS). We then measured the width and length of each strip and calculated their sheet resistances using the following formula
R s = R w l where R s is the 2-probe sheet resistance in the unit of Ω/sq. w and l are the measured width and length of the strip. Resistances and transmittances @ 550 nm of mono-, bi- and tri-layer plain graphene, MGG and CNTs-only were listed in fig. S4. Trilayer MGG has the best conductance with transparency of almost 90%. fig. S4. Comparison of two-probe sheet resistances and transmittances @550 nm of mono-, bi- and trilayer plain graphene (black squares), MGG (red circles), and CNTs (blue triangle). 4-probe sheet resistances were also measured on mono-, bi-, and tri-layer MGG samples. We randomly tested about 10 positions over the sample area of ~ 1 cm x 1 cm. The average sheet resistances of mono-, bi- and tri-layer MGGs are 185, 375, and 637 Ω/sq with coefficient of variations (CV = standard deviations/mean) of 7.2, 5.1 and 1.8%
respectively (fig. S5). The small CVs indicate the scrolls are uniformly distributed over the sample, enabling highly consistent stretchable electrodes. fig. S5. Sheet resistances of mono-, bi-, and trilayer MGGs. Figure S6 compares their optical transmittances of MGGs and multilayer plain graphene. fig. S6. Optical transmittances of MGGs and multilayer plain graphene. We also compared resistances of MGG and plain graphene at strain, which is parallel to the current flow (fig. 2C), and performed cycling tests (fig. S7). Graphenes (~ 5 mm wide and ~ 1 cm long) were transferred onto SEBS elastomer substrates, contact with liquid
metal (EGaIn) directly, and connected to a home-made stretching machine with automatic strain loading. For the samples with scrolls, graphene structures show less resistance change. fig. S7. Normalized resistance change of mono- and bilayer MGGs (black) and G (red) under ~1000 cyclic strain loading up to 40 and 90% parallel strain, respectively. Capacitance measurement gated by MGG under strain To fabricate stretchable capacitors with MGG as back gates, we first transferred MGG structures onto SEBS elastomer substrates and then covered them with SEBS dielectric layers (2-μm in thickness). The thin dielectric film was transferred from a very smooth hydrophobic OTS modified SiO2/Si surface by spin-coating a SEBS toluene (80 mg/ml) solution at 1000 rpm for 1 min. This OTS surface helped obtain uniform and pin-hole free thin film polymer dielectric. For top metal plate, to avoid metal penetration and easily determine the capacitor area, we deposited 5 nm Al/40 nm Au patterned by shadow mask followed by additional liquid metal as feasible contacts during strain test. An LCR
meter (Agilent) measured its capacitance gated by graphene in the frequency of 20 Hz. According to the following formula (57) C = ε r ε 0 A d (Equation 1) Where capacitance (C) is dependent on the area (A) and separation (d) of two electrodes. While stretching, relative areal capacitance is C = ε 1 A rε 0 d 0 (1 νε) (Equation 2) Where ε is strain and ν is poisson ratio. It will increase due to the decreased thickness of dielectric layer. Figure S8 calculates the ideal relative areal capacitance change as a function of strain. Because the area of the measured capacitor is much smaller than the stretched substrate, the real strain on the capacitor is less than the strain we applied on the whole substrate. Therefore, the actual slope of the capacitance change should be flatter than that in fig. S8.
fig. S8. Calculation of relative areal capacitance change as a function of strain. Morphological understanding of graphene after strain. We observed the surface of MGGs on elastomer using a variety of methods. As shown in fig. S9, the scroll on MGG is extremely difficult to be visualized under optical microscope on elastomer substrate due to the lack of color contrast. fig. S9. Optical microscopy image of trilayer MGG on SEBS elastomer. The top white rectangular region is bare SEBS substrate and the bottom red region is covered by trilayer MGG.
SEM detects secondary electrons emitted by atoms excited by electron beam. Under SEM, conductors are easier to conduct electrons showing as dark while nonconductors and the sharp edges tend to accumulate electrons showing as bright. Figure S10 is an SEM image of trilayer MGG on elastomer substrate after strain. Overall, trilayer MGG after strain is still conductive, showing as dark. The conductive scroll is dark as well with bright edges. Cracks perpendicular to the strain are bright. It is clearly observed that this scroll crossed over several cracks in the graphene, bridging the domains. However, samples on nonconductive polymer tend to charge and cause scanning faults and image artifacts, making it difficult to make conclusive observations. fig. S10. SEM image of trilayer MGG on SEBS elastomer after strain, showing a long scroll cross over several cracks. The scroll is arrowed as well as a typical crack. AFM reflects the morphology by touching the surface with a mechanical probe. Figure S11A-D are the AFM images of mono-, bi-, tri-layer MGGs after the underlying SEBS (~1 mm) are stretched up to 100%. Zoom-in image of fig. S11C shows a scroll crossed over a crack (fig. S12). In contrast, no scroll was observed on fig. S11E and its corresponding zoom-in image fig. S11F. Similarly, CNTs also bridged the cracks in graphene. Since it is not easy to distinguish CNTs on a rough surface in topography
image (fig. S11G), we observed them in the phase image. One bridging CNT was marked in fig. S11H. The existence of such 1D conductive structures must contribute to the electrical conductivity in particular when they are under strain. fig. S11. AFM images of various graphene structures on SEBS elastomer after 100% strain. Mono-, bi- and tri-layer G/G scrolls (A-C) bilayer G (E), G-CNT3-G (G) and the zoom-in images (D, F, H) corresponding to the marked regions. Representative cracks and scrolls are labeled. To observe the relative movement of scroll vs. underlying graphene, we specifically zoomed into a scroll-covered region at 20% strain (fig. S12). While the graphene cracks during stretching in order to accommodate strain, the scrolls are very likely not to crack at the same location, continuing to contribute a percolating pathway.
