Graphene e-tattoos for unobstructive ambulatory electrodermal activity sensing on the palm enabled by heterogeneous serpentine ribbons

Electrodermal activity (EDA), also known as galvanic skin response, is widely used as a quantitative indicator of psychological arousal and mental stress. Psychophysiological literature recommends the palm—especially the thenar and hypothenar eminences and finger phalanges—because of the high density of eccrine sweat glands that respond primarily to psychological stimuli. Conventional ambulatory EDA systems use Ag/AgCl gel electrodes on the palm connected via wires to a rigid wrist-worn PCB, which is obstructive, stigmatizing, prone to mechanical delamination, and degrades over time due to gel dehydration increasing skin-electrode impedance. To reduce obstructiveness, dry electrodes and alternative body locations (wrist, forearm, shoulder, back) have been explored, but off-palm measurements suffer from contamination or interruption by thermoregulatory sweat (apocrine glands), leading to lower signal fidelity. A prior dry wireless hand sensor was still thick (~1 cm) and obstructive. Ultrathin, skin-conformal electronics (e-tattoos) offer mechanical and optical imperceptibility and stability; prior work from the authors introduced sub-micron-thin graphene e-tattoos (GET) with low electrode–skin impedance and multiple sensing modalities. However, a key unresolved challenge is forming a robust, stretchable electrical interface between sub-micron-thin GET and millimeter-thick rigid circuit boards for ambulatory use, since conventional interconnect methods (soldering, ACF, z-axis tapes) or pastes/liquid metals are unsuitable due to stiffness, fragility, or encapsulation complexity. The research question is how to engineer a mechanically robust, unobstructive, high-fidelity palm EDA sensor by solving the ultrathin-to-rigid interconnection problem.

Prior EDA sensing approaches include gel electrodes on the palm connected via wires to wrist-worn hardware (standard clinical/gold-standard), which are obstructive and degrade as gels dry. Dry electrodes on non-palmar sites (wrist, forearm, shoulder, back) reduce visibility but yield lower signal quality due to thermoregulatory sweat interference; a thick wireless dry hand sensor remained obstructive. Ultrathin e-tattoos based on graphene and other materials have demonstrated excellent skin conformability, low impedance, and multiparametric sensing (ECG, EMG, EEG, skin hydration, temperature, pressure). Despite advances in ultrathin sensors, robust stretchable electrical connections to rigid DAQ systems remain a bottleneck. Prior soft contacts (silver paste, liquid metals) suffer from stiffness after curing or require encapsulation. This work builds on those findings by proposing heterogeneous serpentine ribbon (HSPR) interconnects and a soft conductive interlayer to create a robust, reusable, adhesive-free interface for ambulatory palm EDA sensing.

