A molecular design approach towards elastic and multifunctional polymer electronics

Next-generation wearable electronics require enhanced mechanical robustness and device complexity. Besides previously reported softness and stretchability, desired merits for practical use include elasticity, solvent resistance, facile patternability and high charge carrier mobility. Here, we show a molecular design concept that simultaneously achieves all these targeted properties in both polymeric semiconductors and dielectrics, without compromising electrical performance. This is enabled by covalently-embedded in-situ rubber matrix (iRUM) formation through good mixing of iRUM precursors with polymer electronic materials, and finely-controlled composite film morphology built on azide crosslinking chemistry which leverages different reactivities with C–H and C=C bonds. The high covalent crosslinking density results in both superior elasticity and solvent resistance. When applied in stretchable transistors, the iRUM-semiconductor film retained its mobility after stretching to 100% strain, and exhibited record-high mobility retention of 1 cm2 V−1 s−1 after 1000 stretching-releasing cycles at 50% strain. The cycling life was stably extended to 5000 cycles, five times longer than all reported semiconductors. Furthermore, we fabricated elastic transistors via consecutively photo-patterning of the dielectric and semiconducting layers, demonstrating the potential of solution-processed multilayer device manufacturing. The iRUM represents a molecule-level design approach towards robust skin-inspired electronics.

described below. All of the transistors were measured using a Keithley 4200 semiconductor parameter analyzer (Keithley Instruments Inc, Cleveland, OH, USA) under ambient atmosphere at room temperature. AFM-IR: Atomic force microscopy paired with infrared spectroscopy (AFM-IR) is a surface characterization technique capable of probing both nanoscale topography and local chemical composition. Like traditional scanning probe microscopy, AFM-IR rasters a nanoscale probe across the surface either in contact or tapping mode. As the probe tracks along the material surface, a pulsed, tunable infrared laser excites the scan area and causes rapid thermal expansion within the material. This thermal expansion is detected by the AFM probe which is then translated into local chemical topography. Local composition can also be acquired by collecting a broadband infrared spectrum on the surface of the material where the AFM probe is positioned over a user-specified location and the infrared laser sweeps through a range of infrared frequencies. Samples are measured using a nanoIR3 AFM-IR from Anasys Instruments (Santa Barbara, CA) coupled to a MIRcat-QT™ quantum cascade, mid-infrared laser by Daylight Solutions (frequency range of 917-1700 cm -1 and 1900-2230 cm -1 and repetition rate of 1,470 kHz). AFM-IR data are collected in tapping mode using a gold-coated, AFM probe (spring constant (k): 40 N/m and resonant frequency (fo): 300 kHz). The pulsed, mid-IR laser is tuned to frequencies unique to each component as determined by FTIR characterization. Acquired images are flattened using Analysis Studio software.

Molecular Dynamics Simulations
Molecular dynamics (MD) simulations were conducted to probe the mixing behavior of BA and DPPTT. MD simulations were conducted with the Gromacs 2018 program 3 . The force field is based on Optimized Potentials for Liquid Simulations all atom (OPLS-AA 4 ) 5 . Each BA chain has n, l, m = 9, 3, 10 with a unique random sequence of different monomers. The length of DPPTT is 5. Initially, polymer chains were randomly inserted into a large cubic box with a side length of 11 nm. Then the simulation box was compressed at a high pressure of 100 bar to reach a density close to the melt. After that, the system was simulated for 20 ns at a temperature of 430 K and a pressure of 1 bar. The snapshot was taken at the end of 20 ns simulation.

Thin-film Preparation and Characterization
(1) iRUM-s-x:y film preparation: 5 mg/mL DPPTT (or 10 mg/mL IDTBT) and 10 mg/mL BA stock solutions were first prepared with anhydrous chlorobenzene. Then, the BA solution and the DPPTT (or IDTBT) solution were mixed with a specific volume ratio to obtain BA/DPPTT (or IDTBT) blend with a designed weight ratio (x:y). (2) Preparation of crosslinked BA rubber film: Newly synthesized BA was drop-casted on OTS-modified substrate, and put in vacuum heater at RT for 5 days, 50 o C for 1 d, 60 o C for 1d, 80 o C for 2 d. The crosslinking process needs to be slow, otherwise bubbles will form and act as mechanical defects. The mechanical properties of as-prepared BA rubber film were tested by Instron. Film dimension: length 1.7mm, width 2.7mm, thickness 0.8mm.
