I. Abstract of the Paper’s Core Research

This study formulated a binary solvent system of ethanol and terpineol to prepare graphene‑based nano‑ink, and employed an aerosol jet printing process to fabricate sensing circuits on three cellulose‑based substrates—chromatographic paper, sticky notes, and weighing paper. By systematically tuning the number of printed layers, the conductive network was optimized, and a paper‑based sensing layer was integrated with a pneumatic four‑finger soft gripper. Comprehensive tests were conducted covering microstructural characterization, electromechanical sensing, and robotic grasping. The key findings are as follows:

• Compatible with aerosol-specific low-viscosity graphene ink systems

Ethyl cellulose serves as the binder, while a blend of ethanol and high-boiling terpineol modulates the evaporation rate. The ink exhibits stable viscosity, compatible with pneumatic atomization, ensuring no nozzle clogging during prolonged printing, and producing uniform single- and multilayer traces with no noticeable material overflow.

• Paper fibers form a three-dimensional conductive network, with tunable high and low sensitivity.

A single-layer ink coating encapsulates only a single paper fiber, while multiple layers stacked together form a continuous conductive film. In ten-layer printed samples, the compressive sensitivity reaches as high as 56.5, and the tensile GF is 24.2, enabling the detection of both minute deformations and large‑scale bending.

• Dual-Response Stable Piezoresistive Sensing Mechanism for Tension/Compression

When the paper is bent or stretched, the conductive pathways are disrupted, leading to an increase in resistance; under compression, cracks close, increasing the number of conductive paths and reducing resistance. After 50 cycles, the hysteresis is minimal, and the multilayer structure leverages microcracks to amplify the magnitude of signal changes.

• Ecoflex packaging significantly enhances environmental and cycling stability.

The silicone rubber thin layer exhibits permeation‑induced conductive cracking, which suppresses particle shedding; its resistance–voltage curve remains smooth under both high and low pressures, and its durability under humid conditions and repeated deformation is significantly enhanced.

• In-situ direct printing and integrated sensing for the robot body

A U-shaped sensing channel is directly printed within the soft gripper’s paper‑based strain‑limiting layer. With a 5 kPa pressure step, it can precisely distinguish the degree of gripper bending, clearly differentiating between the presence or absence of an object, thereby enabling intrinsic tactile feedback from the gripper itself.

II. Complete Process Flow for Paper-Based Strain Sensors via Aerosol Jet Printing

• Process flow:

1) Dust-free, dry pre-treatment of cellulose substrates (for chromatography, sticky notes, or weighing paper);

2) Graphene, ethyl cellulose, ethanol, and terpineol were ultrasonically dispersed for 4 hours to prepare a specialized aerosol ink.

3) Pneumatic atomization aerosol equipment, with a 3 mm nozzle-to-substrate distance and a low‑speed, 0.5 mm/s layer‑by‑layer printing process for depositing conductive traces;

4) The solvent evaporates naturally at room temperature, eliminating the need for high-temperature sintering; single-layer and multilayer gradient samples can be prepared in parallel.

5) Some samples were spin-coated with an Ecoflex silicone rubber encapsulation layer for protection.

6) Calibrate the strain response of the bending testing machine and assemble the pneumatic soft gripper to perform actual grasping tests.

• Five Core Advantages of Aerosol Technology (Comparison with Screen Printing and Inkjet)

Non-contact compatible porous paper fiber substrate

Aerosol droplets can penetrate the internal fiber network of paper, while screen-printing pressure compresses and damages the paper’s pores, preventing the formation of a three-dimensional conductive sensing pathway.

Wide compatibility with ink viscosities; graphene slurry ensures stable atomization.

Pneumatic atomization is compatible with carbon-based inks of broad viscosity, and graphene nanosheets with a high specific surface area enable stable printing; inkjet printing, by contrast, is prone to sedimentation and nozzle clogging.

Maskless Digital Rapid Design of Sensor Trajectories

CAD‑based free‑form design of U‑shaped and serpentine sensor circuits, adaptable to various soft‑finger sizes, enables multiple gripper‑sensor iterations to be completed in a single day.

In-situ direct printing on the substrate eliminates the chip‑mounting process.

