Preparation of waterborne polyurethane/silsesquioxane/carbon nanotube aerogel with rigid-flexible framework for mechanically tough and wide pressure-range properties for high-performance piezoresistive sensing

Abstract

The piezoresistive response of flexible piezoresistive sensors is usually limited to narrow low-pressure regions, with limited linear stability during wide pressure variations, which presents a significant difficulty. To overcome this constraint, waterborne polyurethane (WPU) and hydroxylated silsesquioxane (HSQ) are covalently bonded to form a ‘rigid-flexible’ hybrid framework. The typical problems of low mechanical strength, poor filler dispersion, and weak interfacial bonding in traditional aerogel sensors are all concurrently resolved by this molecular-level design. The resulting WPU–HSQ₈/CNT composite aerogel, created using an eco-friendly freeze-drying procedure, achieves a remarkable combination of mechanical toughness and electrical sensitivity by integrating multi-walled carbon nanotubes (MWCNTs) to form a continuous conductive network. The WPU–HSQ₈/CNTs exhibit excellent dynamic stability under high compression rates, superior fatigue resistance over extended cycling, and stable multi-level piezoresistive response across an ultra-wide pressure range of 0–5 MPa, a capability rarely achieved in prior studies. The practical viability of this aerogel is proven by real-time tracking of human joint movements. This material offers a viable platform for next-generation flexible sensors in wearable health monitoring, electronic skin, and human–machine interfaces by presenting a material design strategy that resolves the long-standing trade-off between operational range and sensing stability.

---

1. Introduction

The development of flexible electronics technology has resulted in the rise of piezoresistive sensors as an intelligent technological apparatus for a wide range of applications, including health monitoring, human–machine interfaces, and electronic skin.^1–3 The sensors' structural simplicity, cost-effectiveness, and rapid reaction time are the driving forces behind their technological spread.^4,5 Conductive aerogels represent an exemplary piezoresistive sensing material, as their tunable three-dimensional (3D) porous networks endow them with the capacity for reversible deformation under pressure.^6,7 Because of this structural characteristic, internal conductive routes may be effectively adjusted, resulting in extremely sensitive and rapidly responding piezoresistive sensors. Thus, developing conductive aerogels with remarkable integrated properties that combine macroscopic mechanical flexibility and microscopic electrical responsiveness has become a critical step toward producing the next generation of high-performance, wearable piezoresistive sensors.

In the material design of aerogel piezoresistive sensors, there is often a struggle to balance all the critical properties of single-component materials. Classical carbon-based compounds with high conductivity and lightweight properties⁸ include graphene^9,10 and carbon nanotubes.^11,12 However, their inherent brittleness restricts their applicability in flexible scenarios. Because polymer aerogels (such as polyurethane,^13,14 polyimide,^15,16 and polyacrylamide¹⁷) provide superior flexibility, stretchability, and processability, they are ideal flexible substrates. Nevertheless, their inherent insulating properties prevent mechanical force from being directly converted to electrical impulses. Therefore, contemporary research frequently uses two types of multi-material composite techniques to overcome the limitations imposed by the intrinsic features of single-component materials.

The first strategy is the ‘synergistic reinforcement of carbon skeletons and nanofibres’ approach,^18–21 which involves adding nanofibers, such as aramid nanofibers²² or other cellulose,²³ to carbon networks to increase their mechanical strength. Nevertheless, it has been demonstrated that insulating nanofibers interfere with conductive routes, which lowers conductivity and sensitivity.^24–26 Additionally, there are problems associated with these nanofibers, such as weak interfacial interaction with the carbon matrix and complicated manufacturing techniques. The second strategy is the ‘incorporate conductive fillers into polymer matrices’ approach, which entails utilising flexible polymers as the matrix and adding carbon-based materials to form conductive networks. The widespread use of flexible wearable sensing devices has been greatly aided by this strategy, which facilitates achieving an ideal balance between conductivity and flexibility. However, the stability and repeatability of the conductive network are affected by the conductive filler's propensity for agglomeration at high content, and thus, it is difficult to achieve uniform dispersion. More importantly, the inability to maintain a linear and stable piezoresistive response with this strategy across a wide range of pressures remains a technological bottleneck that limits its further application in complex stress environments.

Elastomers such as polydimethylsiloxane (PDMS)²⁷ and polyurethane (PU) are ideal substrates for conductive aerogels due to their flexibility, stretchability, and ability to conform to the human body's curved surfaces. The biocompatibility of waterborne polyurethane (WPU) is satisfactory, with outstanding processing flexibility and adaptable mechanical properties.^14,28,29 Its water-dispersible characteristic allows for easy mixing with various aquatic systems. Utilising water emulsions as templates, WPU can be directly produced into lightweight, porous aerogels using an eco-friendly freeze-drying approach. Consequently, it is regarded as the most suitable polymer matrix for constructing flexible conductive aerogels. inorganic nanofillers such as MXenes,³⁰ silica (SiO₂),^31,32 and boron nitride (BN)³³ are commonly used to improve mechanical properties. However, these fillers tend to agglomerate in polymer matrices. Conversely, silsesquioxane (SQ) is a perfect reinforcing phase to improve the strength and thermal stability of composites because of its stiff inorganic Si–O–Si cage structure.^34–36 Furthermore, chemical modification of its peripheral functional groups improves dispersion and interfacial reinforcement in polymers. For imparting electrical conductivity, multi-walled carbon nanotubes (MWCNTs) are frequently utilised as fillers to establish 3D conductive networks due to their lightweight nature, high aspect ratio, exceptional mechanical strength, and outstanding electrical conductivity.^37,38

Although the aforementioned strategies have significantly balanced flexibility and conductivity, existing materials remain inadequate for accommodating mechanical strength and sensitivity. Their piezoresistive response is typically confined to a narrow low-pressure range and fails to maintain linear stability across a wide pressure range, significantly limiting their future application under complicated stress conditions. To overcome these limitations, we employed covalent bonding between hydroxylated silsesquioxane (HSQ) and WPU to construct a ‘rigid island-flexible sea’ hybrid framework at the molecular level, forming a stable and resilient 3D porous structure. In this system, flexible WPU chains provide macroscopic elasticity and deformation capabilities, while the rigid HSQ core supports the structural backbone. The multilayer hydrogen-bond network produced between them improves the material's structural integrity and energy dissipation capacity and provides efficient stress transfer and structural recovery qualities, setting the groundwork for long-term cycling stability.