(3) BAc/DPPTT-x:y crosslinked film preparation: 5 mg/mL DPPTT, 10 mg/mL BAc, and 25 mg/mL Azobisisobutyronitrile (AIBN) initiator stock solutions were first prepared with anhydrous chlorobenzene. Then, the BAc solution and the DPPTT solution were mixed with a specific volume ratio to obtain BAc/DPPTT blend with a designed weight ratio (x:y), followed by adding 5w% (4) BH/DPPTT-x:y crosslinked film preparation: 5 mg/mL DPPTT and 10 mg/mL BH stock solutions were first prepared with anhydrous chlorobenzene. Then, the BH solution and the DPPTT solution were mixed with a specific volume ratio to obtain BH/DPPTT blend with a designed weight ratio. The blend solution was heated and stir under nitrogen at 85 °C for 15 min. The solutions were subsequently spin-coated on OTS-modified substrates at 1500 r.p.m. for 1 minute and annealed at 150 °C for 1.5 h in nitrogen atmosphere. Film thickness is around 30-40 nm.
(5) BF/DPPTT-x:y blend film preparation: 5 mg/mL DPPTT and 10 mg/mL BF stock solutions were first prepared with anhydrous chlorobenzene. Then, the BF solution and the DPPTT solution were mixed with a specific volume ratio to obtain BF/DPPTT blend with a designed weight ratio. The blend solution was heated and stir under nitrogen at 85 °C for 15 min. The solutions were subsequently spin-coated on OTS-modified substrates at 1500 r.p.m. for 1 minute. Film thickness is around 30-40 nm.
(6) Photo-pattern iRUM-s-x:y film: The BA/DPPTT blend solution with a designed weight ratio was prepared following the procedures mentioned above. The solution was subsequently spin-coated on OTS-modified substrates at 1500 r.p.m. for 1 minute. The obtained film was photo-crosslinked in glovebox under UV light (254 nm, power 6W) for 1min (Dose: 270 mJ/cm 2 ) using shadow mask, and then developed in CHCl3 in fume hood for 30 s. The patterned semiconductor film was post-annealed at 150 °C for 30 min in glovebox, in order to fully crosslink and get rid of solvent.
(7) iRUM-d-x:y film preparation: 90 mg/mL SEBS and 100 mg/mL BH stock solutions were first prepared with anhydrous toluene. Then, the filtered solutions were mixed with a specific volume ratio to obtain BH/SEBS blend with a designed weight ratio (x:y). The blend solution was heated and stir under nitrogen at 85 °C for 15 min. The solution was subsequently spin-coated on highly doped Si substrates at 1000 r.p.m. for 1 minute and annealed at 200 °C for 1.5 h in nitrogen atmosphere.
(8) Photo-pattern iRUM-d-x:y film: The BH/SEBS blend solution with a designed weight ratio was prepared following the procedures mentioned above.
The solution was subsequently spin-coated on substrates at 1000 r.p.m. for 1 minute. The obtained film was photocrosslinked by exposure under deep ultraviolet light (wavelength 254 nm, using a Spectrum 100 Precision UV Spot Curing System from American Ultraviolet) for 30min with a dose of 126 J/cm 2 at ambient condition using shadow mask. After this, dodecane was used to dissolve the unexposed areas of SEBS/BH blend, with the photo-exposed areas preserved. The patterned dielectric film was post-annealed at 200 °C for 1 h in glovebox, in order to fully crosslink and get rid of solvent.
(9) Pseudo free-standing tensile tests (Film-on-water): The Si substrate is cleaned with deionized water and isopropanol, followed by 1min O2 plasma treatment. Then poly(sodium 4-styrenesulfonate), PSSNa solution (3 wt.% in deionized water) is spin-coated onto Si substrate at 4000 r.p.m. for 1 min, and annealing at 85 o C for 15 min (thickness: ~30 nm). The semiconductor (or dielectrics) film is directly prepared on top of the water-soluble PSSNa layer. The polymer films were first patterned into dogbone shape according to previous reports 6,7 , followed by slowly dipping into deionized water to release and float the semiconductor film by dissolving the underneath PSSNa layer. Later, the semiconductor (or dielectrics) film was hold with two aluminum tensile grips coated with a thin PDMS layer, and then tensile tests were performed.