There is no need to separately fabricate and bond a sensing membrane; the sensor can be directly integrated into the robot’s strain‑sensing paper layer, thereby reducing delamination and debonding defects.

Green, biodegradable substrate; low-cost, single-use soft device.

The paper-based substrate is inexpensive and biodegradable after use, while the aerosol‑based consumables deposit only on the sensing area, significantly reducing prototype development costs.

III. Five Core Technological Innovations of This Study

Innovation 1: A binary-solvent‑compatible ink specially formulated for graphene aerosols

High-boiling terpineol regulates solvent evaporation, preventing premature drying and clogging during printing and addressing the challenge of line breakage in aerosol‑based carbon‑nanotube ink deposition.

Innovative New Mechanism for Three-Dimensional Conductive Network Sensing Based on Paper Fibers

Aerosol ink penetrates into the fiber matrix, with single- or multi-layer architectures enabling precise control over the density of conductive pathways; high‑sensitivity piezoresistive response is achieved through the reversible opening and closing of microcracks.

Innovation 3 enables the integration of a highly stable paper-based flexible sensing solution.

Ecoflex silicone rubber penetrates and repairs microcracks in the conductive layer, inhibits particle spalling, and extends service life under humid and cyclic operating conditions.

Innovative 4‑D Soft Robot In‑Situ Integrated Printing and Manufacturing Paradigm

A strain‑limiting layer is integrally embedded within the pneumatic gripper, and sensors are directly aerosol‑printed onto it, eliminating the need for external patch‑on devices and enabling intrinsic tactile sensing.

Innovative Multi-Paper Gradient Performance Material Selection System

Clearly define the application scenarios for three types of substrates—high-sensitivity sticky notes, stable chromatography paper, and low-hysteresis weighing paper—and enable the robot to match the appropriate substrate on demand.

IV. Current Industry Challenges and Corresponding Solutions for Overcoming Them

Challenge 1: Traditional sensors require post‑installation adhesion, and soft robots suffer from poor deformation coupling and are prone to detachment.

✅ Aerosol‑based printing directly creates sensor circuits on the robot’s built-in paper layer, achieving structural integration with no adhesive interface.

Challenge 2: Carbon-based inks are prone to nozzle clogging during atomization, resulting in uneven printed lines.

✅ An ethanol–terpineol blended solvent balances the evaporation rate, enabling long‑term continuous aerosol formation without filament breakage or material overflow.

Challenge 3: Single-layer graphene exhibits weak sensing signals, while multilayer graphene is prone to cracking and suffers from poor cycling stability.

✅ Layer-by-layer printing controls crack density, and when paired with an Ecoflex encapsulation, it delivers both high sensitivity and long cycle life.

Challenge 4: Photolithography and screen printing can only be used to fabricate sensing membranes separately, making in‑body integration impossible.

✅ Maskless direct-write technology adapts to irregular shapes and porous paper inner layers, enabling integrated molding of the sensing unit.

Challenge 5: Plastic flexible substrates are difficult to degrade, leading to significant e-waste pollution.

✅ Made from all-cellulose paper substrate, the material is naturally biodegradable when discarded and compatible with single-use flexible automated equipment.

V. Future Diversified Industrial Application Scenarios

1 Industrial Pneumatic Soft Gripper Robot

The flexible gripper on the assembly line features built-in aerosol‑based paper sensors, enabling precise detection of object presence and accurate control of gripping force, thus ensuring non‑destructive sorting of fragile items.

2 Medical Minimally Invasive Flexible Manipulation Instruments

A disposable soft‑tissue surgical grasper with a biodegradable paper‑based sensor eliminates the need to retrieve electronic components after surgery, thereby reducing medical waste.

3. Low-Cost Software Robotics Platform for Education and Research

Desktop aerosol‑based rapid printing of various tactile‑sensing grippers, enabling low‑cost bionic robot development in university laboratories.

4 Wearable Flexible Human Motion-Sensing Patch

A lightweight, paper-based strain-sensing patch adheres to joints, capturing bending angles for rehabilitation‑exercise monitoring.

5 Flexible Soft Robotic Gripper for Agricultural Harvesting

A field‑ready, single‑pass harvesting robot featuring a low‑cost, biodegradable sensing structure and lightweight, outdoor‑optimized operations.