The use of MWCNTs creates continuous conducting channels and improves mechanical qualities. The composite aerogel sensor demonstrates a highly porous network structure with remarkable comprehensive performance based on this material design. Multi-level sensing can be achieved across a very broad pressure range of 0–5 MPa, with compressive stress up to 4.61 MPa and toughness up to 0.73 MJ m⁻³. The core sensitivity of 48.21 MPa⁻¹ overcomes the constraint of conventional conductive aerogels, which are only suited to narrow, low-pressure ranges. After 600 consecutive cycles, the sensor exhibits exceptional dynamic stability and fatigue resistance, while concurrently achieving rapid response (approximately 0.60 s) and recovery (approximately 0.50 s). In addition, the material has been successfully applied to real-time monitoring of human joint movements, displaying excellent thermal stability and protective properties, and establishing a new material platform for the development of a new generation of high-performance, flexible sensors.

2. Experimental methods

2.1 Materials

Polycarbonate diols (PCDL2000, hydroxyl value = 57.2) were purchased from Tangyi Chemical Co., Ltd. 2,2-Bis(hydroxymethyl)butyric acid (DMBA), 1,4-bis(isocyanatomethyl)cyclohexane (HMDI), and dibutyltin dilaurate (DBTDL) were purchased from Shanghai Titan Technology Co., Ltd. Multi-walled carbon nanotubes (MWCNTs, outer diameter: 20–50 nm, length: ≤15 mm) were purchased from XFNANO Materials Tech Co., Ltd. Triethylamine (TEA) and acetone were purchased from Fuyu Fine Chemical Co., Ltd. All chemical reagents listed above were used as received without further purification. Hydroxy-functionalized silsesquioxane (HSQ) was prepared using our previously described method.³⁹

2.2 Preparation of the WPU–HSQ_x emulsion

A series of WPU emulsions with varying HSQ content was prepared via the acetone method. The resulting products are denoted as WPU–HSQ_x, where x denotes the mass fraction of HSQ in the polyurethane. The synthesis was conducted under an N₂ atmosphere in a four-necked flask equipped with a condenser and stirrer. PCDL2000, DMBA, HMDI, and a catalytic amount of DBTDL were charged into the flask and allowed to react at 70 °C until the –NCO content reached the theoretical value. Subsequently, TEA was added to neutralize the carboxyl groups in the segments, followed by the addition of a calculated amount of HSQ for crosslinking. For emulsification, the resultant mixture was gradually filled with deionized water under vigorous stirring. Finally, the solvent was removed by rotary evaporation to afford the WPU–HSQ_x emulsion. The detailed synthesis route and experimental formulations are provided in Scheme 1 and Table S1.

Scheme 1 Synthesis of (a) HSQ and (b) WPU-prepolymer. (c) Schematic diagram of WPU–HSQ_x polymerization.

2.3 Preparation of the WPU–HSQ_x/CNT aerogel

The WPU–HSQ_x/CNT composite aerogels were prepared using the emulsion template method and an environmentally friendly freeze-drying process. A dispersion of multi-walled carbon nanotubes (MWCNTs) was ultrasonically mixed with WPU–HSQ_x emulsion to prevent MWCNT agglomeration through cavitation effects, achieving uniform dispersion. The mixture was then frozen, which allowed ice crystal templates to form in the aqueous phase. This process enriched polymers and MWCNTs at grain boundaries, which resulted in the construction of a porous framework. Finally, freeze-drying induced sublimation of the ice crystals, resulting in an aerogel with a high-porosity 3D network structure. The specific preparation process is detailed in Scheme 2.

Scheme 2 Schematic diagram of WPU–HSQ_x/CNT composite aerogel preparation.

2.4 Characterizations and measurements

The FTIR spectral data for the samples prepared by the potassium bromide (KBr) compression method were recorded using a Bruker TENSOR-27 spectrometer with a spectral acquisition range of 500–4000 cm⁻¹. The ¹H NMR spectral data for the samples were recorded using a Bruker AVANCE-400 NMR spectrometer, using D₂O as the solvent. The crystallinity of the samples was analysed using a Rigaku D/MAX 2550 X-ray diffractometer under the following conditions: voltage 40 kV, current 20 mA, 2θ scanning range 5–90°, scanning rate 10° min⁻¹. A Litesizer 500 nanoparticle size tester was used to measure the size distribution of nanoparticles in sample emulsions.

The pore volume distribution of the samples was determined using the Quantachrome Poremaster system (software version 8.11) according to the mercury pressure method. The pore morphology and elemental distribution of the samples were analyzed by scanning electron microscopy (SEM) and energy dispersive spectroscopy (EDS) in conjunction with it, respectively, and the relevant data were collected using a Hitachi S-4800 field emission scanning electron microscope. The experimental procedure entailed the execution of tests employing a TA SDT Q600 thermogravimetric analyser (TGA) in a nitrogen atmosphere. The tests were conducted from ambient temperature to 700 °C, with a heating rate of 20 °C min⁻¹.

Dynamic cyclic loading and static compression tests were carried out at room temperature (25 °C) to characterize the compressive properties of the aerogels using an Instron 3344 universal materials testing machine. The electrical conductivity was determined by means of a two-electrode system, using copper tape as the contact electrode, using a CHI 760E electrochemical workstation.

3. Results and discussion

3.1 Structural analysis

The FTIR and ¹H NMR analyses jointly confirmed the successful hydroxylation of GSQ to HSQ. As shown in Fig. 1a, a broad and intense hydroxyl characteristic peak appeared at 3400 cm⁻¹, while the characteristic peaks of the epoxy group at 915 cm⁻¹ and 840 cm⁻¹ significantly weakened, fully consistent with the characteristics of an epoxy ring-opening reaction. Furthermore, the disappearance of the –NCO peak at 2217 cm⁻¹ and the retention of the characteristic cage-like Si–O–Si backbone peak at 1100 cm⁻¹ (Fig. 1b) confirmed the successful synthesis of the final product WPU–HSQ. This approach effectively anchors rigid SQ nanoparticles within the flexible polyurethane network, providing a structural basis for improved dispersion and enhanced composite performance.