fig. S12. AFM image of trilayer MGG on very thin SEBS elastomer at 20% strain, showing that a scroll crossed over a crack. Simulation of graphene on SEBS under 20% strain Graphene has a much higher modulus than that of the SEBS substrate. Although the effective thickness of the graphene electrodes is much lower than that of the substrate, the stiffness of the graphene times its thickness is comparable to that of the substrate, resulting in a moderate rigid-island effect. We simulated the deformation of the graphene and substrate under a plane strain condition, when an external strain of 20% is applied on the SEBS substrate. Geometry of simulation is shown in fig. 4I and strain of simulation is up to 20%. This is because when strain is too large, crack will generate. The graphene electrodes are modeled by beam elements with an effective thickness 1nm, and the graphene is modeled as a linear elastic material with Young s modulus 0.9 TPa and Poisson s ratio 0.15 (43, 44). The SEBS substrate is modeled as an incompressible neo- Hookean material with Young s modulus around 6.23 MPa, and therefore shear modulus around 2.08 MPa. The simulation result shows that at an external strain of 20%, the average strain in the graphene electrode, defined as the elongation of the graphene divided by its original length, is 6.6% (fig. 4J). This indicates that the strain applied on
graphene electrode patterns is significantly confined, forming graphene stiff islands on top of SEBS (54 56). Optical microscope observation of graphene on SEBS under strain To verify the above simulation result, we made graphene patterns with 200 µm features and then stretched and looked at them at optical microscope (fig. S13a, b). At a designated strain, graphene elongation is always smaller than SEBS. Figure S14c summarized the length changes of graphene region and SEBS region at different strains. This observation agrees very well with the simulation result (fig. S13d), confirming graphene rigid island effect on SEBS that strain on graphene patterns is confined. fig. S13. Optical microscopy observation and simulation of graphenes on SEBS under strain. Optical microscope images of patterned graphene strips on SEBS at 0% (a) and 30% (b) strain. The darker strips labeled in red are graphene regions, while the lighter ones labeled in green are SEBS regions. (c,d) Experimentally observed (c) and simulated (d) strains applied on graphene and SEBS regions vs. overall strains on the whole substrate.
Additional supporting information Calculation of contact resistance Transfer-line method (TLM) was used to calculate the contact resistance. In the linear regime, the total resistance (R total ) of the channel should be (58) R total = R channel + R sd = L WμC i (V GS V Th ) + R sd (Equation 4) Where R channel and R sd are the channel resistance and contact resistance, μ is the linear mobility, L and W are the channel length and width, C is the gate capacitance. At different (V GS V Th ), if we plot R total W as a function of channel length L, the intercept to the y-axis gives R sd W. We fabricated CNTs transistors with monolayer graphene and evaporated Au film as top contacts on 300 nm-sio2/si substrates and compared their contact resistances (fig. S14). Monolayer graphene shows much better contact with CNTs than using Au as contacts.
fig. S14. Contact resistances of monolayer G/CNTs and Au/CNTs at different gate voltages. Mobility calculation of stretchable transistors The drain current in the saturation regime is given by I D = W 2L Cμ(V GS V T ) 2 (Equation 5) Where L and W are the channel length and width respectively, C is the gate capacitance, μ is the field-effect mobility, and V T is the threshold voltage. The square root of the saturation current could be plotted as a function of the gate voltage. The slope of the plotted straight-line gives mobility μ while its extrapolation to the V GS axis corresponds to the threshold voltage V T. C was measured aforementioned using a planar capacitor model, assuming a constant capacitance of 1 nf/cm 2 at strain up to 120%. Table S1 summarized mobilities and threshold voltages of bilayer MGG-SWNTs transistors at different channel lengths before and after strain.
table S1. Mobilities of bilayer MGG single-walled carbon nanotube transistors at different channel lengths before and after strain. Channel length (μm) On/off V th (V) Saturated mobility (cm 2 /vs) Before Strain After 105% Strain 100 2.5e3-7.3 5.2 150 2.5e3-8.1 5.6 200 1.5e3-7.1 5.0 100 3.7e3-5.3 3.3 150 3.7e3-5.8 4.1 200 2.1e3-4.4 3.3 table S2. Summary of recent work on all-carbon transistors. Publication Minggagng Xia et al, Appl. Phys. Lett. 105:143504, 2014. Feng Xu et al, Nano Lett, 14:682, 2014. Le Cai et al, ACS Nano, 10:11459, 2016. Alex Chortos et al, Adv. Mater, 28:4441, 2015. Jiajie Liang et al, Nat Commun, 6:7647, 2015. Atsuko Sekiguchi et al, Nano Lett, Mobility; on/off Elastic Transmittance Time Response ratio stretchability 23; 10 2 23% Low Fast 10; 10 5 60% (elastic) Low Slow (ion gel) 4; 500 60% (elastic) Low Fast; nonpolar dielectric 0.2; 10 4 100% (elastic) <60% Slow (ionic effects) 32.5; 10 4 50% (elastic) ~90% Slow (ionic effects) N/A; 10 5 110% (elastic) <60% Slow (ion gel) 15:5716, 2015. This work 6; 10 3 120 % (elastic) >90% Fast; nonpolar dielectric