Image and Text Guide

Figure 1. AJP printing system and GNP ink: from ink formulation and deposition mechanisms to print‑path control

• (a)   Formulated GNP ink sample; (b)   Schematic of the self‑made AJP system: pneumatic atomization (using N₂) generates an aerosol, which is focused by the carrier gas and sheath gas and then precisely deposited onto the substrate via the nozzle. (c)   Groove‑shaped deposition nozzle (1.25 × 0.25 mm), designed to maximize trace width, enhance strain gauge stability, and improve electrical conductivity.
• (d)   Effect of print orientation on track width: When aligned parallel to the nozzle width (0°), the track width is approximately 826 μm; when perpendicular (90°), it reaches about 1189 μm. Intermediate angles yield values between these extremes. For consistency, all subsequent tracks are printed perpendicular to the nozzle width.
• (e)   Schematic of the ink deposition cross-section on paper fibers: the carrier/ sheath gas enables the ink to penetrate deep into the fiber network (as previously reported, penetration depths of approximately 24.9–51.0 μm have been observed on paper and cardboard).
• (f)   The four stages of the interaction between ink and paper are: (i) fibers with no ink deposition; (ii) monolayer deposition coating individual fibers; (iii) continuous deposition leading to thickening; and (iv) filling the interfiber pores to form a continuous graphite layer, with subsequent deposition merely increasing thickness.
• (g)   Representative sample: A 20.0 × 1.5 mm sensor fabricated on S-paper.

Figure 2. Optical microscopy analysis of GNP printing on three paper substrates: the fibrous network governs morphology and cracking.

• (a,b) S-paper (sticky notes) : Single-layer and ten-layer printing; the print path is relatively wide (1778 → 1521 μm). At ten layers, noticeable cracks appear, but the overall structure remains fairly uniform.
• (c,d) F-paper (filter paper) : Highly porous, with high wet strength and rapid solvent absorption; the single-layer pore size is 1363 μm, increasing to 2122 μm in the ten-layer structure. Cracks are fine and uniform, with slight ink spreading visible at the edges, and fibers are densely enveloped.
• (e,f) W-paper (weighing paper) : It exhibits strong moisture resistance and slow solvent absorption; at a single layer, the edges are uneven with ink overflow (1235 μm). At ten layers, cracks become markedly more numerous, forming random fractal cracks several micrometers wide (1350 μm).
• Conclusion The surface roughness, wettability, pore size, and ink‑absorption characteristics of different papers collectively determine the trajectory width, ink penetration depth, and crack morphology—providing a microscopic basis for the subsequent differences in electrical and sensing performance.

Figure 3. Resistance as a function of the number of printed layers and paper type: evolution of the conductive network, initially decreasing then increasing.

• (a)   Resistance of F‑paper as a function of the number of layers: the initial resistance of a single layer is very high, but drops sharply to about one-fifth upon adding the second layer (due to the small volume and insufficient conductive pathways in monolayer GNP); as the number of layers increases, the resistance continues to decrease. At 4 layers, it reaches a minimum of 272 kΩ. ; Subsequently, further deposition led to the formation of microcracks, and by the 10th layer, the resistance had in fact rebounded to approximately 840 kΩ.
• (b)   Three types of paper—single-layer and ten-layer—average resistances (red error bars): S-paper 2766 → 2006 kΩ; F-paper 4900 → 840 kΩ (due to the dense fiber network of filter paper and fewer cracks, resulting in the most pronounced reduction); W-paper 946 → 686 kΩ.
• Mechanism : Multi-layer deposition densifies the GNP network (reducing resistance), but crack formation disrupts the conductive pathways (increasing resistance); the interplay between these competing effects gives rise to the “decrease‑then‑increase” trend.

Figure 4. Cyclic bending characterization: Piezoresistive response and hysteresis under both tensile and compressive loading modes.