Fig. 1 (a) FTIR spectra of the WPU prepolymer, GSQ, and HSQ. (b) FTIR spectra of WPU–HSQ_x. (c) ¹H NMR spectrum of HSQ. (d) XRD patterns of WPU–HSQ and WPU–HSQ/CNT aerogels.

As shown in Fig. 1c, the peak at 0.644 ppm was assigned to the methylene group connected to the SQ cage; the single peak at 1.65 ppm corresponds to the hydrogen protons of the secondary hydroxyl group formed upon ring opening and the primary hydroxyl group forming hydrogen bonds. The multiplet in the 2.2–3.0 ppm range originates from the methylene group connected to the tertiary amine after ring-opening and the hydrogen protons of the unreacted epoxy group. The signals between 3.1–4.1 ppm were attributed to the methylene group near the ether bond. These ¹H NMR results are consistent with the FTIR analysis, collectively validating the structure of HSQ.

In Fig. 1d, the XRD pattern of the WPU–HSQ aerogel exhibits only a broad peak near 2θ = 20°, corresponding to the amorphous structure of WPU. Although hydrogen bonding between WPU hard segments and HSQ promotes the ordered arrangement of the hard segments, the system still fails to form distinct crystalline phases, maintaining an overall structure dominated by amorphous phases. After introducing the CNTs, this broad peak persisted, indicating that the amorphous structure of WPU is preserved. Typically, CNTs exhibit two characteristic peaks at approximately 26.1° (002) and 44.2° (100), corresponding to their interlayer stacking and in-plane graphitized structure, respectively.⁴⁰ However, in the WPU–HSQ/CNT composite aerogel, these CNT characteristic peaks are masked by the broad WPU peak, and the diffraction intensity near 26.1° is significantly weakened. This indicates that CNTs are uniformly dispersed within the matrix without significant agglomeration.

To characterize the hydrogen bonding in the WPU–HSQ_x system, peak splitting fitting was performed on the carbonyl peak in the 1600–1800 cm⁻¹ range using the Gauss–Lorentz equation. As shown in Fig. 2a–c, the absorption peak in this range can be resolved into 3–4 subpeaks, with their assignments and origins as follows:

Fig. 2 FTIR peak deconvolution of the C=O stretching region in (a) WPU–HSQ₇, (b) WPU–HSQ₈, and (c) WPU–HSQ₉.

The fitted peak at 1743 cm⁻¹ corresponds to the free-state carbonyl group, representing the C=O stretching vibration in PCDL2000. The fitted peak at 1695–1720 cm⁻¹ corresponds to hydrogen-bonded carbonyl groups, with peaks at 1720 cm⁻¹ and 1707 cm⁻¹ representing disordered hydrogen-bonded carbonyl groups originating from conventional urethane groups and urethane groups formed by HSQ cross-linking, respectively. The peak at 1695 cm⁻¹ corresponds to ordered hydrogen-bonded carbonyl groups, contributed by the urethane structure formed through block copolymerization of PCDL2000 with HMDI. According to the above fitting results, the incorporation of HSQ effectively enhanced hydrogen bonding interactions within the WPU–HSQ system.

A comprehensive analysis of deviations in the peak area of the carbonyl region (Table S2) and the peak position drift reveals that the HSQ content significantly affects hydrogen bond formation. As HSQ content increases, the peak area of the hydrogen-bonded carbonyl group (C=O, HB) gradually increases, while the peak area of the free carbonyl group (C=O, free) decreases. Simultaneously, the peak position of hydrogen-bonded carbonyl shifted toward lower wavenumbers, suggesting that carbonyl groups were more actively involved in hydrogen bonding during polymerization. This is mainly due to HSQ's function as a crosslinking agent, which improves molecular chain packing and promotes the formation of hydrogen bonds. When the HSQ content is excessively high, the high crosslink density and steric hindrance imposed by its inorganic cage structure restrict the mobility of the soft segment molecular chains. Consequently, the formation of ordered hydrogen bonds was impeded, leading to a decrease in the relative proportion of hydrogen-bonding carbonyl groups. As a result, the peak shift toward lower wavenumbers noticeably slowed.

3.2 Emulsion characteristics

The particle size properties of WPU–HSQ emulsions and WPU–HSQ/CNTs dispersions were characterized by dynamic light scattering (DLS). As illustrated in Fig. 3a and Table S3, the appearance of the WPU–HSQ emulsion was transparent and bluish, with an average particle size of less than 80 nm. It exhibited outstanding centrifugal stability and homogeneous dispersion, suggesting that the nanostructure of the emulsion was not affected by the addition of HSQ. Through hydrogen and covalent bonding, the SQ backbone and the WPU polymer chain segments increased the emulsion's stability. When MWCNTs were introduced, the appearance of the emulsion changed from transparent to a black nano-dispersion, with an average particle size of more than 150 nm and a broader particle size distribution. Nevertheless, the system maintained excellent stability. This suggests that the WPU–HSQ matrix effectively disperses and stabilizes carbon nanotubes by physically adhering to and entangling with CNTs to form a stable core–shell composite structure.⁴¹

Fig. 3 Particle size distribution of the (a) WPU–HSQ and (b) WPU–HSQ/CNT emulsions. Pore size distribution of the (c) WPU–HSQ₈ and (d) WPU–HSQ₈/CNT aerogels.

3.3 Morphology of aerogels

The internal pores of the WPU–HSQ aerogel exhibit a typical 3D layered stacked porous structure (Fig. 4a), with relatively smooth surfaces and inner walls of the framework. This lamellar pore structure with significant layer spacing is attributed to the polymer network that was produced and aligned during ice crystal development and sublimation, enriched at the grain boundaries.^42,43 With an average pore diameter of 19.8 µm and a high porosity of 87.51%, its structural morphology is consistent with the pore parameters in Table S3. As shown in Fig. 3c, the WPU–HSQ₈ aerogel exhibits uniform pore size distribution with a low bulk density (0.2498 g cm⁻³) and apparent density (1.0132 g cm⁻³), indicating its lightweight characteristic—it can easily stand atop a flower pistil.

Fig. 4 SEM images of the pore morphology for different aerogel samples: (a) WPU–HSQ₈ aerogel, (d) its magnified view, (b) WPU–HSQ₈/CNTs aerogel before compression, (c) after compression, and (e and f) corresponding magnified views of the compressed and uncompressed WPU–HSQ₈/CNT aerogel.