For stress-strain measurements and cyclic tests, the strain rate is 0.05 s -1 . For stress relaxation, a strain rate of 0.03 s -1 is applied and the strain is kept within the elastic region (5%). The elastic modulus was obtained from the slope of the linear fit of the stress-strain curve using the first 0.5% strain (elastic region).
For cyclic tests and stress relaxation measurement, an additional thin layer of PDMS is directly prepared on top of the semiconductor film by spin-coating a dielectrics solution (2.8 g PDMS, base/crosslinker-6:1 in 10 mL hexane) at 5000 r.p.m. for 2 min, followed by annealing at 70 o C for 1h in air and annealing in glove box at 150 o C for 40 min. The resultant thickness for dielectrics is around 2.4 µm.
(2) "Stretchable transistor": The semiconductor film (neat conjugated polymer or iRUM-s) was prepared on OTS-treated 300 nm SiO2 substrate first. The solution for dielectrics was prepared by diluting 2.8 g PDMS (Sylgard 184 [6:1] base versus crosslinker in weight ratio) into 10 mL hexane, stir for at least 10min before spin-coating. Then the dielectric layer was prepared by directly spin-coating dielectric solution onto the semiconductor layer at 5000 r.p.m. for 2min, and anneal at 70 o C for 1h in air. Then the substrate with the two layers was further annealed in glove box at 150 o C for 40 min. The resultant thickness for dielectrics is around 2.4 µm as confirmed by a Profilometer (Bruker Dektak XT). Carbon nanotube (CNT) solution for gate electrode was prepared by dispersing P2-SWNT (0.2 mg/mL) with P3HT (0.05 mg/mL) into chloroform through ultra-sonication for 30 min at 30% amplitude using a 750 W ultra-sonication probe, followed by centrifugation at 893.8 × g for 30 min. Then the dispersed CNT solution was spray-coated onto OTS- (3) Fully patterned transistor array fabrication: A Si/SiO2 wafer was firstly cleaned with O2 plasma (150 W, 200 mTorr) for 1-2 min, and then PSSNa solution (3 wt. % in water) was spin-coated on top at 2500 r.p.m. for 1min. The wafer was then baked on a hot plate at 180 °C for 30 min and cooled down to 80 o C to fully get rid of water. Then iRUM-d was photo-patterned on top of PSSNa, following the procedures mentioned above, serving as the dielectrics with a thickness of about 1.2 μm. Subsequently, iRUM-s was photo-patterned on top of iRUM-d following the procedures mentioned above, serving as the semiconductor with a thickness of about 35 nm. For source/drain electrodes, CNT dispersed solutions (P3-SWNT, 0.3 mg/mL in isopropanol) were prepared and directly spray-coated onto the semiconductor through a shadow mask (L = 120 µm, W = 400 µm) to obtain the patterned source/drain electrodes. Then, a SEBS (H1221) stretchable substrate was laminated onto the fabricated devices on the Si/SiO2 substrate. Then, the array was transferred onto the substrate by immersing the entire device in water to dissolve the sacrificial PSSNa layer. Finally, the gate electrodes were patterned on top of the SEBS dielectric by spray-coating the same CNT dispersed solution through a shadow mask. Alignment of the shadow masks was performed under optical microscopy.
Difficult to be detected due to resolution To a 50 mL round-bottom flask were added 10 mL anhydrous DCM and 1.2 g hydroxyl-terminated polybutadiene (1 mmol). The mixture was cooled down to 0 o C and stirred for 5 min until homogeneous solution formed. At 0 o C, 290 mg 2-isocyanatoethyl acrylate (2.05 mmol) was first dissolved in 5 mL DCM and was added slowly into the stirring mixture. The solution was then allowed to warm up to room temperature and stirred for 24 h. After the reaction, the solvent was removed under vacuum (40 o C for 30 min). The product was cooled down to room temperature and dissolved in CB to form 25 mg mL -1 solution for further use (BAc can be kept by itself at -10 o C).