Design: The proposed GET–wristband interface comprises two GET serpentine electrodes placed on the thenar and hypothenar regions of the palm. Each GET serpentine partially overlaps a sub-micron-thin Au/PI serpentine to form a heterogeneous serpentine ribbon (HSPR), relying solely on van der Waals adhesion between ultrathin layers and skin. The Au/PI end expands to contact rigid electrodes on a commercial EDA wristband (Empatica E4) through a reusable soft interlayer composed of Ecoflex with embedded conductive silicone rubber vias for vertical conduction. HSPR configurations and geometry: Three configurations were analyzed: heterogeneous straight ribbon (HSTR), HSPR with Au/PI step edge at the crest (HSPR Crest), and HSPR with step edge at the arm (HSPR Arm). Serpentine unit cell parameters (width w, radius r, angle α, arm length l) were explored via FEM to minimize strain in GET; the selected geometry for experiments had w/r=0.2, l/r=0.5, α=20°, with w=1 mm. Fabrication: GET was fabricated via a wet-transfer (CVD graphene) and dry-patterning method, replacing PMMA with PI as support for improved stretchability. Bilayer graphene on PI reduced sheet resistance to ~410 Ω/sq. A 750-nm Au/PI (100 nm Au on 650 nm PI) bilayer was partially laminated over GET on tattoo paper to define HSPR, then laser-patterned. HSPR was transferred to skin by water-release. Au/PI electrical connectors were made by spin-coating PAA on Cu foil, imidizing to PI, etching Cu, transferring to tattoo paper, and depositing Cr/Au. Soft interlayer: A 3×3×0.2 cm Ecoflex pad was molded with two 8-mm-diameter holes filled with conductive silicone rubber (SNE-553). Mechanical properties (modulus 0.19–2.3 MPa) and vertical resistance (<50 Ω under 20% compression up to 10,000 cycles) were characterized. Mechanical testing: Ribbons were mounted on 100-µm Ecoflex substrates and clamped end-to-end. Resistance change (R/R0) under uniaxial tensile strain was measured until fracture for HSTR, straight GET, HSPR Crest, and HSPR Arm. Stretchability was quantified at R/R0=2 and also by catastrophic failure points. Cyclic tests of HSPR Arm were run between 0–20% strain at 0.25 Hz up to 10,000 cycles. SEM validated GET conformation over Au/PI step edges. Modeling: 3D shell FEM (ABAQUS) simulated strain in GET for different configurations under 20% longitudinal strain and under transverse strain. Stiffness mismatch effects were studied for thicker Au/PI (13 µm) and Cu (18 µm). A 2D plane-strain FEM for the soft interlayer assessed strain isolation in Au under 1 mm shear displacement across the interlayer. Electrode–skin interface characterization: Using a HIOKI LCR meter (42–1000 Hz), impedance of GET vs. gel electrodes of identical area (1.5 cm²) on the palm was measured. Analytical conformability modeling determined thickness thresholds for full skin conformal contact given typical skin roughness parameters. EDA data collection and validation: Six healthy adults (ages 25–35; 5 males, 1 female) wore GET and gel electrodes on the same palm connected to separate E4 wristbands (4 Hz sampling). Testing comprised sessions: expectation (30 s), uncontrolled emotional (200 s, blank screen), expectation repeat (30 s), controlled emotional (400 s, scaled affective pictures from EmoMadrid, 10 images with exposure/recovery), and habituation (120 s, one picture repeated thrice). Dry metal electrodes on the wrist were also tested. Signals were synchronized via cross-correlation, low-pass filtered (0.2 Hz), and separated into tonic/phasic components (0.045 Hz high/low-pass). Candidate SCRs were detected via trough-to-peak method (Ledalab) with ≥0.05 µS amplitude threshold. A new event selection policy objectively selected valid SCRs from gel first, then corresponding GET events; features compared included amplitude, t_peak, t_response, t_rise, t_rec50%, and t_rec10%, with mean error, 95% CI, and t-test p-values computed. Wearability and robustness tests: Motion artifacts were evaluated during hand clenching, wrist bending, phone grabbing, and poking; friction robustness via rubbing with a metal key ring and 300-cycle rubbing on metal and wood surfaces; water exposure via pouring water; humidity effects using a chamber varying RH 54–94%; long-term ambulatory monitoring over multiple 15-hour sessions during daily activities (driving, eating, TV, exercise, study, sleep). Liquid bandage was applied as a thin protective layer over GET and Au/PI during wearability tests.

• HSPR dramatically reduces strain concentration compared to HSTR. FEM and experiments show approximately fifty-fold reduction of strain at the Au/PI step edge for HSPR Arm versus HSTR; HSPR Arm maintains strain at the step edge ~0.7% under 20% applied strain, versus ~35% maximum strain in HSTR.
• Stretchability (criterion R/R0=2): HSTR 4.4±1.1%; straight GET 9.8±0.3%; HSPR Crest 32±4.7%; HSPR Arm 42±2.6%. Using catastrophic failure points, stretchability improves from ~1% (HSTR) to ~51.5% (HSPR), i.e., ~50× enhancement.
• FEM under 20% longitudinal strain: maximum GET strain—HSTR 35%; HSPR Crest 6.7% (at inner crest coincident with step edge); HSPR Arm 4.3% (maximum at crest), with step edge at arm only ~0.7% strain.
• Transverse loading (20%): HSPR maximum GET strains ~5.9–6.0% at shallow crests; step-edge strains 3.9–5.3%, still much lower than HSTR (by ~8×), though less effective than under longitudinal strain.
• Stiffness mismatch: Increasing connector thickness/stiffness (13-µm Au/PI, 18-µm Cu) shifts failure toward the step edge and reduces stretchability; experiments show discrepancies due to buckling/delamination not captured by FEM, indicating FEM applicability when stiffness ratio ≲100.
• Soft interlayer efficacy: 2D FEM under 1 mm shear shows negligible Au strains (0.00028% and 0.0049%) with interlayer, versus ~1% Au fracture-level strain without the interlayer under only 2.5 µm displacement. Conductive rubber vias exhibit low vertical resistance (<50 Ω) and stability over 10,000 compression cycles.
• Electrode–skin interface: GET exhibits lower contact impedance than gel electrodes of equal area, attributed to superior conformability; analytical modeling predicts full conformability for membranes thinner than ~475 nm for given skin roughness, consistent with images showing full GET and partial Au/PI conformity.
• EDA validation: GET palm measurements closely match gel-based palm measurements, with statistical analyses generally yielding p>0.05 for SCR feature comparisons, supporting equivalence in event detection. Wrist-based dry metal electrodes failed to capture meaningful phasic components compared to palm-based sensors.
• Wearability and robustness: GET shows slightly smaller motion artifacts than gel; artifact morphology is distinct from SCRs and removable by selection algorithms. GET maintained functionality through 600 rubbing cycles and after brief water exposure (transient spike with recovery). Increased humidity elevates tonic SCL but not phasic SCR responses. Long-term ambulatory monitoring achieved continuous 15-hour sessions (three sessions demonstrated), with GET outperforming gel electrodes that frequently delaminated and required replacements; GET captured sleep-related “EDA storm” events that gels missed due to detachment. Observed mechanical failures occurred in Au/PI beneath the soft interlayer rather than in GET.