The microstructure and formation mechanism of the aerogel were significantly changed by the addition of CNTs (Fig. 4b and e). During the freezing process, the uniformly dispersed CNTs act as a crucial template due to their superior thermal conductivity. This promotes the homogeneous nucleation and growth of ice crystals through stable heat transfer, thereby forming composite aerogels with a uniform 3D porous network structure.²¹ Therefore, the WPU–HSQ₈/CNT composite aerogel maintained a high porosity of 77.83% and a more concentrated pore size distribution, but its average pore size increased to 27 µm. It is clear from the comparison of the changes in internal pore morphology of WPU–HSQ₈/CNT composite aerogels under compression (Fig. 4b and c) that the pore walls gradually densify and elastically flex. The pore diameter decreases as pressure rises, but the pore wall contact points and contact area also rise.

3.4 Thermal performance analysis

The thermal degradation process of the WPU–HSQ hybrid aerogel shows a clear three-stage characteristic, according to the thermogravimetric analysis (TGA) results displayed in Fig. 5 and Table S4. At temperatures below 250 °C, the negligible weight loss in the sample is mostly ascribed to the elimination of residual water and surface adsorbates, while the primary skeletal structure of the material remains unaffected. The pyrolysis of the aerogel's primary structure occurs between 250–500 °C, which can be further separated into two separate sub-stages according to the different degradation impacts. The cleavage of the urethane group (–NH–CO–O–) in the WPU hard segment occurs between 250 and 370 °C.⁴⁴ As the temperature increases, the second major decomposition stage occurs at 370–500 °C. During this stage, the polyester polyols in the soft segments of WPU undergo backbone cleavage and thermal decomposition, while there is volatilization and breakage of the organic substituents (epoxy and hydroxyl groups) surrounding the Si–O–Si cage core. During this process, the rigid backbone of HSQ provides structural support and confinement to the WPU segments, delaying the overall pyrolysis process and significantly raising the initial decomposition temperature of WPU from 328.44 °C to 392.58 °C. Finally, the carbonized residue decomposition stage occurs above 500 °C. During this phase, the carbon residues generated in the previous phases are enriched to further cyclize and crosslink to form a graphite-like carbon layer. This layer combines with the inorganic Si–O–Si component from HSQ to form a C/SiO₂ composite ceramic phase that efficiently prevents thermal oxygen migration and builds a dense barrier of protection on the material's surface.⁴⁵ As a result, the residual carbon content of the aerogel significantly increases from 0.33% to 7.44%. The residual carbon content increased to 10.71% as a result of the addition of CNTs, which further improved the carbon layer's continuity and stability. Therefore, the combination of HSQ and CNTs resulted in a highly efficient thermal insulation barrier, significantly improving the aerogel's overall thermal performance.

Fig. 5 TGA and DTG curves of (a and c) WPU–HSQ₈ aerogel and (b and d) WPU–HSQ₈/CNT composite aerogel.

3.5 Mechanical properties analysis

As is well known, the mechanical properties of aerogels significantly impact the stability and responsiveness of compression sensors. Therefore, their mechanical properties are characterized through compression testing. The stress–strain curves of the WPU–HSQ_x and WPU–HSQ_x/CNT composite aerogels reveal three typical stages within an 80% strain range, as shown in Fig. 5a.⁴⁶ Within the linear elastic area (ε < 28%), the aerogel exhibits a nearly linear increase in stress with strain. This is mostly due to the elastic bending of the pore walls and the reversible supporting effect of the skeleton. As the aerogel reaches the plateau zone (28–43%), its porous structure irreversibly collapses and endures substantial deformation, which decreases the stress rise. In the densification zone (ε > 43%), the pore structure completely collapses, resulting in a dense internal skeleton.

In static compression testing, the WPU–HSQ aerogel exceeds most previously reported PU-based aerogels in terms of compressive strength and resilience (Fig. 6b). From a chemical perspective, the superior compressive strength of the WPU–HSQ aerogel stems from its rigid island-flexible sea hybrid network structure. The EDS elemental distribution maps (Fig. S1a–e) reveal a uniform distribution of C, N, O, and Si elements across the pore wall framework, indicating that rigid HSQ nanoparticles act as reinforcing phases, enhancing the overall structural stability. Consequently, increasing the HSQ content from 7% to 8% promotes phase separation through additional HSQ nanoparticles, which results in a denser, ordered multilevel hydrogen-bond network and a more ordered arrangement of WPU hard segments.⁴⁷

Fig. 6 Mechanical properties of WPU–HSQ_x and WPU–HSQ_x/CNT composite aerogels: (a) compressive stress–strain curves.^{13,46,50–53} (b) Comparison of compressive properties with recently reported aerogels. (c and d) Cyclic stress–strain curves of WPU–HSQ₈ and WPU–HSQ₈/CNT composite aerogels under 100 cycles of 30% strain.

This synergistic interaction between covalent crosslinking and hydrogen bonding significantly enhanced the network's toughness and energy dissipation capacity, and enabled WPU–HSQ₈ (Fig. 6a and Table S5) to achieve the highest compressive stress (3.43 MPa) and toughness (0.54 MJ m⁻³). However, when the HSQ concentration was increased to 9%, the excessively high crosslink density and steric hindrance effects of the inorganic cage structure hampered polymer segment mobility and the development of an ordered hydrogen-bond network.⁴⁸ This reduced the network's toughness and energy dissipation capacity, resulting in a compressive stress of 2.98 MPa and a toughness of 0.45 MJ m⁻³. However, these figures remain superior to the mechanical characteristics (2.24 MPa, 0.29 MJ m⁻³) of WPU–HSQ₇. Macroscopically, the SEM images reveal a 3D layered porous structure (Fig. 4d), with sufficient interlayer spacing that facilitates stress dispersion and efficient energy dissipation under compression.

As illustrated in Fig. 4b, leveraging the high modulus and high strength properties of CNTs further enhanced the stress-bearing and stress-transfer capabilities of the aerogel.⁴⁹ During compression, the aerogel pore walls elastically buckled and gradually densified, resulting in a decrease in pore size (Fig. S2). This indicates the structural responsiveness and adaptability of the composite network to external stresses by illustrating that the pore walls developed from elastic to plastic deformation to prevent structural collapse. With a compressive stress of 4.61 MPa and a high toughness of 0.73 MJ m⁻³, the WPU–HSQ₈/CNTs demonstrate optimal compressive performance in Fig. 6a and Table S5. This indicates that this composite aerogel possesses excellent compressive elasticity and strength, proving its potential as a pressure-sensitive sensor device.