Yield: ~99%.  To a 250 mL double-neck round-bottom flask were added 100 mL anhydrous ethyl acetate (EA) and 4.8 g hydroxylterminated polybutadiene (4 mmol). The mixture was degassed by N2 and stirred for 30 min until homogeneous emulsion formed. Then 500 mg Pd/C (dry, type 487) was added to form suspension and the N2 atmosphere was purged with H2 balloons. The solution was stirred at room temperature for 96 h during which H2 balloons were frequently refilled to ensure enough H2 pressure. After the reaction, the suspension was filtered off and the solvent of obtained solution was removed under vacuum (50 o C for 1 h). The product was cooled down to room temperature to yield white, non-flowable wax hydrogenated hydroxyl-terminated polybutadiene. Yield: ~50%.
To a 50 mL round-bottom flask were added 10 mL anhydrous DCM and 620 mg hydrogenated hydroxyl-terminated polybutadiene (0.5 mmol). The mixture was stirred at room temperature for 5 min until homogeneous solution formed. 260 mg 4-azido-2,3,5,6-tetrafluorobenzoyl chloride (1.02 mmol) was first dissolved in 5 mL DCM and the solution was added slowly into the stirring mixture. The solution was then stirred at room temperature for 16 h.
After the reaction, the solvent was removed under vacuum (40 o C for 30 min). The product was cooled down to room temperature and dissolved in CB to form 100 mg mL -1 solution for further use. (BH is not as reactive as BA due to the lack of double bonds but it would be better to be kept in dilute solution state) Yield: ~99%.
Difficult to be detected due to resolution Supplementary Figures 1-68 Film dimension: length 2mm, width 2.7mm, thickness 0.8mm. The extracted Young's modulus is 6.86 MPa, while the strain at break is 145.85%. Cyclic stress-strain curves with the strain rate of (B) 20%/min and (C)100%/min. Film dimension: length 1.7mm, width 2.7mm, thickness 0.8mm. There is no residue strain after multiple loadingunloading cycles and the stress-strain curve at different cycle overlap with each other very well, indicating the elastic behavior with no strain-rate dependence. The first cycle is omitted due to the clamp sliding.

A B C
Supplementary Note 1. The rationale for good mixing.
The well blending of polybutadiene-derived precursors with conjugated polymers is the premise to enable our iRUM approach. This allows uniform crosslinking and prevents large-domain phase separation, which may result in stress concentration at domain boundaries and more likely to have mechanical failure under strain.
As confirmed by AFM and IR-AFM of BF/DPPTT-x:y blend films, no micrometer-sized domains due to crystallization was observed at different BF ratios. The high flexibility of polybutadiene and matched surface energy with semiconducting polymer contributes to their uniform mixing with conjugated polymer nanostructures, which are essential for high charge carrier mobility. However, in some small molecule/conjugated polymer blend systems, the small molecules have strong tendencies to crystallize, thus making large non-uniform phase separation more favorable.
BF/DPPTT was used as an analog to mimic the non-crosslinked blend and it showed similar morphology to the crosslinked one (iRUM-s-x:y, BAc/DPPTT-x:y and BH/DPPTT-x:y crosslinked films). The uniform distribution of DPPTT nanostructures was still maintained. Therefore, we assumed that the morphology of composite film was mainly determined during solution mixing and subsequent solution deposition (spin coating), which was not affected too much by crosslinking process. Furthermore, the low surface roughness of iRUM-s films is also beneficial when incorporated into electronic devices where the charge transport is across or along interfaces between different device components.