• (a–c) Single-layer, and (d–f) ten-layer S/F/W paper sensors; for each sample, (i) the relative resistance change ΔR/R₀ after 50 cycles of tensile bending and 50 cycles of compressive bending, and (ii) the hysteresis loops for the first cycle (red) and the last cycle (blue) during the final ten cycles.
• Maximum bending strain: for S, F, and W papers, it is 0.77%, 0.82%, and 0.44%, respectively. For the ten-layer sample, the maximum ΔR/R₀ is: S-paper under tension, 14.2%; Compressed by 43.4% ;F-paper 12.8% / 17.1%; W-paper 5.7% / 11.4%.
• Mechanism : During stretching, the conductive particle network is pulled apart, leading to an increase in resistance (ΔR is positive); during compression, the particles and cracks close, forming shorter conductive pathways and resulting in a decrease in resistance (ΔR is negative). In ten-layer samples, crack propagation and closure dominate, so the compressive mode exhibits a stronger response; in single-layer samples, hysteresis is smaller and reversibility is improved.

Table 1. Strain‑gauge factors (GF, mean ± standard deviation) for each sample: Sensitivity is generally higher in compression mode.

• GF = (ΔR/R₀)/ε, which represents the average slope of the relative resistance change with respect to bending strain.
• All samples The compressive modulus GF is consistently higher than the tensile modulus. ; Ten layers are superior to a single layer.
• S10 Maximum: Tensile 24.2 ± 1.8, Compressive 56.5 ± 4.5 ; S1 is 4.1 / 11.5, F1 is 4.7 / 14.2, F10 is 17.6 / 22.8, W1 is 3.1 / 12.9, and W10 is 9.1 / 27.2—values comparable to those reported in the literature, such as GF ≈ 7, 42, and 46.

Figure 5. Variable-curvature compression mode test: Screening the ten most stable S-paper configurations.

• (a–f) Three types of paper samples—single-layer and ten-layer—under compression testing: (i) four levels of increasing curvature, each subjected to 10 cycles (41.5 / 83.0 / 124 / 166 m⁻¹); (ii) quasi-static tests at stepped curvatures, with each step held for 10 s in both the ascending and descending directions; (iii) stability at the maximum curvature for 10 s. The vertical axis shows ΔR/R₀ normalized to the value at the maximum curvature.
• Some samples exhibited significant drift during cycling and stability tests, whereas… The ten-layer S-paper (Fig. 5d) exhibits the greatest stability in both cyclic and quasi-static tests. 。
• Accordingly, the author has selected   S10 (Ten-layer S-paper) As the final integration solution for the soft robot, it will undergo subsequent packaging and full-system validation.

Figure 6. Ecoflex elastomer encapsulation: sacrificing some amplitude in exchange for a smooth, stable signal.

• (a)   Optical micrograph of S10 after spin coating with Ecoflex: the GNP layer remains visible beneath the thin elastomer. (b)   Schematic of elastomeric coating penetration into cracks for reinforcement.
• (c,d)   Comparison of tensile/compressive cycling with and without Ecoflex: The signal from the uncoated sample exhibits random spikes (due to loosely adhering particles and transient on/off events), whereas after coating, the signal… More uniform and smooth, with improved reversibility. 。
• Data Ecoflex infiltration increases the substrate resistance by approximately 300% (disrupting conductive pathways), leading to a reduction in ΔR/R₀—resulting in a decrease from 16.5% to about 4.7% in tension and from 44.5% to roughly 21.5% in compression; however, the relative response remains well-defined, and the elastomeric reinforcement enhances the sensor’s durability, making it waterproof and moisture‑resistant in real soft‑tissue environments.

Figure 7. Soft gripper integration demonstration: Sensors are directly printed onto a strain‑limiting paper layer.

• (a)   Design of the four-finger Pneu-net pneumatic soft robotic gripper and its strain‑limiting paper‑layer dimensions; leveraging the printing flexibility of AJP, U-shaped GNP sensing tracks (2 mm line width, 1 mm pitch) are directly printed on the paper layer. , the lead wire exits from the base of the gripper.
• (b)   Pressure‑driven testing: At pressures ranging from 0 to 60 kPa in 5‑kPa increments, the resistance exhibits a distinct step response; as pressure increases (and curvature grows), the signal becomes more pronounced. Moreover, the system delivers low‑noise, low‑hysteresis performance during both pressurization and depressurization, enabling precise tracking of the gripper arm’s bending—monitoring up to approximately 80 m⁻¹ of curvature.
• (c)   Gripping interaction testing: When an object is grasped, bending strain decreases, and ΔR is significantly reduced under the same pressure—demonstrating that the sensor can reliably and reproducibly detect whether an object has been successfully grasped.