To evaluate the service life, WPU–HSQ₈ and WPU–HSQ₈/CNT composite aerogels were compressed 100 times at 30% strain. Nonlinear stress–strain curves with observable hysteresis loops were obtained for both materials (Fig. 6c and d), revealing the typical energy dissipation properties of viscoelastic materials under significant deformations.⁵⁴ After the initial 10 cycles, the curves became highly overlapping, indicating rapid stabilization and excellent fatigue resistance. After 100 cycles, the maximum compressive stress of WPU–HSQ₈ slightly decreased from 0.20 MPa to 0.17 MPa, while that of the WPU–HSQ₈/CNTs decreased from 0.48 MPa to 0.39 MPa, demonstrating satisfactory structural robustness and elastic recovery capability.

As the cycles progressed, the hysteresis loop area progressively decreased, reflecting reduced energy dissipation and internal structural stabilization.⁵⁵ The energy dissipation of WPU–HSQ₈ decreased from 30.46 kJ m⁻³ to 13.40 kJ m⁻³ due to weak hydrogen bond breakage, plastic movement of polymer segments, and localized compaction of pore walls. The energy dissipation of WPU–HSQ₈/CNTs significantly decreased from 17.37 kJ m⁻³ to 6.41 kJ m⁻³. Aside from the aforementioned mechanisms, this reduction benefited from interfacial slip between the CNTs and the matrix, which was combined with a synergistic reconfiguration of the CNT network.⁵⁶ The extremely low steady-state energy dissipation (6.41 kJ m⁻³) indicates structural stability and consistent response during long-term cycling, providing critical assurance for the sensor's long-term reliability and measurement accuracy.

To visually demonstrate the comprehensive performance of the WPU–HSQ aerogel, we conducted three demonstration experiments (Fig. S3). As shown in Fig. S3a, the aerogel easily lifted a 1.5 kg weight without cracking after stretching, indicating excellent load-bearing strength and ductility. In the drop-weight impact test (Fig. S3b), a mere 2 mm-thick aerogel layer effectively protected the glass from damage, demonstrating outstanding energy absorption and dissipation capabilities. Furthermore, after hundreds of folds (Fig. S3c), the aerogel developed wrinkles but remained intact, demonstrating its remarkable fatigue resistance and flexibility. Collectively, these results indicate the material's broad application potential in flexible protective and wearable devices.

3.6 Compression-sensing performance evaluation

Because of its remarkable compressive elasticity and enormous deformation capacity, which are provided by its 3D porous structure, the WPU–HSQ₈/CNT composite aerogel displays application potential as a wide-stress, high-sensitivity piezoresistive sensor material. As seen in Fig. 7, as the applied pressure increases, the aerogel undergoes larger compressive deformation, resulting in a decrease in resistance and a significant increase in the brightness of the series-connected LEDs. This typical negative piezoresistive effect is caused by the increased contact points and contact area between the aerogel pore walls during compression, which multiply and shorten conduction routes.⁵⁷ The phenomenon fully confirms that the resistance of the WPU–HSQ₈/CNT composite aerogel is sensitive and reversible in response to compressive stress.

Fig. 7 (a) Schematic illustration of the WPU–HSQ₈/CNT composite aerogel under light and heavy compression. (b) Photographs demonstrating the corresponding change in conductivity.

The pressure-sensitive coefficient S = (ΔR/R₀)/ΔP was used to comprehensively assess the impact of changing HSQ concentration on piezoresistive performance in Fig. 8a–c. This revealed how HSQ adjusts the stiffness–flexibility balance of the aerogel framework by modulating hydrogen bond density, which influences its sensing behavior across various pressure ranges. At 7% HSQ content, the decreased hydrogen bond density results in a flexible network capable of continuous deformation across a wide pressure range (0.5–5 MPa). As a result, the conductive network exhibits poor sensitivity (25.34 MPa⁻¹) but a broad detection range as it evolves through mostly tunnel-dominated contacts. A 9% HSQ concentration causes excessive hydrogen bonding, which results in an excessively rigid framework. It exhibits ohmic behavior with sensitivity as low as 2.43 MPa⁻¹ and suppresses deformation in the low-pressure region (0–1 MPa). Beyond the yield threshold (1 MPa), fast collapse of pore walls causes tunneling-dominated rearrangement, resulting in high sensitivity (102.38 MPa⁻¹) within a restricted range (1–2 MPa). This great stiffness allows for excellent sensitivity within a restricted pressure range, but low-pressure responsiveness and wide-range detection capacity are lost.

Fig. 8 (a–c) Pressure sensitivity S of WPU–HSQ_x/CNT aerogels at 7%, 8%, and 9% HSQ. Piezoresistive sensing characteristics of the optimized WPU–HSQ₈/CNT aerogel sensor: (d) gauge factor (GF) in low- and high-strain regions; (e) real-time resistance response under low pressure (0.0268–0.0609 MPa); (f) real-time resistance response under high pressure (1.1414–2.9994 MPa); (g) dynamic sensing response at loading rates of 150–300 mm min⁻¹; (h) response time at 300 mm min⁻¹; and (i) cycling stability over 600 cycles at 0.4 MPa.

In Fig. 8b, the WPU–HSQ₈/CNTs obtained the most optimal balance of stiffness and flexibility, allowing the material to support a wide pressure range while supporting rapid conductive network reconfiguration. ΔR/R₀ showed a distinctive three-stage change as pressure increased, which corresponded to different conductive processes and structural evolution. The CNT network underwent reversible elastic bending with a minor increase in pore wall contact area in the low-pressure elastic zone (0–1.0 MPa). Geometric restructuring is the main cause of resistance changes, while ohmic contact dominates the sensing mechanism, resulting in a sensitivity of 14.37 MPa⁻¹ (R² = 0.955).