BA/DPPTT (wt/wt) 1:9 3:7 1:1 3:1 Crosslinking density (10 -4 mol/g) 1.2 3.6 6 9 Supplementary Table 1. Crosslinking density of iRUM-s films. The crosslinking density is calculated by using the molar percentage of azide over the total weight of composite film. Note: The association between DPPTT sidechain and BA is stronger than the association between DPPTT backbone and BA. As BA concentration increases, the association between BA with both DPPTT backbone and side chain decreases, which is due to the increased association among BA molecules. Figure 7. AFM height and phase images of DPPTT unannealed film and BF/DPPTT-x:y blend films, where x:y is the BF-to-DPPTT weight ratio. The surface roughness is obtained from AFM height image. Note: DPPTT+BF was used as an analog to mimic the non-crosslinked blend and it showed similar morphology to the crosslinked one, DPPTT+BA. Therefore, we assumed that the morphology of composite film was mainly determined during solution mixing and subsequent solution deposition (spin coating). Note: The different reactivity of BA and BH can be seen by the time needed for the azide characteristic peak to disappear in FTIR. The drop-casted BA and BH both have sharp azide stretch peak at 2120 cm -1 . After thermal crosslinking at 150 o C, the azide peak of BA completely disappeared. On the other hand, the azide peak of BH still remained, and it disappeared after thermal crosslinking for another one hour, which indicates that BH needs higher energy to be fully activated, thus a lower crosslinking efficiency. The same trend can be observed in BA/DPPTT and BH/DPPTT blends. Referring to this, we decided the thermal annealing condition for iRUM-s film and BH/DPPTT crosslinked film preparation. Note: a − * transition was recorded around 450nm for all the films. Their internal charge transfer peaks between 700 and 900 nm showed two distinct vibronic bands: the 0−1 and 0−0 transitions, with the lower energy 0−0 peak (higher wavenumber) typically attributed to DPPTT aggregation. When normalized to the 0−1 peak, an increase in the intensity of 0-0 peak and a slight red-shift are indicative of a higher degree of DPPTT aggregation. For BF/DPPTT-x:y blend films, with BF proportion increasing, DPPTT aggregation increases. This is due to the introduction of a secondary component into DPPTT film, resulting in higher interaction between DPPTT polymer chains, which is also observed in previously reported conjugated polymer/elastomer blending systems (ref). For BA/DPPTT-x:y, BAc/DPPTT-x:y and BH/DPPTT-x:y blend films, such an increase in DPPTT aggregation can also be observed. However, after crosslinking, DPPTT aggregation changed in different manners. BAc/DPPTT-x:y crosslinked films exhibited higher DPPTT aggregation than blend films, which is due to the formation of a rubber matrix that contributes to stronger DPPTT chain interactions than that from precursors. For BA/DPPTT-x:y (iRUMs-x:y) crosslinked films, a slight increase in DPPTT aggregation can also be observed than blend films. However, the degree of DPPTT aggregation increase is lower than that observed in BAc/DPPTT-x:y crosslinked films. This can be explained by the reactivity difference between azide/double-bond cycloaddition and azide/C-H insertion. A higher proportion of BA will react with the double bonds on itself to create a rubber matrix, contributing to increase in DPPTT aggregation. However, a small proportion of BA can also react with DPPTT side chains, thus disrupting DPPTT aggregation. This claim can be supported by the fact that BH/DPPTT-x:y crosslinked films exhibited obvious lower DPPTT aggregation than blend films. Without such reactivity difference, the DPPTT aggregation was continuously disrupted with BH proportion increasing.

Supplementary Note 2. GIXD analysis (Supplementary Figure 25-28, Supplementary Table 2-5).
GIXD provides information regarding the crystalline region of conjugated polymer thin film. As shown from 2D diffraction patterns, all of the blend/crosslinked films exhibit an edge-on orientation, with out-of-plane (h00) lamella stacking peaks and in-plane (010) π-π stacking peak. Compared with neat DPPTT film (with or without thermal annealing), there is negligible change in π-π spacing (~3.6 Å) but an increase in lamella spacing, suggesting the more favorable interaction between polybutadiene-derived precursors with DPPTT alkyl side chains than conjugated backbones. However, this change in lamella spacing is small relative to the length of polybutadienederived precursors, thus, it is unlikely that the precursors inserted into the crystalline region. Instead, they are distributed mostly into the amorphous regions of conjugated polymers. For BF/DPPTT-x:y blend films, a decrease in FWHM of the (200) lamella stacking peak is observed with BF proportion increasing, which corresponds to an increase in coherence length. For the same value of x:y (precursor/DPPTT), the coherence length of BH/DPPTTx:y crosslinked film is smaller than that of iRUM-s-x:y and BAc/DPPTT-x:y crosslinked films. This supports the claim that a higher proportion of azide/C-H insertion undergoes after blending DPPTT with BH than BA, thus chain aggregation and packing of polymer semiconductor is disrupted more in BH/DPPTT-x:y crosslinked films than iRUM-s-x:y films.