The study addresses the key barrier to deploying ultrathin e-tattoo sensors in real-world ambulatory settings: forming a robust, low-profile, and stretchable interface to rigid electronics. By combining heterogeneous serpentine ribbons (placing the Au/PI step edge at the serpentine arm) with a compliant vertically conductive soft interlayer, the design minimizes strain in the ultrathin GET and in the Au nanomembrane, effectively eliminating the classic step-edge strain concentration problem. The strain relief achieved by HSPR Arm yields stretchability comparable to homogeneous GET serpentines despite the heterointerface, enabling reliable palm-based EDA acquisition in daily-life conditions. Equivalent-circuit analysis confirms that added interfaces (Au–graphene, soft interlayer) introduce negligible additional resistance relative to the dominant skin interface impedance, and current flows through the GET–skin interface. Experimentally, GET provides lower electrode–skin impedance than gel due to superior conformability, which contributes to high-fidelity phasic EDA detection. In controlled and uncontrolled emotional stimuli, GET measures SCR features comparably to gel electrodes on the palm, while wrist dry electrodes lack phasic content, reaffirming the importance of palmar placement. Under motion and during long-term wear, GET’s unobtrusive, substrate-free, ultrathin construction reduces motion artifacts and avoids delamination issues common to gel electrodes, enabling continuous ambulatory monitoring and detection of nocturnal EDA phenomena. These results demonstrate a generalizable mechanics and integration strategy for ultrathin skin electronics interfacing with rigid systems, applicable beyond graphene to other ultrathin materials and sensors.

This work introduces a mechanically robust, unobstructive palm EDA sensing system by engineering a heterogeneous serpentine ribbon (HSPR) interface between sub-micron-thin graphene e-tattoos and rigid wristband electronics, augmented by a soft conductive interlayer. HSPR with the Au/PI step edge at the serpentine arm reduces strain by ~50× compared to heterogeneous straight ribbons and achieves stretchability comparable to homogeneous serpentines. The soft interlayer effectively isolates shear from rigid electrodes, protecting fragile ultrathin conductors. The GET-based sensor delivers lower skin-contact impedance than gel electrodes, captures phasic EDA events reliably, and enables long-term ambulatory monitoring with fewer motion artifacts and greater comfort than gel-based systems. Future directions include optimizing interconnect transparency (e.g., ultrathin PEDOT:PSS), improving robustness under transverse and multi-axial strains, extending FEM frameworks to account for buckling/delamination at higher stiffness mismatches, and translating the HSPR–interlayer strategy to other ultrathin epidermal sensors and modalities.

• HSPR effectiveness diminishes under transverse loading compared to longitudinal stretching; while still beneficial, strain reduction is smaller.
• FEM predictions align well for small stiffness mismatches but diverge for very stiff/thick connectors (e.g., 18-µm Cu) due to unmodeled buckling/delamination; current FEM is applicable up to stiffness ratios ~100.
• The Au/PI connector beneath the soft interlayer remains a mechanical weak point under prolonged shear; observed long-term failures occurred in Au/PI, not GET.
• Dry electrodes can exhibit motion artifacts, although GET showed slightly smaller artifacts than gel; careful event selection is still required.
• Differences in electrode location/size between GET and gel introduce variability in amplitude and may affect some statistical comparisons.
• Environmental humidity affects tonic SCL (baseline) though phasic SCRs remained stable; thick overlays (e.g., 47-µm Ecoflex) can induce GET delamination, emphasizing the need for substrate-free, ultrathin construction.