The rigid structure starts to support loads once it enters the high-sensitivity structural transition region (1.0–3.0 MPa). The aerogel skeletons are pulled closer together by intense compression, which results in quick contact between the pore walls. This dramatically decreases the distance between conductive fillers while also greatly increasing the number of contact points and contact area. As a result, the number of conductive paths and the conductive efficiency are significantly increased due to the enhancement of the quantum tunneling effect. This stage is dominated by tunneling effects, with sensitivity rising to 48.21 MPa⁻¹ (R² = 0.986).⁵⁸ In the high-pressure densification zone (3.0–5.0 MPa), pores completely collapse, the conductive network approaches saturation, and resistance changes flatten.⁵⁹ The conductive mechanism returns to primarily ohmic contact, and the sensitivity drops to 1.63 MPa⁻¹ (R² = 0.951). The segmented sensitivity characteristics stated above show that this composite aerogel exhibits a phased piezoresistive response over a large pressure range, making it appropriate for multilevel pressure sensing.

In Fig. 8d, the gauge factor (GF), which indicates strain for the WPU–HSQ₈/CNTs, increases from 31.63 in the low strain region (0–30%) to 526.81 in the high strain region (60–75%), indicating a shift in the conduction mechanism from geometric rearrangement to quantum tunneling dominance. Simultaneously, ln(ΔR/R₀) demonstrates excellent linearity with ln(ε) across the full testing range (R² = 0.970), with a fitted power exponent b = 2.24, further showing the dominance of tunnelling effects in the growth of the conductive network. Fig. S6 illustrates that, despite its moderate pressure sensitivity (S), the WPU–HSQ₈/CNT aerogel displays a stable piezoresistive response over an extraordinarily broad pressure range of 0–5 MPa in comparison to previously reported piezoresistive materials.^60–65 This exceptional combination of sensitivity and ultra-wide detection range is attributed to the aerogel's optimized structure, which achieves an ideal balance between rigidity and flexibility.

The relative resistance change (ΔR/R₀) reflects the piezoresistive response sensitivity of the WPU–HSQ₈/CNT composite aerogel. As shown in Fig. 8e and f, during cyclic compression of the WPU–HSQ₈/CNT composite aerogel, when the pressure is within the low stress range (0.0268–0.0609 MPa), the corresponding ΔR/R₀ values range from −7.38% to −20.98%. At this stage, reversible elastic bending of the pore walls occurs, causing them to approach each other. The contact points and contact area increase with pressure, shortening the conductive pathways and exhibiting sensitive piezoresistive characteristics. Upon entering the high-pressure zone (1.1414–2.9994 MPa), ΔR/R₀ further changes from −1134.42% to −3762.22%. This corresponds to plastic collapse and gradual densification of the pore walls, where previously isolated conductive pathways interconnect to form a dense network, causing a sharp decrease in resistance. These results demonstrate that the material exhibits high sensitivity, excellent reversibility, and stable cyclic piezoresistive sensing performance across a wide pressure range.

As shown in Fig. S4a and 8g, the WPU–HSQ₈/CNT composite aerogel exhibits stable dynamic piezoresistive performance across a wide compression rate range (20–300 mm min⁻¹) at 0.15 MPa pressure. Furthermore, the aerogel rapidly responds during dynamic loading–unloading cycles (Fig. S4b and 8h), exhibiting a compression response time of approximately 0.60 s and a resistance recovery time of approximately 0.5 s, even at the high compression rate of 300 mm min⁻¹. Benefiting from the material's inherent viscoelasticity, the WPU–HSQ₈/CNT composite aerogel exhibited highly consistent piezoresistive response–time curves across varying compression rates, demonstrating excellent signal repeatability and response stability. This further validates its potential for wide-rate, high-dynamic sensing applications.

The WPU–HSQ₈/CNT composite aerogel's pressure-sensing response at 0.15 MPa pressure during 600 cycles is shown in Fig. 8i. The composite aerogel maintained a stable dynamic response under long-term dynamic loading without apparent damage, which is evident by comparing the curves of 18 cycles at the initial and final stages of the cycle. This suggests that the material possesses excellent cyclic stability and structural integrity.

The results show that there is a wide pressure range and great sensitivity for the WPU–HSQ₈/CNT composite aerogel, allowing for effective detection of transient signals in complicated situations. To visually validate its practical application in wearable sensing, the material emitted stable, distinguishable electrical signals when pressed with Morse code for ‘SDU’ and ‘SFM’ (Fig. S5). As illustrated in Fig. 9, it can track and record distinct and repeatable piezoresistive signals in real time under a variety of bending motions when affixed to important human movement locations (such as the wrist, elbow, ankle, and knee joints). These preliminary tests demonstrate its suitability for physiological signal acquisition scenarios such as human motion monitoring.

Fig. 9 Monitoring human wrist, elbow, ankle, and knee joint movements with WPU–HSQ₈/CNT composite aerogel sensors.

4. Conclusions

WPU–HSQ/CNTs were successfully prepared by chemically bonding and hydrogen-bonding flexible WPU chain segments with a rigid HSQ skeleton and introducing MWCNTs, which possess excellent mechanical properties, superior dynamic stability, and a wide pressure response range. The ‘rigid-flexible’ matrix structure of WPU–HSQ/CNTs allows for exceptional mechanical properties, including an ideal compressive stress of 4.61 MPa and a toughness of 0.73 MJ m⁻³. The structure forms a highly efficient thermal barrier that significantly improves the thermal stability of the material and effectively absorbs and disperses impact energy, thereby providing excellent substrate protection. Additionally, the continuous conductive network inside the material enables a stable multi-stage sensing response in the pressure range of 0–5 MPa.

The WPU–HSQ/CNTs displayed exceptional dynamic stability and fatigue resistance under high-speed loading circumstances of 300 mm min⁻¹ and over 600 cycle tests, respectively, further demonstrating their dependability and longevity in real-world applications. Furthermore, the aerogel was able to precisely track the minor pressure activities of many joints in the human body in real time. The results indicate that this study provides a stable material substrate as well as innovative concepts for the development of high-performance flexible sensors suitable for complicated dynamic scenarios, particularly wearable health monitoring and wide pressure sensing.

Ulrich Schubert dedication

On the occasion of his 80th birthday, I (Hongzhi Liu) gratefully dedicate this paper to Professor Ulrich Schubert, my postdoctoral advisor. His insight, patience, and encouragement during my stay at the Vienna University of Technology (2009–2011), supported by the Lise Meitner Scholarship, were instrumental in my development as a researcher. It is a privilege to honor his lasting influence through this work. My coauthor joins me in expressing respect and good wishes to Professor Schubert on this milestone.