Supplementary Figure 30. Crack on-set strains of DPPTT and iRUM-s-x:y films, as measured by "film-onelastomer" using optical microscope.
Note: "Film-on-elastomer": the semiconductor film is transferred onto a supported PDMS substrate (1 mm), and its crack formation is monitored under optical microscope with increasing strains. This measurement is relevant, as the semiconductor film is usually deformed on a supported dielectric in electronic devices. Our parameter of interest is crack on-set strain, which is usually used to evaluate the stretchability of a semiconductor film. As demonstrated in previous reports, the main failure mechanism for a semiconductor film during stretching is crack propagation, which can hinder charge transport and the device will suffer from dramatic electrical performance drop.
Supplementary Figure 31. Summarized mechanical properties characterization results for iRUM-s-x:y films. The elastic modulus and fracture strain are extracted from pseudo free-standing tensile tests (film-on-water). The crack on-set strain is obtained from "film-on-elastomer" measurement. The PDMS fractured when stretched to 120% strain, thus the crack on-set strain for iRUM-s-1:1 and iRUM-s-3:1 is defined as 120%.
Supplementary Figure 33. The representative stress-strain curves for DPPTT, iRUM-s (BA/DPPTT) films, BF/DPPTT blend films, BAc/DPPTT and BH/DPPTT crosslinked films, where the precursor-to-DPPTT weight ratio is 1:1, which were obtained from pseudo free-standing tensile tests 7,11 . Water-soluble PSSNa sacrificial layer (thickness: ~30 nm) was firstly prepared on Si substrate, followed by directly spin-coating precursor/DPPTT-1:1 blend solutions and thermal crosslinking in nitrogen atmosphere (thickness: ~40 nm). The polymer films were first patterned into dog-bone shape according to previous reports, followed by slowly dipping into deionized water to release and float the semiconductor film by dissolving the underneath PSSNa layer. Later, the semiconductor film was hold with two aluminum tensile grips coated with a thin PDMS layer, and then tensile tests were performed. The strain rate is 0.05 s -1 .
Note: with the same value of precursor-to-DPPTT weight ratio, all of the films exhibit similar elastic modulus with each other, yet lower elastic modulus than neat DPPTT films, which is determined by the amount of secondary component added. BF/DPPTT doesn't show any improvement in fracture strain than DPPTT, due to the lack of energy dissipation mechanism. Simply softening DPPTT through the plasticizing effect of polybutadiene precursor cannot delay crack formation. For iRUM-s, BAc/DPPTT and BH/DPPTT, they show higher fracture strain (crack on-set strain higher than 100%) than DPPTT (crack on-set strain of 50%), due to the formation of a rubber matrix. However, the extent of improvement is different among them. BH/DPPTT shows the highest fracture strain, iRUMs is in between, while BAc/DPPTT is the lowest. The reason lies in network structure. BH/DPPTT has one uniform network, BAc/DPPTT has double network, in which there only exists physical interaction between the two networks.