Author contributions

Yiyang Zhang: investigation, data curation, funding acquisition, conceptualization, formal analysis, writing – original draft, writing – review and editing. Hongzhi Liu: investigation, data curation, writing – review and editing.

Conflicts of interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Data availability

All possible experimental and analysed results have been included in this manuscript explicitly. No other new data has been generated by any further experiments/analysis.

Supplementary information (SI) is available. See DOI: https://doi.org/10.1039/d6ta00815a.

Acknowledgements

This work was supported by the National Natural Science Foundation of China (Grant No. 21975144 and 22111530285).

References

1. M. Mahmoudpour, J. E.-N. Dolatabadi, M. Hasanzadeh and J. Soleymani, Adv. Colloid Interface Sci., 2021, 298, 102550.
2. Z. Tan, Q. a. Hu, B. Yang, W. Liu, Z. Zhang, L. Shu and X. Qiu, Adv. Funct. Mater., 2025, 35, e11831.
3. P. Min, X. Li, P. Liu, J. Liu, X. Q. Jia, X. P. Li and Z. Z. Yu, Adv. Funct. Mater., 2021, 31, 2103703.
4. Y. Liu, Z. Zhong, C. Liang, F. Wang, H. Xu, Y. Wan, X. Ma, G. Tian and D. Qi, Chem. Eng. J., 2024, 496, 154069.
5. L. Chen, S. Chen, J. Li, C. Hu, M. Zhu, R. Xiong and C. Huang, Compos. Sci. Technol., 2025, 262, 111062.
6. Y. Han, Y. Cao, S. Bolisetty, T. Tian, S. Handschin, C. Lu and R. Mezzenga, Small, 2020, e2004932.
7. S. Shrestha, K. J. Barvenik, T. Chen, H. Yang, Y. Li, M. M. Kesavan, J. M. Little, H. C. Whitley, Z. Teng, Y. Luo, E. Tubaldi and P. Y. Chen, Nat. Commun., 2024, 15, 4685.
8. M. Jiang, B. Li, Y. Zhao, W. Mo, Z. Peng, W. Wang and F. Liao, Adv. Funct. Mater., 2025, e13514.
9. Y. Zhi, H. Zhang, L. Zhang, Q. Li, X. Kuang, W. Wu, Q. Zhou, P. Li, W. Li and H. Zhang, Adv. Fiber Mater., 2024, 7, 541–553.
10. Y. Lv, J. Wei, W. Wang, H. Deng, Z. Huang, J. Zhou, Z. Chen, J. Xie, X. Huang, Y. Guo and Y. Chen, Int. J. Biol. Macromol., 2025, 295, 139553.
11. X. Du, Q. Chen, Q. Zhou, Y. Zhou, F. Wang, W. Xu, Y. Zhan and M. Jiang, Compos. Sci. Technol., 2025, 260, 110976.
12. H. Li, R. Luo, J. Hu, K. Yang, B. Du, S. Zhou and X. Zhou, J. Mater. Sci. Technol., 2024, 182, 22–32.
13. Y. Wang, X. Li, H. Cheng, B. Wang, X. Feng, Z. Mao and X. Sui, ACS Sustainable Chem. Eng., 2020, 8, 8977–8985.
14. Y. Wang, Q. Qi, G. Yin, W. Wang and D. Yu, ACS Appl. Mater. Interfaces, 2021, 13, 21831–21843.
15. J. Lin, J. Li, Y. Song, W. Chu, W. Li, F. Liu, X. He, Q. Zhao and H. Zhao, ACS Appl. Mater. Interfaces, 2024, 16, 16712–16723.
16. X. Hou, Y. Mao, R. Zhang and D. Fang, Chem. Eng. J., 2021, 417, 129341.
17. C. Wu, L. Zeng, G. Chang, Y. Zhou, K. Yan, L. Xie, B. Xue and Q. Zheng, Adv. Compos. Hybrid Mater., 2023, 6, 31.
18. W. Ye, L. Meng, J. Xi, W. Chen, H. Bian, L. Zhang, H. Xiao and W. Wu, Chem. Eng. J., 2024, 500, 529–539.
19. Y. Tong, F. Xue, X. Lu, X. Liao, J. Meng, L. Zhong and H. Zhu, Int. J. Biol. Macromol., 2025, 320, 145681.
20. W. Ye, L. Meng, J. Xi, H. Bian, Z. Xu, H. Xiao, L. Zhang and W. Wu, J. Colloid Interface Sci., 2024, 666, 529–539.
21. J. Cao, G. Sun, P. Wang and C. Meng, ACS Appl. Mater. Interfaces, 2024, 16, 54652–54662.
22. J. Song, G. Wang, L. Chen, C. Zhang, R. Zan, Z. Wang, Z. Rao and L. Fei, J. Colloid Interface Sci., 2025, 677, 512–520.
23. Y. Zou, Z. Chen, X. Guo, Z. Peng, C. Yu and W. Zhong, ACS Appl. Mater. Interfaces, 2022, 14, 17858–17868.
24. S. B. Choi, T. Noh, S. B. Jung and J. W. Kim, Adv. Sci., 2024, 11, e2405374.
25. C. Cao, P. Zhou, W. Qin, M. Liu, P. Wang, T. Zhang, J. Wang and Y. Qi, Chem. Eng. J., 2025, 509, 161521.
26. X. Wang, A. Bao, B. He, X. Han and J. Hong, Chem. Eng. J., 2025, 523, 168747.
27. D. Li, H. Wang, Z. Han, Q. Wu, X. Lv, Y. Zhang, M. Wang, Z. Li and M. He, Composites, Part B, 2025, 291, 112028.
28. W.-w. Hu, X.-y. Shi, M.-h. Gao, C.-h. Huang, T. Huang, N. Zhang, J.-h. Yang, X.-d. Qi and Y. Wang, Compos. Commun., 2021, 28, 100980.
29. F. Sun, J. Yang, H. Zhang, L. Yi, K. Luo, L. Zhao and J. Wu, Compos. Sci. Technol., 2018, 165, 175–182.
30. Z. Qin, Z. Wang, D. Li, Y. Lv, B. Zhao and K. Pan, ACS Appl. Mater. Interfaces, 2024, 16, 32554–32565.
31. X. Li, Y. Xu, X. Y. An, L. Gong, R. Wang and Z. M. Liu, Int. J. Biol. Macromol., 2025, 304, 140947.
32. C. Ortega-Portas, E. Pinilla-Peñalver, L. Sánchez-Silva, E. Torrecilla-Sádaba, J. Esteban and J. J. Aguilera-Correa, Colloids Surf., A, 2025, 721, 137204.
33. S. Ren, J. Yan, M. Li, Z. Tao, M. Yang and G. Wang, Ceram. Int., 2023, 49, 8945–8951.
34. G. Mirchandani, G. Waghoo, R. Parmar, S. Haseebuddin and S. K. Ghosh, Prog. Org. Coat., 2009, 65, 444–449.
35. Q. Zhang, H. He, K. Xi, X. Huang, X. Yu and X. Jia, Macromolecules, 2011, 44, 550–557.
36. B. Zhao, K. Wei, L. Wang and S. Zheng, Macromolecules, 2019, 53, 434–444.
37. J. Tan, Y. Fang, K. Wang, M. Wei, Z. Zhang, L. Wang, S. Chen and J. Pan, Adv. Funct. Mater., 2025, 35, 2417798.
38. G. Yang, S. Lei, K. Chang, L. Ma, Z. Li, S. Yang and J. Wang, Chem. Eng. J., 2024, 489, 151378.
39. J. Dong and H. Liu, Int. J. Biol. Macromol., 2025, 296, 139686.
40. Z. Huang, Y. Zheng, H. Zhang, F. Li, Y. Zeng, Q. Jia, J. Zhang, J. Li and S. Zhang, J. Mater. Sci. Technol., 2021, 94, 90–98.
41. B. Liu, J. Wang, Z. Li, Z. Sun, C. Li, J.-M. Seo, J. Li, Y. Guo, H. Yao, S. Guan and J.-B. Baek, Nano Energy, 2024, 126, 109611.
42. S. Deville, A. P. Tomsia and S. Meille, Acc. Chem. Res., 2022, 55, 1492–1502.
43. C. Fu, Z. Sheng and X. Zhang, ACS Nano, 2022, 16, 9378–9388.
44. Y. Zhang, R. Zeng, T. Ban, M. Guo, Y. Wang, J. Zhang and X. Zhu, Colloids Surf., A, 2024, 682, 132816.
45. J. Pagacz, E. Hebda, B. Janowski, D. Sternik, M. Jancia and K. Pielichowski, Polym. Degrad. Stab., 2018, 149, 129–142.
46. J. Zhai, Y. Zhang, C. Cui, A. Li, W. Wang, R. Guo, W. Qin, E. Ren, H. Xiao and M. Zhou, ACS Sustainable Chem. Eng., 2021, 9, 14029–14039.
47. K. N. Raftopoulos, M. Jancia, D. Aravopoulou, E. Hebda, K. Pielichowski and P. Pissis, Macromolecules, 2013, 46, 7378–7386.
48. X. Lin, M.-X. Nie, H. Liu, D.-L. Zhou, S.-R. Fu, Q. Zhang, D. Han and Q. Fu, Chem. Mater., 2023, 36, 575–584.
49. Z. Ye, B. Zhao, Q. Wang, K. Chen, M. Su, Z. Xia, L. Han, M. Li, X. Kong, Y. Shang, J. Liang and A. Cao, Adv. Funct. Mater., 2023, 33, 2303475.
50. Y. Zhang, Y. Wu, Z. Liu, Q. Zhang, J. Lu, Z. Dong, X. Cao and S. Li, Int. J. Biol. Macromol., 2025, 298, 140015.
51. R. He, C. Xie, Y. Chen, Z.-X. Guo, B. Guo and X. Tuo, Compos. Sci. Technol., 2022, 228, 109622.
52. X. Zhao, M. Su, S. Yu, J. Zhang, X. Liu, K. Qiu, X. Yi, J. Zhang, G. Dou and M. Wang, ACS Appl. Mater. Interfaces, 2024, 16, 29282–29290.
53. M. Liu, X. Liang, X. Zhang, Z. Hu, P. Gu, X. Yang, G. Zu and J. Huang, Chem. Eng. J., 2024, 502, 157865.
54. E. Hu, Y. Zhu, X. Cheng, C. Deng, L. Zhao, B. Cui, K. Gu and M. Zhu, React. Funct. Polym., 2024, 201, 105944.
55. T. Xu, Q. Song, K. Liu, H. Liu, J. Pan, W. Liu, L. Dai, M. Zhang, Y. Wang, C. Si, H. Du and K. Zhang, Nano-Micro Lett., 2023, 15, 98.
56. R. Zhu, D. Zhu, Z. Zheng and X. Wang, Nat. Commun., 2024, 15, 1344.
57. X. Li, C. Chen, Z. Li, P. Yi, H. Zou, G. Deng, M. Fang, J. He, X. Sun, R. Yu, J. Shui, C. Pan and X. Liu, Nano-Micro Lett., 2024, 17, 52.
58. J. Feng, Z. Zhuang, Y. Zhou and C. Li, Adv. Funct. Mater., 2024, 34, 2315188.
59. M. Xie, G. Qian, Y. Yu, C. Chen, H. Li and D. Li, Chem. Eng. J., 2024, 480, 148203.
60. S. Zhang, X. Zuo, S. Yun, J. Qin, G. Zhang and X. Shi, Chem. Eng. J., 2026, 529, 173050.
61. J. Li, W. Li, J. Lin, W. Chu, Z. Zhao, Y. Lu, X. He and Q. Zhao, Compos. Sci. Technol., 2023, 235, 109953.
62. Z. Wang, C. Wei, L. Fang, H. Zhang, Y. Luo, Q. Zhang, J. Wu, Z. Song, W. He, R. Zhang and P. Wang, Adv. Mater., 2026, 38, e15619.
63. C. Li, R. Xu, D. Han, P. Li, W. Liu, M. Guang, X. Chao and P. Wang, Nat. Commun., 2026, 17, 721.
64. J. Zhou, W. Wang, J. Xie, S. Ding, Y. Chen, J. Xi, Y. Guo and Y. Chen, Ind. Crops Prod., 2026, 241, 122833.
65. A. Huang, Z. Yang, Y. Zhu, B. Tan, Y. Song, Y. Guo, T. Liu and X. Peng, Appl. Surf. Sci., 2023, 618, 156661.

---