In the case of iRUM-s, the BA network is chemically connected to DPPTT network. The more homogeneous of a crosslinked semiconductor film is, the higher fracture strain will be obtained. These were obtained from pseudo free-standing tensile tests. Water-soluble PSSNa sacrificial layer (thickness: ~30 nm) was firstly prepared on Si substrate, followed by directly spin-coating BA/DPPTT-x:y blend solutions and thermal crosslinking at 150 °C for 40 min in nitrogen atmosphere (thickness: ~40 nm). An additional thin layer of PDMS is directly prepared on top of the semiconductor film by spin-coating a dielectrics solution followed by annealing at 70 °C for 1h in air and annealing in glove box at 150 °C for 40 min (thickness ~2.4 µm). The multilayer film was slowly dipped into deionized water to release and float the iRUM-s-x:y-PDMS bilayer film by dissolving the underneath PSSNa layer, then was hold with two aluminum tensile grips coated with a thin PDMS layer and performing tensile tests. For stress-strain measurements and cyclic tests, the strain rate is 0.05 s -1 . For stress relaxation, a strain rate of 0.03 s -1 is applied and the strain is kept within the elastic region (5%).  Figure S34). The PDMS dielectric solution was spin-coated and cured directly on top of semiconductor. Therefore, the dielectric layer is of the same size as the semiconductor layer. After transferring semiconductor/dielectric together with CNT gate layer, there is a large overlapping area between the semiconductor and the CNT gate, thus resulting in high gate leakage current. However, this doesn't affect the extracted mobility of the semiconductor much. For future circuit fabrication, both the semiconductor and the gate will be patterned into small structures to minimize overlapping area to minimize leakage current.  film after multiple stretching-releasing cycles at 25% strain under strain released state. IDTBT suffers from nearly one order drop in mobility even after cycling at 25% strain, which is due to plastic deformation.  Note: Due to the polybutadiene backbone structure, interfacial crosslinking can be created between BF/IDTBT blend semiconductor film and PDMS dielectrics through hydrosilylation during Si-H/vinyl curing process. In addition, due to the plasticizing effect, BF/IDTBT exhibits reduced elastic modulus compared to IDTBT. However, BF/IDTBT blend film still suffers from mobility drop after multiple stretching-releasing cycles, which confirms that the cyclic durability cannot be improved through simply softening IDTBT with plasticizer or form interfacial crosslinking with PDMS dielectrics. BF/IDTBT blend film still undergoes plastic deformation under strain, which is similar as IDTBT film. Therefore, the improvement in cyclic durability of iRUM-IDTBT originates from its improved elasticity.

Supplementary
Supplementary Figure 53. The representative stress-strain curves and extracted mechanics parameters for IDTBT and iRUM-IDTBT-x:y films, which were obtained from pseudo free-standing tensile tests.
Supplementary Figure 54. Crack on-set strains of IDTBT and iRUM-IDTBT film, as measured by "film-onelastomer". PDMS (base:crosslinker =12:1 in weight ratio).  profilometer before-and after-solvent treatment (depositing chlorobenzene or chloroform on film surface and stay for 10s, followed by spin-coating at 1500 r.p.m. for 1 min). Compared to pristine SEBS film which showed substantial swelling after solvent treatment (roughness increased from 7 nm to 178 nm), iRUM-d-1:5 and iRUMd-2:5 showed less swelling. However, the surface roughness still increased from 6 nm to 22 nm for iRUM-d-1:5 film, and 4 nm to 13 nm for iRUM-d-2:5 film. This explains the relatively low charge carrier mobility after depositing and photo-patterning of iRUM-s film on top of them compared to iRUM-d-4:5 film. Therefore, increasing the crosslinking density in SEBS dielectrics resulted in improved solvent-resistance and more smooth film surface after multiple solvent washing steps.

PDMS & BA rubber PDMS & PDMS
Supplementary Figure 60. Transfer curves from the bottom-gate top-contact transistors, with iRUM-s directly photo-patterned on iRUM-d-x:y, highly-doped Si as gate electrode and MoO3/Au as source/drain electrodes (W= 1000 µm , L= 50 µm) (solid lines: drain current; dash lines: square root of the drain current).   Supplementary Figure 64. Bright field and dark field optical microscope images of photo patterning iRUM-s directly on top of (A) unpatterned iRUM-d (B) photo-patterned iRUM-d. Mask 1 and 2 are the shadow masks for photo-patterning iRUM-s with different feature sizes. For mask 1, the size of patterned iRUM-s film is smaller than patterned iRUM-d film, cannot see clearly under optical microscope due to color similarity. For mask 2, the rectangle with dash edge represents photo-patterned iRUM-s, while the square represents photo-patterned iRUMd. Therefore, iRUM-s can be successfully photo-patterned on top of iRUM-d (unpatterned and patterned) without observing swelling under dark field.